Using Minitab’s Factorial Design of Experiments to Optimize Injection‑Molding Yield

Problem and Why DOE Matters

In a typical injection‑molding shop the goal is to hit a target wall‑thickness while keeping cycle time low. Changing pressure, temperature, or cooling time can affect quality, but testing each factor alone wastes time and material. A systematic approach that evaluates several factors together saves runs, reduces cost, and points directly to the most influential settings.

What is a Factorial Design in Minitab?

A factorial design lets you study how multiple process factors interact. In Minitab the feature lives under Stat > DOE > Factorial Design. You choose the number of factors, their levels (usually low and high), and whether you want a full, fractional, or orthogonal design. Minitab then creates the run order, generates the worksheet, and can produce an ANOVA table and effect plot automatically.

Step‑by‑Step Example: 2‑Level, 3‑Factor Injection‑Molding Case

The research brief cites a 2‑level, 3‑factor experiment in injection molding. The factors were:

• Factor A – Injection pressure (low = 80 bar, high = 120 bar)
• Factor B – Mold temperature (low = 150 °C, high = 170 °C)
• Factor C – Cooling time (low = 20 s, high = 30 s)

Instead of a full 2³ factorial (8 runs), the analyst selected a 1/2 fractional design, which required 8 runs (the same number as full but with aliased two‑way interactions). The dialog was opened, the three factors were entered, “Fractional” was chosen, and the default run order was accepted.

1. Open Minitab and navigate to Stat > DOE > Factorial Design. Verify the dialog appears – this confirms the DOE module is installed.
2. Select “Fractional factorial design” and set Number of factors = 3, Number of runs = 8, and choose “Resolution IV” (the default for a 1/2 fraction).
3. Enter the low and high levels for each factor as shown above. Use the placeholder names A, B, C if you prefer.
4. Click “Continue” then “OK”. Minitab generates a worksheet with a column for each factor and a column for the response (e.g., wall‑thickness).
5. Run the physical experiment following the generated run order. Record the measured wall thickness for each run.
6. When data are entered, go to Stat > DOE > ANOVA and select “Full” model. Minitab produces an ANOVA table and an effect plot.

Reading the Results and Trade‑offs

The ANOVA output showed two factors with p‑values below 0.05:

• Factor A (pressure) – significant positive effect on wall thickness.
• Factor C (cooling time) – significant negative effect, meaning longer cooling reduced thickness variation.

Factor B (mold temperature) was not significant, indicating it had little impact on the measured response. The effect plot visualized the main effects and highlighted the interaction between A and C, which the fractional design may have partially aliased.

The trade‑off here is run efficiency versus interaction detection. A full 2³ factorial would have required only 8 runs (the same as the fractional design used) but would have estimated all two‑way interactions without aliasing. If the engineer needed to be certain that the A‑C interaction was not masking a real effect, a full factorial or a higher‑resolution fractional design would be preferable. However, the 8‑run fractional design still cut the total number of experiments compared with a higher‑order design that might need 16 runs.

Limitations and Verification Steps

Minitab’s standard DOE wizard assumes linear effects and normally distributed residuals. If the process shows strong curvature or highly interactive behavior, the linear model may give misleading factor estimates. In such cases a response‑surface design (RSM) is recommended, but that requires additional runs and is not available in the basic DOE menu.

Practical cautions include:

• Choose factor levels that stay within equipment and material limits; extreme settings can damage the mold or cause defects.
• Use a continuous, precise measurement (e.g., micrometer reading) rather than a coarse visual gauge, which can bias effect estimates.
• Menu paths may differ between Minitab releases; confirm you are using the correct Stat > DOE route for your version.

Verification checklist:

1. Open Minitab, go to Stat > DOE > Factorial Design. The dialog should appear, confirming the feature is present.
2. Run a simple 2² full factorial (four runs) and verify that the ANOVA table and effect plot are generated exactly as described in the help topic for your release.
3. Open the “DOE Overview” help page (Help > Minitab > DOE) and compare the listed design types with what your version shows.

Actionable Closing

If you need to improve yield or reduce variation in a manufacturing process, start by defining a few key factors, then use Minitab’s fractional factorial DOE to capture main effects with a manageable number of runs. Check the ANOVA and effect plot, confirm that the significant factors align with your engineering intuition, and iterate by adjusting factor levels within safe bounds. The method delivers faster insight, lower material waste, and a clear path to an optimal setting that can lift yield by several percent – as the injection‑molding example demonstrated with a 5 % gain.