Resources >
Case Study

This case study uses data from real parts made with a real mold on a real press by real people.

Introduction

This case study demonstrates only a few of the fundamental ways you can benefit by using Algoryx’s technology. Also please note that correlation charts are only a small fraction of the useful output generated by Algoryx’s Correlation Master (CM) software when doing a Mold Characterization Study analysis or simulation. 

Case Study Analysis Results

The below correlation chart shows, for this case study mold, the correlation between the 2.020″ dimension in cavity 1 (the predicted dimension) and the 5.000″ dimension in cavity 1 (the Predictor Dimension). You can click on the chart to enlarge it. 

CM Run In Analysis Mode - Click to Enlarge

As you can see, the regression line is completely outside of the spec box. That means you cannot make good parts no matter how much you “jockey” the knobs on the press. This statement holds true for all possible combinations of press settings. Just this information, alone, can be extremely valuable and can save valuable time during mold development or when revalidating transfer molds. 

Note that many moldmakers/molders will have only the round circle data points (nominal press settings) available when they use scientific/decoupled molding results. Algoryx’s analysis expands this single-point perspective into a multi-point perspective. Algoryx’s analysis is also 100% compatible with the “min-max-nom” window study done by some moldmakers/molders. 

Project Team Tuning Choices

For this mold, the only feasible option was to tune the mold. Relaxing tolerances (per the customer) and other tuning choices were not an option. The CM output verified that GD & T and measurement capability were not a problem.

If only the nominal (process window center point) data is available, then the moldtuner would cut steel on the 2.020″ dimension in cavity 1 by 0.015″. This would drop the regression line down by 0.015″. The moldtuner would also cut steel on the 5.000″ dimension in cavity 1 by 0.008″. This would move the regression line to the left by 0.008″. The nominal (round circle) data points would be superimposed over the target intersection (the yellow dot). 

However, with Algoryx’s technology, we deliberately change one or more press settings that affect cycle time. For this case study, setup no. 1 (the brown cross data points) had a 6% lower cycle time than nominal. The team therefore selected this point (setup no. 1) along the regression line as their operating point. 

We still cut steel on the 2.020″ dimension by 0.015″ to drop the brown crosses down to the 2.020″ target value. However, we cut steel on the 5.000″ dimension by only 0.006″. This will move the brown crosses so they are superimposed over the target intersection. These distinctions can become critical for tight-tolerance part designs.

By making these changes to the steel, we maximize Cpks/Opk while simultaneously reducing cycle time by 6%. 

Case Study Simulation Results

It is good practice to use Correlation Master’s capability to quickly and easily simulate the team’s tuning choices. The below correlation chart shows Algoryx running this case study data in the simulation mode with the above-mentioned steel tuning changes. The simulation output confirms to the team that they made the right choices and didn’t have any logic or computational errors. 

As you can see, the brown crosses from setup no. 1 are superimposed on the target intersection.

CM Run in Simulation Mode - Click to Enlarge

The normal press variation for this particular press was about +/- 0.002″ of Predictor Dimension movement. This gave an Opk/Cpk of 2.25. The molder could only make good parts after the Algoryx tuning change. The 2.020″ dimension did not need to be measured in production even if it was previously designated as an in-process dimension. 

Algoryx’s simulations are based on actual part measurement data obtained during the Induced Variation Study along with team tuning choices.

Take the Algoryx Challenge – Do an Algoryx mold analysis or simulation using your data!