How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

chvrajus said:

1. They are looking for Histogram (Can U please provide me a Histogram chart with the same sample data ??) with limits and Cp&Cpk values

Click to expand...

I see that Bob has created one for you. However, this is a skill you should learn: what software do you have to create a histogram? the formula for Ppk is very straightforward and easy to find so you shoudl be able to calculate this for yourself. Although I would caution that the practice of capability indexes is not always easy and you should take the tiem to study thsi subject. you can always come back here to ask pertinant questions.

chvrajus said:

2. The UCL,LCL & CL only tells you that the readings you have got are in statistically under control or Not,

Click to expand...

True - and you are 'in control'.

chvrajus said:

you have to calculate the Process capability & index then only you will find out that the obtained values are good enough to satisfy the customer specification.CP & CPk value tells that the process is capable of producing the parts per the customer specification ( How can I tell my process is capable and enough to satisfy the customer specification??)

Click to expand...

without goign in to all of the nuances, the simple answer is: your customer should have provided you with a desired - or required - minimum Cpk (or Ppk) value. If your value is greater than that you are capable of meeting the Customer requirements. On the other hand you acn look at the I, MR chart that also includes the specification limits and see that for this data set you are no where near the spec limits. IF you maintain this performance over time you can be certain that you are capable.

chvrajus said:

3. If my data always plot with in the limits, Can I still tighten the limits?? if so how?

Click to expand...

You can do what ever you want. However, that is NOT the intent of SPC. you should never tighten the limits until you have taken specific action to improve the process AND your process is running consistently better than before.

I also strongly urge you to get yourself a couple of very good SPC books and start studying them...

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

I have few questions on your comments.

How much data is good or need to calculate Cpk values. Our is a low volume manufacturing assembly process. We manufacture 25-30 parts per month. Can you give some example how we can show our process is with in the limits and process is capable ( If you need I can give you data)

I didn’t understand your comments on tightening the limits. (You can 'run tighter', but it goes back to what really controls the process. Do you dial it in? If so, with what parameter? If you compress the limits, you can meet them by adjusting the process more)

How you created the histogram chart. Did you used any tool or excel.
If it’s free tool or excel can you give me the steps to create the chart by my own.

Thanks again for your help and inputs.

bobdoering said:

See attached.



It does not look like there is enough data to fill out your distribution - which may be one reason why the Cpk appears to be a tad low (1.2 - 1.33 would have been more comforting) Because you have so much data just below the mean, it skews the Cpk value low. Cp at 1.4 is not too bad.


Was the current data created without adjustment? Was it affected by setups between data points?
You can 'run tighter', but it goes back to what really controls the process. Do you dial it in? If so, with what parameter? If you compress the limits, you can meet them by adjusting the process more. Is that cost effective? Do you want a person (the adjuster) to "become" the process? You can...but at a cost.

Click to expand...

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

chvrajus said:

How much data is good or need to calculate Cpk values. Our is a low volume manufacturing assembly process. We manufacture 25-30 parts per month. Can you give some example how we can show our process is with in the limits and process is capable ( If you need I can give you data

Click to expand...

30 to 50 pcs of data usually generates an adequate histogram.

chvrajus said:

I didn’t understand your comments on tightening the limits. (You can 'run tighter', but it goes back to what really controls the process. Do you dial it in? If so, with what parameter? If you compress the limits, you can meet them by adjusting the process more)

Click to expand...

If you tightened the limits, how would you meet them? The limits are set by your current process results. To make them tighter, you have to reduce the variation, and that requires changing something - raw material, more adjustment, etc. - depends on what influences that characteristic.

chvrajus said:

How you created the histogram chart. Did you used any tool or excel.

Click to expand...

I used QI Macros. Not free, but cheap (qimacros.com).

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

Thanks again –

By showing your histogram chart, how I can explain someone what are the CL and how can I define the process is in control or not?

Can you refer any online link to find info about histogram and to understand the process capability?

Can you share your personal email id?

Thanks

bobdoering said:

30 to 50 pcs of data usually generates an adequate histogram.



If you tightened the limits, how would you meet them? The limits are set by your current process results. To make them tighter, you have to reduce the variation, and that requires changing something - raw material, more adjustment, etc. - depends on what influences that characteristic.



I used QI Macros. Not free, but cheap (qimacros.com).

Click to expand...

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

Bev D said:

calculate the moving range between each value: MR1 = the absolute absolute value of the second value - first value. MR2 = the absolute absolute value of the third value - second value and so on. you will have 29 of these values.
then calculate the average of these 29 values. this is the average moving range, MR Bar

The CL = is the average of the 30 readings.
LCL = average - 2.66*MRbar
UCL = average + 2.66*MRbar

this will ONLY work if the tensile values are in order of PRODUCTION of hte parts. if you dont' know the order of production and you only know the order of the tensile measurements you cannot do the control chart. teh dat must in production time sequence.

But I have to add that I am concerned (no offense intended) that this is a very basic question that can be readily answered by a simpel google search or any SPC book that is lying around. You provide very little background as to the nature of the problem. You seem to have almost no knowledge of SPC and I wonder why you are tasked - feel tasked - to do this on your own (my assumption since you're here)?

Click to expand...

hw did u get 2.66

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

UCL = X_average + 3* SIGMA_within

SIGMA_within = MR/d2(sample_size=2) = MR/1.128

2.66 = 3/1.128

UCL = X_average + 2.66 * MR

But it is without autocorrelation

for an autocorrelated system

SIGMA_within_auto = MR/d2 * (1/(1-r^2)^0.5)

so

UCL = X_average + 2.66* MR * (1/(1-r^2)^0.5)

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

You got the right answer but used the wrong formula.

The actual formula for control chart averages is the average + 3*std error
(std error is the standard deviation of the subgroup averages).

In control charts the std error is calcualted by estimating the total standard deviation (from the within subgroup variation, modified by the appropriate factor: d2 for ranges and c4 for standard deviations) and dividing by the square root of the sample size.
So for Moving Ranges, the formula is: 3(MR_bar/d2)/sqrt

For an I, MR chart, d2 comes from the sample size for the Moving range which is 2; d2 = 1.128. The subgroup sample size is 1; so the sqrt = 1.

3*(MR_bar/1.128)/1 = 2.66MR_bar....

Help im so lost. i am trying very hard to understand what is being asked of me and im to my wits ends.

the questions is as follows:

· Write a 800 words (maximum) explanation of the control limits including the calculations and data used to determine them.
· Discuss the effect of any seasonal factors using the process performance data collected each week.

o Discuss the confidence intervals and their usefulness based on the number of data points.
the collected measure that i observed since week 1 was productivity. it was to observe something going on in my job. so i collected data for Funds Received week 1-5, Funds Obligated Week 1-5, and Labor Hours Used Week 1-5.

the data is below;
Funds Received
Week 1 : 5.3Million
Week 2: 3.6 Million
Week 4: 2.3 Million
Week 4: 4.1 Million
Week 5: 11,000

Funds Obligated
Week 1: 2.6 Million
Week 2. 4.1 Million
Week 3: 3 Million
Week 4: 2.1 Million
Week 5: 21,000

Labor Hours Used
Week 1: 25
Week 2: 31
Week 3: 28
Week 4: 22
Week 5 21

i attached the process flow chart. i wouldnt be asking for help if i didnt really need it and didnt exhaust all other options. the professor gave me an extension because im in the hospital right now (and have been for the past 4 days) but i only have until tonight. can someone please please help me?

even help understanding this would be helpful. or getting the data charts and data done so i can possibly write this. i dont know what else to do. most sites are blocked in the hospital and with me going in and out of the mri rooms and dialysis, i am completely desperate. please help.

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

I can't help you with your specific numbers, but these existing discussion threads have discussions of Control Limits and how to calculate them: Control Limits

Re: How to Calculate UCL (Upper Control Limit) & LCL (Lower Control Limit) & CL?

Bev D said:

calculate the moving range between each value: MR1 = the absolute absolute value of the second value - first value. MR2 = the absolute absolute value of the third value - second value and so on. you will have 29 of these values.
then calculate the average of these 29 values. this is the average moving range, MR Bar

The CL = is the average of the 30 readings.
LCL = average - 2.66*MRbar
UCL = average + 2.66*MRbar

this will ONLY work if the tensile values are in order of PRODUCTION of hte parts. if you dont' know the order of production and you only know the order of the tensile measurements you cannot do the control chart. teh dat must in production time sequence.

But I have to add that I am concerned (no offense intended) that this is a very basic question that can be readily answered by a simpel google search or any SPC book that is lying around. You provide very little background as to the nature of the problem. You seem to have almost no knowledge of SPC and I wonder why you are tasked - feel tasked - to do this on your own (my assumption since you're here)?

Click to expand...

hi,

i'm new to spc.

just curious to know, where do u got 2.66???
is it a fixed value?

thanks