How to Start SPC from the ground up!

  • Thread starter Mary - QA Manager
  • Start date
M

Mary - QA Manager

I am starting the SPC project at my facility. The only direction I have is to use "rainbow" charts. I am working on these charts but was wondering if anyone here has taken on such a task and how they completed it? I am using excel at the moment and asking for Minitab.
Thanks!
 

Project Man

Involved In Discussions
Mary,
We currently have a very successful SPC program in our shop. It has taken 3 years to get it where it is today. It took two years to completely change the culture on the shop floor (and that was starting with a pretty good culture). That is where you have to start and that is where it will or won't be successful (The software you use just makes it easier to use the data collected)
We used "Quality starts at the machine" as the mantra to get get culture changed. It takes diligence from the supervisors and buy-in (read $$) from upper management.

Rainbows can help (but don't make it work), Minitab can help (but doesn't make it work), using a software specifically designed for SPC can help (but doesn't make it work). The ONLY thing that makes it work or not is the culture on the floor.

We use: 1. A great shop culture (45 employees on the floor), 2. Formulas in excel will graph output., 3. Quality analysis at the beginning, in-process and at the end.
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
It's been an ongoing implementation and evolution for me over the past 23 years on SPC. I'd say an important starting point is Dr. Deming's Red Bead Experiment. I have a video link, and several other useful SPC documents at http://www.efcog.org/wg/esh_es/Statistical_Process_Control/

By the way, I've just used Excel over the years, starting with Excel 4.0 with dbaseIII+ to do the data processing. Now I primarily use Access for the data processing.
 

Miner

Forum Moderator
Leader
Admin
Also, take a VERY hard look at where SPC will actually help you. I have seen poor implementations back in the 80's, where Ford came in and forced the supply base to use SPC everywhere, and it was a disaster.

There are definitely processes that can benefit from SPC. These are processes that are historically out-of-control and producing rework/scrap in that operation or downstream operations. On the other hand, there are well behaved processes that will see little benefit, or processes that have automated controls already in place.

Where SPC will be a benefit, take a close look at the TYPE of SPC chart used and the rational subgrouping. Xbar/R charts are not always the best choice.
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Also, take a VERY hard look at where SPC will actually help you. I have seen poor implementations back in the 80's, where Ford came in and forced the supply base to use SPC everywhere, and it was a disaster.

There are definitely processes that can benefit from SPC. These are processes that are historically out-of-control and producing rework/scrap in that operation or downstream operations. On the other hand, there are well behaved processes that will see little benefit, or processes that have automated controls already in place.

Where SPC will be a benefit, take a close look at the TYPE of SPC chart used and the rational subgrouping. Xbar/R charts are not always the best choice.

Definitely a true story about Ford. My father worked for a supplier to Ford and with SPC crammed down their throats, there was very little value seen. Ford could force entities to make SPC charts, but without them actually being "used" by Ford or the suppliers or by folks with knowledge, they were just pretty pictures on the wall.

I will state that even if you currently have a stable process, doing SPC on it can be an ounce of prevention - so you can detect (inexpensively) if the process has an upset. I liken it to all of the warning lights you have on the dashboard on your modern car. Usually at least two dozen lights monitoring various things, and likely most of them will never come on during the life of the car. However, by monitoring these critical measurements, you can save a big maintenance bill later.
 

Paul Simpson

Trusted Information Resource
I am starting the SPC project at my facility. The only direction I have is to use "rainbow" charts. I am working on these charts but was wondering if anyone here has taken on such a task and how they completed it? I am using excel at the moment and asking for Minitab.
Thanks!

Hi, Mary. The only suggestion I'd make is that you be absolutely sure what problem it is you are trying to fix. It is very easy to go a long way down the 'use "rainbow" charts' road without knowing exactly where you are trying to get to.

The question I'd go back with is: 'What do you want SPC for and how will I know when it has given you what you need?'
 
G

Grimaskr

I agree with all of the above, but will add my own subject to really look at: scope.

Before you actually start, try to imagine the end goal of your SPC implementation project and get a realistic feel for how big the scope is going to be.
  • Will you be doing SPC on process inputs (like machine temperatures, pressures, speeds)?
  • Will you being doing SPC on outputs (dimensions of fabricated parts, functional testing on parts)?
  • How many inputs/outputs do you think you might need in your organization?
  • How will you get the measurements into a computer and what might be some of the problems/restraints based on resources and formats?
  • Are your measurement systems good enough to even use for SPC (e.g., might be good enough to tell the difference between good/bad parts, but not good enough to zoom in further)?
  • Who will have the expertise to analyze these charts?
  • Will you expect production to react live to out of control situations?
  • How many Western Electric rules do you plan to use and react to on the production floor? And will your staff need training?
Thinking about these kinds of questions up front can really help your planning for rolling SPC implementation out in a successful manner.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Here are the steps I recommend:

1. Develop the total variance equation
2. Determine which variance factors are adjustable, which are noise, and which can be set as a constant.
3. Minimize the variation of each of the participating variables - get the process in a steady state and capable.
4. Accurately determine the correct distribution of each of the remaining variances
5. Determine which of the remaining adjustable variance factors you are going to chart
6. Pick the correct chart to evaluate each of those variance factors (variables)
 

stevegyro

Involved In Discussions
Mary, 'Grimaskr' above has really defined very well, what most 'management teams' have a challenge with. Best regards. Pls. Keep us posted ;-)
 
Top Bottom