I need help verifying formula's in a database for LCL UCL Cp Cpk PPk

[chiragtilara]

Hello everyone:

I have developed a MS access database which captures data from our legacy LIMS (I would not really say it is LIMS). The visual basic code behind the scene does all statistical calculations like determining LCL;UCL;Cp;Cpk; Ppk and Normality Studies.

Since I do not have sophisticated statistical program which can automatically figure out these parameters, I need your help to find out whether my calculations are correct.

I have attached a spreadsheet having 3 products and their viscosities by batch number. I treated them as individual measurements (for statistical calculations) and plotted Run Chart, Range Chart (individual), calculated LCL, UCL, Cp, Cpk, Ppk and plotted a normal probability plot using a equation that I found in one literature. Also the LSL and USL are 250,000 and 600,000 cP respectively

It would be great if you can punch this data into software you have (minitab etc.) and send a snap shot of results. It would be great if the snapshot shows important parameters like control limits; capability and normality parameters.

I would like to compare the results that you got against mine. If any one is interested in my results let me know.

Thank you in advance.

Chirag Tilara

[Miner]

Here you go.

[chiragtilara]

Hi Miner:

Thanks for your prompt reply. My results are comparable against those of yours. I am assuming that you use Minitab. Attached are results from my database. Let me know what you think, I should change to make the database better.

Thanks

Chirag

[Darius]

I checked the report

Maybe I miss something but looks OK, there is a small issue, altho I may be agree with you..., it's not usual to represent target and spec limits in a control chart.

The reason of this idea (not showing specs on a SPC Chart) is because the control has nothing to do with specs, the SPC chart show your process behabiur, the variation and whereabouts.

Looks like SPC take in account that you are centered on specs or leave it to the capability index. You can even check on patterns and nobody will tell you as a pattern out of specs (We are talking on SPC, not precontrol).

[Jim Wynne]

In Reply to Parent Post by Darius

I checked the report

Maybe I miss something but looks OK, there is a small issue, altho I may be agree with you..., it's not usual to represent target and spec limits in a control chart.

The reason of this idea (not showing specs on a SPC Chart) is because the control has nothing to do with specs, the SPC chart show your process behabiur, the variation and whereabouts.

Looks like SPC take in account that you are centered on specs or leave it to the capability index. You can even check on patterns and nobody will tell you as a pattern out of specs (We are talking on SPC, not precontrol).

Darius,

In my experience, it's fairly common to show spec limits on control charts. In fact, seeing the spec limits can obviate the need for Cpk calculations. At the same time, with a relatively wide specification band and a well-centered process, the x-axis might get too big. At any rate, they usually do no harm, and might be useful. We should assume that people using control charts have been trained and understand the difference between control and specification limits.

[Darius]

JSW05 Wich stat software do you use?, maybe I have to change.

[Jim Wynne]

In Reply to Parent Post by Darius

JSW05 Wich stat software do you use?, maybe I have to change.

I don't use any at the moment; I was just making the point that the inclusion of specification limits is not something that needs to be avoided unless there's a special reason (such as an overly spread-out x-axis) for doing so. Lots of people use Excel or paper or some other means where it's possible to include the spec limits; I don't know how the various commercial packages treat the subject.

[dfirka]

Regarding the results shown by Miner done with minitab:

Does anyone knows which formula is being used for the standard deviations in the example?

1) At product A, Minitab gives Overall StdDev: 78265, but using Excel and also in Chirag's Soft, we get 78095. The other products show similar discrepancies.

2) In addition, I dont understand the value "within variation", that I suppose should be based on Moving Ranges... but if I calculate Avg(RMov)/d2 for Product A, it's 65120, very different from Minitab's 73901.

Maybe Minitab is using another formula (for 1) or arrange the data in a different way (for 2)
Any ideas?
Thxs
Daniel