Spec Limits in SPC charts

[JasnahKholin]

Hi All,

Is it OK to include the spec limits on control charts? This was found by one of my peers during an internal audit and is apparently a common practice that has been in place for years here. My peer insists that this is an NC and that spec limits are not allowed to be included on control charts. Production's opinion is that it is relevant to include on the charts because it affects what OCAP operators follow depending on if you are outside control limits vs. outside spec limits. For example, something that is outside control limits but inside spec limits may still be a good product but if you are outside spec limits then it may require rework or scrap. I cannot find anything in the AIAG SPC manual that states that spec limits can't be included on the charts. Is this a common practice anywhere else?

We are both internal auditors in the quality department. We are certified to ISO9001, IATF16949 and AS9100.

Thanks,

 

[Jim Wynne]

I can't think of a reason that anyone would not want to see the spec limits on an SPC chart. The idea that it's somehow considered a bad idea is frankly ludicrous. The whole idea is to be able to compare process behavior with the specifications.

 

[Matt Savage]

The job that control limits do is completely different than the job that spec limits are used for. Are you working with an Individuals & Moving Range (X-MR) chart of X-bar?I ask because if it is X-bar chart, I would never put spec limits on the chart.
Lets assume you are working with Individuals data. One negative consequence is that if it is "In Spec" yet out-of-control, why should I change anything? I mean, the measurement is good, correct? In short, the idea of improvement may not occur when specs are included on the control chart.

 

[Miner]

Another issue with adding spec limits to sub-grouped control charts is that it can mislead an operator into thinking that because the point was inside the specification that all is well. The problem is that the point is an average, and individual results may well be out of spec.

 

[JasnahKholin]

I ask because if it is X-bar chart, I would never put spec limits on the chart.

These are not X-bar charts. Each data point is a measurement from that specific part.

Another issue with adding spec limits to sub-grouped control charts is that it can mislead an operator into thinking that because the point was inside the specification that all is well. The problem is that the point is an average, and individual results may well be out of spec.

Our MES system stops the machine from running when there is a violation of WECO rules for the feature being measured and operators/engineers must address the issue before continuing production. The OCAP processes also have different procedures on what operators should do if the point is outside the UCL/LCL but still inside spec limits vs. a point which is either above or below the spec limit.

For example, if a hole diameter is outside the UCL/LCL the OCAP is to change the tool but product is still OK to use. A hole diameter smaller than LCL and LSL has an OCAP to rework while a diameter larger than UCL and USL would mean the product is scrap.

 

[AMIT BALLAL]

Use of specification limits on a control chart leads to tampering (over-adjustment). And it causes more problem than benefit.
So I would suggest to stop such a practice.

As you mentioned, every time the hole diameter point is outside a control limit, considering such a problem due to special cause related to tool and change of tool seems like tampering. I recommend revisiting the procedures in such a case.

@Miner will be able to give more insights.

 

[Jim Wynne]

Another issue with adding spec limits to sub-grouped control charts is that it can mislead an operator into thinking that because the point was inside the specification that all is well. The problem is that the point is an average, and individual results may well be out of spec.

Here we're talking about incompetent operators. Also, before subgroup results can be charted, the individual samples must be measured. Anything anomalous should be evident when the parts are measured.

 

[Miner]

Here we're talking about incompetent operators. Also, before subgroup results can be charted, the individual samples must be measured. Anything anomalous should be evident when the parts are measured.

Just because all of the parts in a subgroup are in spec does not mean that all parts produced at that time are in spec.

If I recall, you have a machining background which likely has a very tight distribution. My background includes processes with a lot of variation, so I have seen tis occur.

 

[Ron Rompen]

it is definitely NOT a nonconformance to put specification limits (or any other limits you like) on a control chart. A nonconformance is a failure to comply with the standard, and to the best of my knowledge there is nothing in any of the standards you mentioned that would prohibit this.

There may be valid reasons not to do it (several have been cited) however I always like to have the specification limts AND the control limits shown on charts - that way I can determine if parts are still acceptable to print spec, even if they have gone out of control.

 

[Jim Wynne]

Just because all of the parts in a subgroup are in spec does not mean that all parts produced at that time are in spec.

I suggested nothing of the kind. I said that the majority of anomalous conditions will be seen when parts are being measured. I will also reiterate that arguments against spec limits on control charts assume operator incompetency.

If I recall, you have a machining background which likely has a very tight distribution. My background includes processes with a lot of variation, so I have seen tis occur.

I've worked in machining, metal stamping, sheet metal, precious-metal plating, and injection molding molding job shops. I've also worked for OEMs in manufacture of toxic-gas monitoring systems and motorcycles.