posted 18 May 1999 04:33 PM               

quote:

---

Our company recently obtained QS9000 certification. Our registration company cited statistical techinques (lack of them)as a weakness in our system.

---

Typical. To see the lack of a tool as a weakness without understanding the process is absurd. For example, to drive a nail, I need a hammer. I have many types to choose from, so I select the best tool for the job. But, I digress.

Under the assumption that the statistical techniques you wish to add to your system add value, consider this.

All processes must have inputs (flow rates, rates per hour, pressure, etc). You could start with your inputs and apply statistical techniques there. I would not begin to assume to assign a particular technique without knowing more.

All process must have outputs (weights, content purity, etc). These can also be assigned a statistical technique that adds value.

Start with inputs to the process. Identify Key Input Variables (KIV) and determine which have a direct affect on your Key Output Variables (KOV) (an Ishikawa or affinity diagram works nice for this).

Next, determine which statistical technique is applicable to the particular KIV. For example, if pressure is determined to be a KIV, you could monitor that with an X-MR chart by measuring the pressure once per hour. You could monitor pressure by a subgroup of four readings every 15 minutes (or every four hours, if applicable), calculate Xbar and range and use the typical Xbar-R chart.

Once the KIVs are determined and brought under control, do the same thing with the KOVs.

BTW, being a bulk processor, I would wager you are already using a statistical technique: sampling. But, I could be wrong.

Regards,
Don

------------------
Just the ramblings of an Old Wizard Warrior.

Check Out dWizard's Lair:
*** Dead Link Removed ***