Mold Validation Procedure Part-3

Option 4 Description:
This option does not evaluate the inherent material variation. It does investigate the differences attributed to molding the same part in a number of colorants along with the complication of using regrind. The Qualification study requires a representative virgin resin lot of the mean specification. In addition, it requires regrind and adequate amounts of all colorants which will run on the mold.

All the steps during the procedure that involve intimate contact with the blow molding machine are to be done by a qualified injection molding machine operator.

Option 4 Procedure:




Shape of Sample Population for Qualification Study

• Two histograms (6 and 7 classes) indicate that the observations are likely follow a Normal Probability Distribution.
• The 9-class histogram fails the two-question test (see Histogram Construction on page 67) as the population appears to be slightly skewed toward the high end of the distribution.
• Assume that the parent population is indeed normally distributed (represent a more conservative estimate) and proceed with the Capability Analysis (next page).












4.6 Verification (30-Day Run)

Purpose:

The purpose of the Verification study is to determine if the process can sustain key part dimensions within specified tolerances. The Verification consists of monitoring 30 days of production. During this time product quality measures must be met; i.e., Cr, Tz & Cpk.

All the steps during the procedure that involve intimate contact with the blow molding machine are to be done by a qualified blow molding machine operator.
Confirm quality through warehouse checks, trade surveys, and consumer comments.
If all criteria is met during a 30 day time span, the mold is validated.
Procedure:
4.6.1 Begin shipments to the trade, assuring individual component quality using established supplier/plant procedures.
4.6.2 Confirm quality through warehouse checks, trade surveys, and consumer comments.
4.6.3 If all criteria is met during a 30 day time span, the mold is validated.



5.2 CONTROL CHART INTERPRETATION

THREE BASIC RULES

C, P, R


C If any data points are outside the control limits, treat them as a special cause.
Caution: With 3 sigma limits, 3 out of 1000 times will not be a special cause,
but 997 times out of 1000 it will be!


P Since the data should be normally distributed about the mean, look for any non-normal patterns.
The easiest way to do this, is to divide the distance between the UCL and the mean or the LCL and the mean into 3 equal parts or zones.
- If about 68% of the data points fall within the first zone above and below the average, then there are no special causes.
- If there are more than 68% (say 90%) or less than 68% (say 45%) then there are special causes acting on the process.


R If there is a consecutive run of 7 points in a row above or below the mean,
treat the 7 or more points as a special cause, i.e., the process has "shifted".
Think of it as flipping a coin and getting 7 heads in a row.
A very unlikely event!! (Actually, only 8 chances out of 1000!)

5.3 HISTOGRAM CONSTRUCTION


Terms:
N - number of observations.
M - smallest unit of measure.
R - range= high - low.
K - number of classes (square root of N with K_12)
H - class width (R/K).

What:
Histograms are "snap-shot" pictures of a system's output over a defined period of time and shows central location (average), variability (spread), and distribution (shape).



Shape:
To determine if the shape of the histogram is Normal (bell shape curve) as a prerequisite to the capability analysis, answer the following two questions:

1. Does the highest bar of the histogram contain the mean of the sample population?
2. Do the bars on each side of the mean get equally smaller as you move away from the mean?

• If you answer "yes" to both questions, assume a Normal population.
• If you answer "no" to at least one question, determine how non-Normal the population may be, e.g., skewed or truncated. At this point, you may need a computer software capable of performing the capability analysis with a clearly non-Normal population.


5.4 CAPABILITY ANALYSIS COOKBOOK


1. MUST have determined that the Process is stable via Control Charts.
2. MUST have determined that the population is Normal via Histograms of observations.