Thursday, July 07, 2005

"By the Numbers" @ "The Vision Thing"

The following Blogpost is provided by my Blogpal Ethan.

I was asked to provide some input.

Go to the post below before reading my response.
Quality By the Numbers

Below is my response:
E @ TVT,

2 cents worth – PUNT!

2 dollars worth,

I hate to do this (the correct way is to meet with Ethan and clarify request) but I’m going to make some assumptions for Learning sake (I should not ask questions without additional knowledge, aghh who cares people do it all the time).

1. There is a process/system for performing this task, it was stable at one time and the process is no longer capable of meeting current and future customer requirements.
2. There is a process/system for performing this task, was stable prior to the previous three batches. If this process is Stable, informing the worker about the defects and to run additional batches is TAMPERING to the Big Dog!
3. There is a process in place and we have no idea if it’s stable (sometimes it works, sometimes it doesn’t, we just change as we go…. and the QAs have no idea of the requirements since they think Cs are OK.
4. There are NO prevention type process measures at critical points in the production process to let worker no there is a problem without waiting for inspection (1-10-100 rule), think of this as dollars, $100 means end of process, $10 means caught in mid production and $1 defect caught at beginning of process. Unknowable cost if product is shipped to customer(s).
5. QAs (Assurance) are really QCs (Control). Jack Welch would fire; qg would ask Management to inquire into process/system since they own it. There is a cost for being stupid and the “Leaders” of this company are to blame. P/S – I don’t want to hear any crap that it’s the worker’s fault.
6. Company is measuring to customer specification without knowing if process is stable (must be stable before capable of meeting customer specs). So many do this mistake, use/hire a Statistician (I’m not one, just an applicationist).

There is more but that would be in the 200 dollars range.

Where does one start with such a mess? Do we start over and Design a new process using a “Methodology” that must include all internal players as well as customer requirements and suppliers input? Example: qgMethodology:

1. Define and initiate the Scope and Plan the project using Project, Process, Technical, Management Tools & Principles
2. Then we thoroughly must understand and specify process owners, users, suppliers and customer requirements (Critical to Quality Requirements).
3. Develop a High-Level Design based on CTQs.
4. Develop “detailed” Design and any corresponding code, work plans or procedures.
5. Verify Design meets CTQs and implement new design with associated process and quality measures.

The previous example is probably an unlikely solution or starting point for this company.

qualityg is going to go out on a limb and say items 2-6 above are all involved.

Tampering is action on the process/system without identifying the “probable” cause(s) of the trouble.

A process may be stable (i.e., B’s results are within control limits), yet turn out faulty items and mistakes (Stable can mean good or bad, depending on the situation). To take action on the results of a stable process in reaction to production of a defective machine part or a mistake is to tamper with the process. This can be a “very” costly mistake (1-10-100 rule).
Please remember that customer specs are not control limits. Severe losses occur when a process is continually adjusted one way and then the other to meet specifications.
In order to improve a stable system the complete process/system must change, not just a part of the process. One must understand Variation.

Let me recommend Dr. Deming’s “The New Economics.” – to learn more about Tampering (chapter 9 – Funnel Experiment) and Variation (chapter – 10 – Lessons in Variation). Deming Books

Hope this helps,

No comments: