Monitoring a shift of the average in an SPC chart

T

tww11

#1
I have a question that I need cleared up for me. I'm not experienced in SPC at all. I have been to a few seminars and pretty much self taught alot of this material to myself. Here is a scenario I'm trying to figure out.

1. When I'm starting SPC for the first time and I gather say 100 data points. I enter them into my software and get a control chart that shows that I'm within statistical control. As data is entered on an hourly basis, the data points slowly start to shift and over time my average shifts with it indicating a fundamental process change is occurring (say tool wear). Now, my chart isn't going to visually show me an average shift is it? Yeah, i can read the actual numbers and tell that the average is shifting, but since my centerline is always readjusting with new data points, I don't visually see any shift such as line breaks at the centerline up or down.

What I want to know is if in order to see this shift, do you get your process average where you want it and then set the program to use only historical control limits and average. This way the control limits and average never change, but if you start to see data points making a trend upwards or downwards then that could indicate an average shift for which you would need to recalculate controls?


Hope this makes sense. Thanks.
 
Elsmar Forum Sponsor

Jen Kirley

Quality and Auditing Expert
Staff member
Admin
#3
I drew a blank (bringing this back up to the top). I do not really understand the question. Tim or Miner, are you there??
 
B

Bob_M

#4
Are you starting SPC collection on a new project without a well establish average/high/lows?
You might also want to specifically tell the experts what type of SPC chart you are using.
(No I'm not one of the experts)
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
#5
When I'm starting SPC for the first time and I gather say 100 data points. I enter them into my software and get a control chart that shows that I'm within statistical control. As data is entered on an hourly basis, the data points slowly start to shift and over time my average shifts with it indicating a fundamental process change is occurring (say tool wear). Now, my chart isn't going to visually show me an average shift is it? Yeah, i can read the actual numbers and tell that the average is shifting, but since my centerline is always readjusting with new data points, I don't visually see any shift such as line breaks at the centerline up or down.

What I want to know is if in order to see this shift, do you get your process average where you want it and then set the program to use only historical control limits and average. This way the control limits and average never change, but if you start to see data points making a trend upwards or downwards then that could indicate an average shift for which you would need to recalculate controls?
I think the answer is in the "tool wear" comment. What is your process?
 
G

Geoff Withnell

#6
I have a question that I need cleared up for me. I'm not experienced in SPC at all. I have been to a few seminars and pretty much self taught alot of this material to myself. Here is a scenario I'm trying to figure out.

1. When I'm starting SPC for the first time and I gather say 100 data points. I enter them into my software and get a control chart that shows that I'm within statistical control. As data is entered on an hourly basis, the data points slowly start to shift and over time my average shifts with it indicating a fundamental process change is occurring (say tool wear). Now, my chart isn't going to visually show me an average shift is it? Yeah, i can read the actual numbers and tell that the average is shifting, but since my centerline is always readjusting with new data points, I don't visually see any shift such as line breaks at the centerline up or down.

What I want to know is if in order to see this shift, do you get your process average where you want it and then set the program to use only historical control limits and average. This way the control limits and average never change, but if you start to see data points making a trend upwards or downwards then that could indicate an average shift for which you would need to recalculate controls?


Hope this makes sense. Thanks.
Yes it makes perfect sense. The goal of the control chart is to separate random variation from non-random variation. You want to be able to tell if a given change is just "noise" or is something actually going on with your process. So you have run a 100 data points or so, and your process is in control. So this average and these control limits are what the process "normally" does, when it is in control. So set your software to use these numbers, and measure any trend or shift accordingly. Now, when do you recalculate the mean and limits? When you have a change that you understand, and that is in a positive direction, so this in now where you want the process to be running normally. If there is a shift that you don't understand (don't know the root cause) or is in the wrong direction, do not recalculate, investigate! Find out what is happening, and make the apporpriate decisions once you understand what is happening in the process.

Geoff Withnell
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
#7
Actually, if he is precision machining, he is using the wrong chart, and with the correct charting methodology he will not need to recalculate at all. That is why it is important to know the process. :cool:
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#8
whoa...let's go back to the OP's original post. this isn't (yet) a question of the right chart type or not:

"but since my centerline is always readjusting with new data points" and What I want to know is if in order to see this shift, do you get your process average where you want it and then set the program to use only historical control limits and average."

This is a fundamental yet too common mistake perpetuated by SPC software - the automatic recalculation of limits.

The intent of SPC is to determine the stable period, calculate statistical control limits that quantify that stability. Then plot future points against the baseline limits. The trending, shifting and drifting and pattern rules are intended to be applied to current data aginast the baseline limits. In this way you will detect changes to your process. Do not recalculate the limits until a permanent and intentional improvemetn has been made.

most "online" SPC software allows you to turn off the auto recalculate function. if it does not you shoudln't purchase or use the software. A mistake this fundamental typically means that other fundamental and not so obvious mistakes are resident in the software. Offline statistical software, such as Minitab or JMP are not meant for ongoing monitoring and should not be used as such...even tho autorecalculate can be turned off any one entering the data can turn it back on and you'd never know.

Now once you have your limits locked down, we can deal with the essential questions concerning the correct chart selection...tool wear isn't the only thing that causes drifts and many of these are unwanted and controllable. As such one wouldn't use the uniform distribution, a standard Xbar R cahrt may in fact be correct.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
#9
...tool wear isn't the only thing that causes drifts and many of these are unwanted and controllable. As such one wouldn't use the uniform distribution, a standard Xbar R cahrt may in fact be correct.
Yep - that is why knowing the process would be handy.
 
T

tww11

#10
Wow, you guys are great. You hit the nail on the head for me. I wanted to know if setting historical limits was correct and when to recalculate my data. I may have not been clear enough with my initial question, but I got great answers.

Actually, I used tool wear just for an example. I don't have any process that I'm monitoring at the moment. I have just been wrestling with how I would actually peform SPC in the future by pretty much teaching myself. I have reached a point where I understand most of the underlying statistics, but I'm still trying to sort out the real world applications. I haven't dove into any of the other charts other than XbarR.

I knew that you take a bunch of points and let the software calculate the control limits and average, but I wasn't quite sure what to do after that and the software gives you lots of options and its easy to be misled on how to perform it. I just recently had a black belt tell me that you're suppossed to set control limits to customer specs. They clearly had little understanding of SPC vs. Process Capability. I've never even done real world SPC and I know that's not right. So, it's makes it hard to sort out truth from fiction.

I appreciate your patience with me and I'm very thankful for your answers.

A few more questions. Guys and Gals I would really appreciate your input on this. A range chart. I know that it measures the difference between the first data point and the last data point in a subgroup. 1. If my range chart goes out of control above the upper limit, is this telling me that my distribution is now flatter(aka larger STD) since I have a wider dispersion of data in that subgroup? 2. Following that logic, does a range chart then tell me that my process capability is varying too much?

3. I always thought of this as the old accuracy vs. precision debate. My thinking is that the range chart tells me how precise around the target I am and the control chart tells me how accurate to the target I am. But, then I hear that you should look at the range chart first. Don't you want to be accurate before you're precise?
 
Last edited by a moderator:
Thread starter Similar threads Forum Replies Date
S Monitoring is 1/shift. Is this the minimum required? CQI-11 Expert needed. IATF 16949 - Automotive Quality Systems Standard 3
lanley liao Question regarding the calibration of monitoring and measure equipment. Oil and Gas Industry Standards and Regulations 0
K Monitoring the identified side-effects and contraindications ISO 14971 - Medical Device Risk Management 4
C ISO 14001:2015 6.1.3 Compliance Obligations - Legal requirements monitoring ISO 14001:2015 Specific Discussions 0
J Calibration cycle for monitoring & measuring tools used in medical device manufacturing General Measurement Device and Calibration Topics 5
C Monitoring and Control Instruments RoHS, REACH, ELV, IMDS and Restricted Substances 3
E Insulation diagram for ECG monitoring and diagnosis system. IEC 60601 - Medical Electrical Equipment Safety Standards Series 1
A How to reduce the process SPC monitoring Capability, Accuracy and Stability - Processes, Machines, etc. 3
R Calibration lab environmental monitoring General Measurement Device and Calibration Topics 4
S User evaluation for self monitoring blood glucose test systems US Food and Drug Administration (FDA) 4
S HIPAA-compliant monitoring software (advice needed) Hospitals, Clinics & other Health Care Providers 1
P ISO Class 8 particle count (annual certification vs monitoring) ISO 13485:2016 - Medical Device Quality Management Systems 4
CPhelan Best metrics for monitoring low volume (<1000) parts per vendors Supply Chain Security Management Systems 1
qualprod Monitoring of lead time - Good KPI identification? ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 14
B Bioburden monitoring for surgical instrument provided non-sterile EU Medical Device Regulations 3
A How to set up Continuous CpK monitoring of an injection mold process Reliability Analysis - Predictions, Testing and Standards 7
P IATF 16949 8.4.2.4 Supplier Monitoring IATF 16949 - Automotive Quality Systems Standard 2
B Measuring and monitoring equipment - Understanding which procedures to be compliant with ISO 13485 ISO 13485:2016 - Medical Device Quality Management Systems 6
C ISO 9001:2015 Monitoring and measuring resources. Application a service industry ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 7
O Monitoring performance - How do I determine performance measurement basis within my organization? Misc. Quality Assurance and Business Systems Related Topics 4
D Monitoring and measuring resources - Example of how section ISO 9001 7.1.5 would apply to a manufacture of software ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 4
J Textiles needed for home monitoring device - any suggestions? Other Medical Device Related Standards 1
G Problem Resolution Report Monitoring - Customer complaint or PRR as general motors use Customer Complaints 12
E ISO 13485 7.6 Control of monitoring and measuring equipment - Assess the Validity ISO 13485:2016 - Medical Device Quality Management Systems 4
R ISO 13485 Software validation procedure and Quality Objectives Monitoring wanted Document Control Systems, Procedures, Forms and Templates 1
Q AS9120B - ESD monitoring devices part of 7.1.5 AS9100, IAQG, NADCAP and Aerospace related Standards and Requirements 7
Q AS9120 Vendor of ESD Monitoring Equipment needs to be in the Approved Supplier List? AS9100, IAQG, NADCAP and Aerospace related Standards and Requirements 8
J ISO 17025 - Please share a sample procedure for monitoring the validity of results ISO 17025 related Discussions 1
P IATF 16949 - Monitoring Suppliers for Premium Freight IATF 16949 - Automotive Quality Systems Standard 12
qualprod ISO 9001 Clause 9.1 - Monitoring measurement analysis and evaluation ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 6
T ISO 13485 8.2.5. Monitoring and measurement of quality management system processes ISO 13485:2016 - Medical Device Quality Management Systems 1
A Records required for ISO 9001 clause 7.1.5.1 - Fitness for purpose of the monitoring and measurement resources ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 10
chris1price Cleanroom Monitoring Plan - ISO14644-2:2015 - Risk Assessment Other Medical Device Related Standards 3
dhuot1990 Issues with REES Environmental Monitoring System Manufacturing and Related Processes 1
M Informational USFDA draft guidance – A Risk-Based Approach to Monitoring of Clinical Investigations Questions and Answers Guidance for Industry Medical Device and FDA Regulations and Standards News 0
I FDA requirement for 0.5 and 5 micron particle monitoring US Food and Drug Administration (FDA) 0
qualprod Complying with ISO 9001:2015 - 9.1.1 b - Monitoring, measuring, analyzing and evaluation ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 2
Marc New Posts and New Threads - Monitoring (aka "Watching") Elsmar Xenforo Forum Software Instructions and Help 1
L Monitoring transmitted data from a bluetooth medical product IEC 60601 - Medical Electrical Equipment Safety Standards Series 8
T How to plan the monitoring of suppliers ISO 13485:2016 - Medical Device Quality Management Systems 1
M CQI-9 monitoring of temperature in die Various Other Specifications, Standards, and related Requirements 2
qualprod ISO 9001:2015 Cl. 4.1 and 4.2 - Evidences for Monitoring ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 6
G IATF 9.1.1.1 Monitoring and Measurement of Manufacturing Processes IATF 16949 - Automotive Quality Systems Standard 5
qualprod ISO 9001 Cl. 4.1 and 4.2 - What does monitoring and review mean? ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 13
T Supplier monitoring - The meaning of "8.4.2.4 " Customer disruptions IATF 16949 - Automotive Quality Systems Standard 1
K ISO 13485:2016 Cl. 4.2.3 - Determine QMS Processes, Monitoring, Measuring, etc. ISO 13485:2016 - Medical Device Quality Management Systems 4
E IATF 16949 Cl. 9.1.2 - Monitoring Process Performance IATF 16949 - Automotive Quality Systems Standard 1
C Method of Monitoring our Internally Produced Non-Conforming Products Quality Tools, Improvement and Analysis 7
B IATF 16949 Section Clause 8.3.4.1 - Monitoring - Design and Development Input(s) IATF 16949 - Automotive Quality Systems Standard 3
G ISO 9001:2015 - Management Review 9.3.2 c) 5) - Monitoring and Measurement Results ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 7

Similar threads

Top Bottom