Why is Pre-Control not a Process Control tool?

The rules of pre-control address adjust process when two yellow points in a row (same zone) and stop the process and investigate when two yellow points in a row ( opposite zone ). --- it seems that pre-control can tell us what to do on the process when some circumstances occur.
But SPC says it is NOT a process control tool, why?

Re: Why pre-control is not process control tool?

Because, at least according to what I've learned about pre-control, the "control limits" are set arbitrarily from the specifications, not from the data.

I've had success in setting average and control limits from as few as five points. So why would someone even need "pre-control" (I'm sure someone will answer that ).

jelly1921 said:

But SPC says it is NOT a process control tool, why?

Click to expand...

There is a great controversy about this thema, the radicals say so, there are the people of the middle that say, ok, if it fit is a tool, and there are the precontrol followers.... I don't think ever will be a consence.

The argument is that Precontrol is set as Steve said, and that it don't look for the improvement of the process.

As Steve said there are some instances where it could apply, as far as I can see:

*When the variation inside the specs limits don't affect the quality of the product, nor the capacity of the process, desto as a result we don't case to reduce variation as long as specs are ok.
* When we can't reduce variation, because other process determines the variation of the process, and accordingly to it's variation the specs where fixed.
*To fit on specs is far more important (critical) than reduce variation.

Maybe Atul could enlight us, as I remember he show us great information about precontrol

My observation that the more vocal proponents of the various methods have chosen to promote their preference by denigrating the other methods. And they do this for very self serving reasons: pay me, buy my book, follow me. This approach is driven by pure greed and provides no insight to those of us who need to control our processes and improve our quality. But it sells professional journals and magazines and helps get various people tenure!

I have used Shewhart SPC, Precontrol, Mistake proofing, alternative SPC (EWMA, 3D charts, etc.) simultaneously (uniquely for different processes) and quite successfully. Each of these approaches are useful IF they are applied to the correct situtation. None of them is a universally applicable tool.

For example, Shewhart charts work well when you have a homogenous process stream. Precontrol works well when there are known causes and corrective actions for movement from target. (it's not tampering when you know what you are doing - a lot of machining and molding processes have this).

Re: Why pre-control is not process control tool?

Steve Prevette said:

Because, at least according to what I've learned about pre-control, the "control limits" are set arbitrarily from the specifications, not from the data.

Click to expand...

When you are not certain of the kind of process variation you are dealing with, it is an adequate tool for ensuring the process is controlled based on the premise you are keeping it from making bad parts, and the signals you are looking for are process points heading too close to the specification. It could very well reduce overcontrol - especially if you process knowledge is too limited to understand long term process variation (the kind of understanding that allows you to assure your sample sizes, frequencies, distributions, etc. are correct for proper control.)

Note the name: pre-control. Get it?

That's the quick overview - a lot can be added to that....

In my mind, Shewhart charts are process monitoring, and pre-control provides more signals for adjustment or control. But, that's because I am contrary to the pack.

Re: Why pre-control is not process control tool?

Steve Prevette said:

Because, at least according to what I've learned about pre-control, the "control limits" are set arbitrarily from the specifications, not from the data.

I've had success in setting average and control limits from as few as five points. So why would someone even need "pre-control" (I'm sure someone will answer that ).

Click to expand...

Hello
I agree with you but I have one question:
For the new or unstable process is it correct to use at first pre-control for a while till the CP&CPK near 1 or 1 than to take a sample and build a control chart X bar or IMR?

Thanks

Oleg Volfson
CQE,CQM

Re: Why pre-control is not process control tool?

oleg_volf said:

For the new or unstable process is it correct to use at first pre-control for a while till the CP&CPK near 1 or 1 than to take a sample and build a control chart X bar or IMR?

Click to expand...

imho, NO, you can't. precontrol focus is to mantain the status quo, not to obtain a better Cp/Cpk, not the reduction of variation but to keep it at aceptable level.

Re: Why pre-control is not process control tool?

oleg_volf said:

Hello
For the new or unstable process is it correct to use at first pre-control for a while till the CP&CPK near 1 or 1 than to take a sample and build a control chart X bar or IMR?

Click to expand...

I think the misunderstanding is that "Pre" doesn't really mean before. In other owrds Pre-Control was not intended to be used before SPC or before the process is under statistical control. The two techniques are not intended to do the same thing. A further misunderstanding is that a Cpk of 1 doesn't mean the process in statistical control. you can have a Cpk < 1 and be in control and you can have a calculated Cpk>1 and not be in control.

There is also not a 'correct' choice that is always 'correct'. There are choices that will work for you and choices that won't.

For example Pre-control is intended for processes where the largest component of variation is NOT piece to piece and the causal mechanisms leave obvious patterns and corrections are known. Classical Shewhart SPC charts (such as I, MR and Xbar, R) are approapriate where the largest component of variation IS piece to piece (otherwise known as a homogenous process stream). In neither case is the capability of the process relevent to the decision....
Precontrol is a Product control method that will that will enable you to produce in specification parts. (it is NOT based on the Normal distribution anymore than Shewhart charts are). Shewhart SPC charts will enable you to detect if your process is stable and detect when assignable causes occur.

Re: Why pre-control is not process control tool?

Bev D said:

I think the misunderstanding is that "Pre" doesn't really mean before. In other owrds Pre-Control was not intended to be used before SPC or before the process is under statistical control. The two techniques are not intended to do the same thing. A further misunderstanding is that a Cpk of 1 doesn't mean the process in statistical control. you can have a Cpk < 1 and be in control and you can have a calculated Cpk>1 and not be in control.

There is also not a 'correct' choice that is always 'correct'. There are choices that will work for you and choices that won't.

For example Pre-control is intended for processes where the largest component of variation is NOT piece to piece and the causal mechanisms leave obvious patterns and corrections are known. Classical Shewhart SPC charts (such as I, MR and Xbar, R) are approapriate where the largest component of variation IS piece to piece (otherwise known as a homogenous process stream). In neither case is the capability of the process relevent to the decision....
Precontrol is a Product control method that will that will enable you to produce in specification parts. (it is NOT based on the Normal distribution anymore than Shewhart charts are). Shewhart SPC charts will enable you to detect if your process is stable and detect when assignable causes occur.

Click to expand...

Hello
Thank you for your replay.
Maybe I explained myself not too good/
will try again:

1.For the parts in machine shop I have in mass or serial production first I used a C-chart along with pareto to detect 3 or 4 most critical dimensions.Second I measure 50 samples those dimensions and prepare a great IMR chart that works fine in production floor according to the SPC rules we detected and fixed a lot of issues and improved fixtures and instruments.
2.For the new units that entering the production and the units that has small batches of 10 or 15 units, there is not enough data to build any of the Shewart chrts so I intend to do the following:
Use the engineering to define the critical dimension,prepare the precontrol charts for those dimension and for other ones collect the data by registering the results in the report,this way I collect the data,control to avoid the nonconforming and when I have enough data I will move to Shewhart charts.

What do you thing regarding my procedure?

Please if you
have any suggestions for the paragraph 2 I will be more then happy to know?

Thanks
Oleg
CQE,CQM

Re: Why pre-control is not process control tool?

oleg_volf said:

For the parts in machine shop I have in mass or serial production....

Click to expand...

When you refer to "machine shop" if you are precision machining, none of the charting methodologies mentioned so far are correct for controlling dimensions. The X hi/lo-R chart uses the correct statistics for the correct distribution for precision machining.

oleg_volf said:

first I used a C-chart along with pareto to detect 3 or 4 most critical dimensions.

Click to expand...

I recommend following these steps to determine what to chart. For machining, you should be charting one dimension (the tightest) per finishing tool.

oleg_volf said:

2.For the new units that entering the production and the units that has small batches of 10 or 15 units, there is not enough data to build any of the Shewart charts so I intend to do the following:
Use the engineering to define the critical dimension,prepare the precontrol charts for those dimension and for other ones collect the data by registering the results in the report,this way I collect the data,control to avoid the nonconforming and when I have enough data I will move to Shewhart charts.

Click to expand...

Again, based on the type of variation you are tracking, first you need to determine if Shewhart charts are even applicable. They are not applicable in every case, as they have very specific statistical limitations based on the type of variation you expect.