Cpk Calculation - Tolerance to maintain 1.33 & 1.67 Cpk respectively

[gianni]

My customer wants us to hold 1.33 & 1.67 Cpk on 2 dimensions respectively. We have not made any parts yet and I have to setup our inspection paperwork to reflect a tolerance that will hold the 1.33 & 1.67 Cpk and submit to them. How do I know what tolerance I will need to hold to maintain the 1.33 & 1.67 Cpk respectively? Help!

 

[Miner]

Re: Cpk Calculation

Assuming your process will be centered, I will use Cp to simplify the math.

Cp = USL - LSL / 6 StdDev

USL - LSL = Total Tolerance = Cp * 6 StdDev

Tolerance = 1.33 * 6 * StdDev or 1.67 * 6 * StdDev

This is all based on the presumption that your customer will allow you to modify the tolerance to meet your process variation. Normally the tolerance is fixed and you must improve the process to meet it.

 

[bobdoering]

Re: Cpk Calculation

My customer wants us to hold 1.33 & 1.67 Cpk on 2 dimensions respectively. We have not made any parts yet and I have to setup our inspection paperwork to reflect a tolerance that will hold the 1.33 & 1.67 Cpk and submit to them. How do I know what tolerance I will need to hold to maintain the 1.33 & 1.67 Cpk respectively? Help!

It does kind of depend on the process, too. What is the process for the dimensions?

 

[brahmaiah]

Re: Cpk Calculation

Assuming your process will be centered, I will use Cp to simplify the math.

Cp = USL - LSL / 6 StdDev

USL - LSL = Total Tolerance = Cp * 6 StdDev

Tolerance = 1.33 * 6 * StdDev or 1.67 * 6 * StdDev

This is all based on the presumption that your customer will allow you to modify the tolerance to meet your process variation. Normally the tolerance is fixed and you must improve the process to meet it.

Right indeed,but first you have to find the std.deviation of the process by a trial.And the process trial should be under statistical control.SPC is a complicated subject easily misunderstood by many.
V.J.Brahmaiah

 

[Stijloor]

Re: Cpk Calculation

Right indeed, but first you have to find the std.deviation of the process by a trial. And the process trial should be under statistical control. SPC is a complicated subject easily misunderstood by many.
V.J.Brahmaiah

Can you provide an example of such a "trial?"

By the way, our friend Miner is an expert at this, and I doubt very strongly if he misunderstands this subject.

Stijloor.

 

[AMITRAAJSHARMA]

Re: Cpk Calculation

I agree with Bramiah - you would have to perform trial runs before you get into the GDNT bit for setting tolerancing levels for the products.

 

[brahmaiah]

Re: Cpk Calculation

Can you provide an example of such a "trial?"

By the way, our friend Miner is an expert at this, and I doubt very strongly if he misunderstands this subject.

Stijloor.

My suggestion of trial run is not adressed to Mr.Miner.I have no doubt abovt his expertise of SPC.The suggestion was for the original person who wanted some help on this.Being an expert Mr.Miner does not need any example from me.
V.J.Brahmaiah

 

[Manoj Tripathi]

Re: Cpk Calculation

Hi,
As you mentioned there is no history available for this process. In this sese you need to performs trials using the recommended machine settings. use control chart to see whether the process is in statistical control or not. Once it is stable find out the UCL, LCL, process average and sigma. This UCL and LCL will provide 6 sigma tolerence band with Cpk=1.00

For Cpk = 1.33 increase the tolerence band by 2 sigma. (USL = UCL + sigma) and (LSL =LCL- sigma).

For Cpk = 1.67 increase the tolerence band by 4 sigma. (USL = UCL + 2*sigma) and (LSL =LCL- 2*sigma).


Manoj

 

[bobdoering]

Re: Cpk Calculation

If you are precision machining, and controlling your process correctly, you would set your control limits at 75% of your tolerance for 1.33 and 60% of your tolerance for 1.67.

 

[jperez25]

Re: Cpk Calculation

My customer wants us to hold 1.33 & 1.67 Cpk on 2 dimensions respectively. We have not made any parts yet and I have to setup our inspection paperwork to reflect a tolerance that will hold the 1.33 & 1.67 Cpk and submit to them. How do I know what tolerance I will need to hold to maintain the 1.33 & 1.67 Cpk respectively? Help!

Gianni, I am not sure if you have already discussed this with your customer, but it is very unlikely (at least in my field in automotive) for a customer to change a product tolerance for the sake of showing a capable process.

The rest of the technical how-to is well explained by the other people in the forum:

1) confirm that you have a capable measurement system for this characteristic
2) run a process trial to understand the variability of this characteristic
3) confirm stability, normality, remove outliers
4) calculate the standard deviation
5) calculate what dimension tolreance you should have if Cpk had to be 1.33 or 1.67

One final comment (hopefully not too disruptive - this is generally quite confusing) is why your customer is asking you to demonstrate Cpk if the process is new. Are you sure it is not Ppk?

Javier