We use these values at work quite often. I'm still a green process engineers only about 1.5 years in my current role. This video has cleared up what I am actually calculating when analyzing the data collected from the manufacturing trials. I am in awe how much I have learned about my own role and how well you explained this topic. Thank you.
Great stuff Andy. Really appreciated that you highlighted that a process needs to be in control for us to utilize the capability calculations. Count me in as a subscriber.
I suspected that one of our processes was not in statistical control so I told my manager I wanted to collect data to measure Pp. I had to talk to human resources.
Andy, You are good. i am not in the profession you are discussing but your explanations are very satisfying and easy to follow. Thank You. I fall on you for some statistical explanations.
Wow! Your explanation of process capability is so clear and helpful that now I actually understand it. You have contributed amazing content and quality with this video, thank you!
Excellent presentation. Made me to understand without any difficulty, by simply following your lucid explanation. Thank You. My further request to you is need your similar video on calculating the defects per million (DPM) with any given Cpk values. I hope I made my requirement clear. Thanks in advance
Thank you Andy for posting this video! I am working on getting my ASQ Six Sigma Green Belt and this video helped make things a bit more clear. Can't wait to watch your other videos.
Thank you very much for this video, you have said it all. I need some clarifications please, 1.we manufacture several batches per year on the same machine, and we are arranged with each batch to make a new adjustment, how I can control the performance of the machine, with what parameter Cp/CpK or Pp/Ppk? Knowing that the batches are manufactured in slightly different periods 2. we have parameters that are not at the upper or lower specification limit, how can we control the capability
Hey Rezgui! Okay, so one of the baseline assumptions of control charts is that the process is consistent, and on-going. If you only run a batch several times per year, then the typical control chart methodology doesn't apply. Because each and every unique setup is essentially "special cause variation", you're intentionally changing the process, so the old control limits don't apply. There is a concept in SPC (Statistical process control) called short-run SPC, which is exactly what you're describing. You may want to do a quick youtube search on short run SPC and see if that answers your question. Check that out and see if that helps you out.
Great video Andy. Two days ago, I spent over an hour learning Cpk on my own through self study. Had I watched this video first, I would have saved a lot of time. Do you cover statistical formulas this same way in your CSSGB prep course? Thanks for the great content!
Your videos are awesome. I will be taking my CRE exam next Sunday. I chose the course Alpha Training and Consulting by John Lee. Both of your guys’ videos are awesome! Great refresher!
Wow thanks Souhaib!! Actually, I did write a book (and I created a course) on Lean Six Sigma - that course (and book) is called the Green Belt Master Class, and I'm currently working on the Black Belt Master Class to build on that initial course (and book).
@@CQEAcademy Great, Will definitely look into those. I'm currently looking into what certifications are worth pursuing, but still in the process of acquiring the knowledge. I believe your course/book would probably get me intrigued 👍.
@@souhaibsebbane5623 Man I'd love to help you learn and grow your career - let me know if you ever have any questions about quality or continuous improvement (Green Belt, etc)
@@CQEAcademy Thanks a lot for your assistance, actually I'd like to know whether or not your program can prepare me for the Council for Six Sigma certification, because I am not qualified enough to take the ASQ (according to their requirements). Thanks again.
hello, dear teacher, May I ask you a question ? In this video, you said before we calculate the CPK and PPK, we should make sure our process is under control, however, when we make a new product and set up new product lines, we do not know if our process is under control, so we would calculate the CPK and PPK, and judge our process is under control or not by the calculated CPK or PPK, is it right?
Hello Andy, I am trying to find out what is the minimum amount of the measured parts for evaluation Cpk and Ppk. I just remember from many years ago that there was specification how many parts have to be measured for evaluating short and long capability of the product. now I can not find. Thank you for your help.
Hey Lukas!!! Okay, so there aren't any hard and fast rules for how many samples you must take to calculate process capability. If you're taking data from a control chart, and converting that into Cpk, I'd recommend at least 25 sub-groups of data before you make any calculations.
Hey Juan!! That's such a great question, okay so the way that you tell if your process is centered, is to compare your Cp against your Cpk. If these two values are identical, then your process is perfectly centered. So in this case, your cp value (1.205) is very close to your cpk (1.198), which implies that your process is very close to the center.
@@CQEAcademy Thanks a lot for your answer Andy, this question is part of the six sigma Black belt certification examn , and I answered wrong .... now Im sure of the right answer thanks to you.
It's v gud you have cleared my doubt ,I having that doubt for 5 years but I have one doubt about cp and its sigma value please explain it with example as u explained in pp sigma value...please sir....
The video show detail information. It's really interested. I hope that you could share more and more video in the next time. By the way, Do you have documents relative lean six sigma topic ? Could you please share me? Thanks so much for you sharing!
Glad you liked it! Here's a link to a page with some content on lean six sigma: www.cqeacademy.com/cqe-body-of-knowledge/continuous-improvement/lean-manufacturing/
Hey Jeremy, it's very difficult, if not impossible to determine if your process is stable just using CP and PP values. Generally, the closer your two numbers are (Cp and Pp) the more stable your process is, but you should really use a control chart to make that determination.
Hey Jacob!!! Great question, okay so this factor comes from control charts. Remember Cpk calculations should originate from a control chart, and the c4 factor allows you to convert the average sample standard deviation (s-bar) from an X-bar and S chart into an estimate of the population standard deviation.
Hi Andy great lectures... but is there a mistake here on your last slides 'interpreting CPk. Should the vertical lines be upper and lower CONTROL limits? the upper and lower spec limits are not related to the process as they are set by the customer, so they are constant as they are not related to the process standard deviation. Am I right here? Also - not fully clear on when you would want to use PPK rather than CPK. Be ver grateful if you can address these points, Andy - and many thanks :)
Andy.... on second thought - and i have thought this through now so maybe apologies for my knee-jerk comment. I am really trying to understand this.. So after further thinking here is what i understood and would like your 'OK' on it. The spec limits are indeed fixed by the customer , so it ay be +/- 3 mm or whatever - that is fixed. But what is NOT fixed is the 'distance ' between the mean of the characteristic and the USL or LSL in terms of process standard deviations. The desire is to have the process so on point that the variability is very low is the standard deviation to be very low. This would mean that the USL for exmaple may be 4 or 5 or 6 standard deviations away from the mean (depending on the criticality v cost tradeoff). This the USL itself does not move but the 'fatness' of the normal distribution ie the process capability is what moves. The control limits (UCL or LCL) are a reflection of the fatness and reflects the measure of the characteristic (eg product length) at 3 standard deviations from the mean. At this point we will have captured 99% of the area under the curve . The question then is the extent of the gap between this point and the USL - if this gap is a posive number then the process is in control as we are able to deliver the product within limits. If the USL is inside the UCL then, 'mission control, we have a problem'. Is this right?
What I find a little bit confusing is that in the last section (where the sigma levels are explained) it looks like the output value remains the same but the specification limits change. (the red bars move outward in the graph) It is the opposite way around right?: the specification limits remain the same but the amount of output that fits within these limits changes. The more output fits within the limits, the higher your sigma number will be. Can you please confirm?
Hey There! You're correct most often the process is updated and the variation is reduced and the limits stay the same, and the width of the distribution shrinks.
Hey Karthik, Remember that when we're calculating CPK, we divide the distance from the nearest spec limit by 3 sigma. So if we were 1 sigma away from the nearest spec limit, our CPK would be 1/3 (0.333), hence 1 sigma is equal to a CPK of 1/3
Hello, thanks for this video. but I have doubt regrading last chart in which you have shown Cpk and sigma level,DPMO and defect rate. As per my understanding DPMO of six sigma is 3.4 but you have mentioned it 0.002. Please clear my doubt. thanks
Hey There!! Okay, so I didn't get into this, because it would have added substantial length to the video, but the estimates for DPMO are the "short term estimates" of failures. Six Sigma practitioners often apply a 1.5 sigma shift to estimate the long term capability and estimates of DPMO. I've explained the 1.5 sigma shift over at CQE Academy: www.cqeacademy.com/cqe-body-of-knowledge/continuous-improvement/continuous-improvement-techniques/ There's another video there, and I'd jump forward to 9:54 to hear the explanation -Andy
@@CQEAcademy I 100% get what you are saying and referring to, but I do think for the sake of clarity that at least you should have mentioned this digress from 3,4 in this vid (which is great btw ;o) ), as this really trips many, I would assume..
Hi Andy! I am following a six sigma foundations course on LinkedIn and noticed that the values of %Conforming and DPM's in the table at 14:22 don't match those as taught in six sigma. The course I'm following for example gives a %Conforming of 99.999 66% and DPMO of 3.4 at a sigma level of 6. How come that they are different? Are they calculated differently?
Hey Bram!!! Great question, okay so there is "short-term" capability, which is what I'm showing here in this slide. Then there is "long-term" capability, where the estimates of the defect rates are "shifted" by 1.5 sigma. That's where the difference comes from. That 1.5 sigma shift is "empirical", meaning that it is derived from experience
Hey Ahmed! That conformity % comes from the normal probability tables. The CPK calculation is very similar to the z-score transformation which will tell you the percentage of non-conforming material. Check out this video to learn that process: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-6hH7LwCw8qY.html
Hi Andy, i have seen a scenario where If PP and PPK fail to achieve the minimum requirement e.g. 1.67 but CP and CPK passes. The big factor is the tolerance is so small it doesn't leave much room for any process variation as PP and PPK is long term. What are your thoughts on this due to tight tolerances?
That's a great question, to be honest when you have tight tolerances it can be very difficult to have a capable process. Which is why when you're designing a new component, you have to pick the tolerance of your part with the manufacturing process in mind so that capability is achievable.
Hello Andy, please let me ask a very elementry question. If we want to increase Cpk value we should increase the numerator of that fraction. Since the specification limits are directly influenced by customer how to increase it? Thanks.
Hey Andy - so as you mentioned at the end, different Cpks may be acceptable depending on the risk. My question is how do you establish what that acceptable limit is for different products / industries? I've been looking into various standards (medical device) with little luck as most text is left vague and open. Lets say a 1.00 Cpk is acceptable for a certain product when for another product its not and would need 1.33 Cpk? I know its based on risk but struggling to find anything to guide internal acceptance criteria. Thanks!
Hey Dylan!!! I understand the struggle - I've also search for more definitive guidance in the past, and haven't found it. You're right in that the general guidance is risk based, and again that's tough to create industry standard guidance once, because we all have different scales for the impact (severity) of our failure modes. Some peoples worst failure mode would result in multiple deaths, while others might only impact the performance of the product without introducing any safety risk. Think about your own organizations scales for severity and probability, and then think about the severity associated with the quality attribute that you're measuring. Then ask yourself, what probability of occurrence should we achieve, for that severity - and that should give you a decent target for Cpk.
Hey Gayathri, I don't currently have a free giveaway from Cpk/Ppk, but honestly a great way to practice using these equations is to create your own excel spreadsheet
@CQEAcademy Hello. I think I made a mistake. Spec limits are used in here to calculate the Cpk and some maths, control limits are used in control charts.
Great question - are you referring to a specific comment in the video? In general, this topic is related to process capability, not SPC, so in this situation we would be focused primarily on the specification limit (Voice of the Customer), not the control limits (Voice of the Process). -Andy