Injection Moulding Minitab DOE - All P Values are 0

[Mikeybhaba]

Hi All. I am very new to both DOE and Minitab. We have run a 4 factor design on the overall length of a part. My query is that all the factors seem to be coming up as significant on the Pareto and all the P values equal 0.000


Estimated Effects and Coefficients for OAL D1 (coded units)

Term Effect Coef SE Coef T P
Constant 0.704323 0.000896 786.03 0.000
Mould T 0.017032 0.008516 0.000896 9.50 0.000
Melt T 0.019819 0.009909 0.000896 11.06 0.000
Inj speed 0.010278 0.005139 0.000896 5.74 0.000
Pack press 0.009071 0.004536 0.000896 5.06 0.000
Mould T*Melt T -0.018512 -0.009256 0.000896 -10.33 0.000
Mould T*Inj speed -0.010953 -0.005476 0.000896 -6.11 0.000
Mould T*Pack press -0.011035 -0.005518 0.000896 -6.16 0.000
Ct Pt 0.010853 0.005738 1.89 0.068


S = 0.00566714 PRESS = *
R-Sq = 93.45% R-Sq(pred) = *% R-Sq(adj) = 91.81%


Analysis of Variance for OAL D1 (coded units)

Source DF Seq SS Adj SS Adj MS F P
Main Effects 4 0.0087080 0.0087080 0.00217701 67.78 0.000
Mould T 1 0.0029010 0.0029010 0.00290103 90.33 0.000
Melt T 1 0.0039278 0.0039278 0.00392782 122.30 0.000
Inj speed 1 0.0010563 0.0010563 0.00105633 32.89 0.000
Pack press 1 0.0008228 0.0008228 0.00082284 25.62 0.000
2-Way Interactions 3 0.0058442 0.0058442 0.00194805 60.66 0.000
Mould T*Melt T 1 0.0034268 0.0034268 0.00342682 106.70 0.000
Mould T*Inj speed 1 0.0011996 0.0011996 0.00119960 37.35 0.000
Mould T*Pack press 1 0.0012177 0.0012177 0.00121774 37.92 0.000
Curvature 1 0.0001149 0.0001149 0.00011492 3.58 0.068
Residual Error 32 0.0010277 0.0010277 0.00003212
Pure Error 32 0.0010277 0.0010277 0.00003212
Total 40 0.0156948

Can somebody explain the reason for all the zero P values please? Any help would be extremely helpful!

 

[Miner]

The format is illegible. Can you paste it into a Word or text file and attach it?

 

[Mikeybhaba]

Hi Miner, thanks for getting back. I have attached the info in the Word file.

 

[Miner]

Thanks. That is much easier to read. What type of design was this? A full factorial, 1/2 or 1/4 fraction? What resolution?

Two things: First, you should investigate Observation #29. It has an unusually large residual, and may affect the results of your analysis.

Second, after confirming observation 29, you should remove the Center Point from the model. You didn't state what alpha you are using, so I assumed 0.05. That means that curvature is not significant and the center point can be removed. This may change the p-values of the remaining terms.

It is not that unusual for an injection molding process to have these factors and a number of interactions show as significant. Sometimes, you have to make a call on whether something is of PRACTICAL significance. That is, it may be statistically significant, but the practical effect is too small to be of practical use in the process.

 

[Mikeybhaba]

Hi Miner, thanks again! We ran a 1/2 fraction with a resolution of IV and you were correct in the assumption that the alpha used was 0.05.

We have since ran a second DOE. This time we ran a 3 factor full factorial. From our practical analysis of the process and the experience of the first DOE, we choose holding pressure, injection speed & mould temperature.

This time we have seen the mould temperature to be critical for this particular part. This is across all eight cavities of the mould. My query now is with a multi-cavity mould, what is the standard method in Minitab for obtaining nominal settings for the validation runs. Is one particular cavity choosen or does Minitab have some method of assessing multiple cavities to obtain a nominal settings?

 

[Miner]

I would treat the multiple cavities as REPEAT measurements. Enter them across the row for a particular experimental run. That is, if you have 6 cavities, enter the repeats across six columns for that experimental run.

Then use the Pre-process Responses for Analyze Variability. This will create three new columns containing the Mean, Standard Deviation and n for each experimental run. You can then analyze the mean and the variability.

 

[Mikeybhaba]

Hi Miner, Thanks for all your help on this query!!

 

[supreecha]

Low Pressure Die Casting Minitab DOE - All P Values are 0
Box-Behnken Design . (4 Factors 2 Levels with Center Point.)