วีดีโอ

You take either 50-100 individual samples or 30-60 sample subgroups to determine your mean and control limits. These should be collected over a medium time interval - that means don't use consecutive samples, but rather sample across some time, batches and/or operators. My rule of thumb is to sample at the same frequency as your intended control plan or up to twice as frequent. During this time, try to minimise 'corrections' to the settings or process. If you have real process shifts happening during this sampling period, you'll see that the actual mean may not be very relevant, but the SD and control limits will still be useful. In any case, you'll first set temporary control limits while you're removing special causes of variation and improving process stability. Then you repeat the sampling process with a new set of 50-100 / 30-60 samples and calculate your definitive mean and control limits from them.

The concept of One Piece Flow is still very valid - it's a good system, just not as overpoweringly good as many of those animations make it out to be. I would say the main benefit is early detection of problems - if you pull one piece through your entire production process, you'll know if you have the right materials, design and work stations set up, before you commit a whole batch of materials to the first couple of operations. The other benefit is flexibility in production - Work In Progress isn't worth anything (you've committed the materials, but haven't made the thing your customer will pay for), so it's best to finish a product once you start on its materials. With OPF, you'll get to each finished product as soon as possible. You may find, however, that some operations are simply easier to do in small batches. Usually because the setup of a work station takes some time or if an operation can treat a batch of products in basically the same time as a single piece (example of the oven). That's why many companies don't go for literal One Piece Flow, but rather use Small Batch Production. These batches probably should not be larger than the customer's order (unless you're making to stock), but they can definitely be smaller - pick a batch size that still allows for easy handling and transport and also doesn't take up too much time. If you're working on larger stuff (like a motor cycle, or even just an engine or pump) one piece is probably your best batch size, but if your products are small enough that several fit in a box (which won't require a forklift to move 😉) than go up to roughly what you can still move manually and up to what takes a matter of minutes to produce (up to about 30 min maximum for a small batch - shorter is often better for flow).

@@TomMentink Good feedback. We make custom kitchens and builtin furniture. I am a career carpenter but I am self taught regarding cabinetry. It started with just, "can you make a cabinet" (yes), to a small shop (3,000 SF) and 3 other employees besides myself. I created and continue to develop a system and process for building, tracking labor and materials, analyzing the process, bidding and estimating, and, most recently, upgrading to industrial machinery which none of know how to use yet, including a CNC. The point is, Im always researching and exploring new and better ways to do EVERYTHING, and a big part of that is using the internet. But it sometimes surprises me what other people have to say regarding literally everthing, but specifically concerning workflow efficiency, best practices for building, and in general how things should be done. My instincts tell me to find what works best for our specific situation. Learn what I can from other skilled and knowledgable people, but chew the meat, spit out the bones. One challenge of that however, is when you have a tendency to get stuck in analysis paralysis. It kinda comes with the territory of being a bit obsessive over the details. It can make for a good craftsman, but it also can get in the way of production. Regardless, I appreciate your video and feedback. Its quite helpful to hear and see people breakdown things to a level thats easily digestable.

My pleasure, happy you found my video useful

Specifically the stuff I share in this video, probably not really from college (maybe Business Administration). But QA/QC skills can be found in several studies: for food industry, there's a range of Food Technology studies at professional education, college and university levels; similarly automotive employs Automotive Engineers from different education levels; other industries have their own specialised studies. In general, any Technical Business Administration or Engineering studies that are more or less in the field you want to work in will do - your eye for quality, ability to think in processes and general STEM fields will make you successful in the QA/QC field, your specific formal education is mostly your way into the career and there are Quality Managers coming from many different backgrounds and educations.

Thanks for sharing that praise ☺ Providing others the possibility to learn from my experiences and insights is why I started this channel. I'm a firm believer in learning from several sources. So spread the word, let anyone who might benefit from my videos know about them - it'll help them and me both 😉

@TomMentink That is very correct 👏 I will, for sure. Thank you 😊

@@TomMentink Tom, where can I get your email address, Sir?

@@BNT05 it's on the channel page (but won't show on mobile devices for some reason): youtube@tommentink.com

@@lastcactus6854 that’s great to hear - wishing you the very best in your new role! This video specifically is a very strong concept for managing the CI process/programme within our organisation, so take it to heart in this new function 😉

Thanks @@Oekiero for those nice words - it means a lot to be that my work brings value to others. If you’d like to help, spread the word - feel free to send my videos to colleagues and your broader professional network.

@@leratotakalo5997 my pleasure, glad to see you like my video

Happy to see you liked it, thanks for sharing that. If you’d like me to cover a certain topic in a future video, feel free to request 😉

Happy to hear you found value in my videos. Do remember that this is only my view on it, and it would serve you well to check out multiple viewpoints to make your own style.

That depends on what you're averaging - if you take the beginning + end of one run and use the average of those two to say something about that run; that's subgroup size 2. If you adjust the process based on each individual sample; that's subgroup size 1. Subgroup size 15 would be if you average all begin samples over a month and use them to steer the process after that month, not super likely to be used for process control, but that's how you'd get to such a sample size.

@@TomMentink Hi Tom, Thanks again for the response, i can see your point for having either a subgroup size of 1 or 2., would this also depend if there is variation in your process whether you choice subgroup size of 1 or 2?

​@@deanopenn you can use a larger sample size to get a better measurement of process centreline, so yes: if you've got more variation in your process, use a larger sample size. It can also depend mainly on process setup - if you've got two lanes producing the same product, you'll have to decide on if and how you'd want to spot any difference between those lanes. If you're mostly interested in the general process shifts, lob them together (do check for the spread between both samples in the subgroup when using this for process control). This is especially useful if you've got a structural difference between the lanes that you can't fix (for now). So one of the things for you to consider: is averaging the begin and end sample a better predictor for the whole batch than just the begin? If there's no significant difference between begin and end samples (do a paired t-test), then averaging becomes less useful (can still be used with high variance). You might get better practical benefit from just using the begin of each run, so that you can still steer the process when needed.

@@TomMentink great response as usual. its on one process, same tool etc. ii have done a Test on the 2 data sets measurement 1 and 2 (Beginning and end of the run), using 2-Sample t. the P value is 0.770 which indicates indicates the null hypothesis is true.

@@deanopenn so you don't have to use both values due to having to average a mean value over the whole run - then it's up to you to decide if the variation between samples is so high that you'd rather average out the two samples into one sample subgroup (statistically, that brings you 1,4 times closer to the actual process centreline, but it will make your calculations a bit more involved and you'll only be able to react after the run is already finished)

Absolutely - that's how to use the Fishbone effectively in the whole RCA process.

My pleasure Karin, and happy to hear you liked it. Still working on that Early Product and/or Early Equipment management overview - they will also get their video at some point, but haven't yet condensed my thinking on them enough to make for an interesting explanation ☺

When Cpk is good, but Ppk is not fully up to your demands (Ppk 1.67 is pretty high, mind you), that means your process' average is shifting over time. Maybe there are large and unpredictable differences between material batches, which influences your process outcome, but most other factors should be controllable (not per se easy to control, but that's what you should work on). In a more recent video, I also explained why I think that for many industries, Cmk >1,67 Cpk >1,33 and Ppk >1,00 is sufficient. You might want to up all of them by 0,33 for more demanding customers, but the tolerances you place on your system do indeed seem very strict. th-cam.com/video/Fi6WPCgwhpg/w-d-xo.html

Happy to hear that you liked the video, hope it will help you understand Process Capability better and allow you to effectively use it in practice.

Yes, Cmk is just the start - if it's good, then you continue with Cpk (and maybe Ppk). Cmk is not enough for PPAP, because it doesn't tell you much about the final variation the customer is getting. For this, you'd need at least Cpk, which is a good indication of variation within each of your batches, or Ppk, which describes the longer term variation also when your customer gets different batches from you (any probably assumes that they will all be the same). 50 consecutive pieces is a subgroup size of 50, but that doesn't figure into Cmk calculations - just calculate standard deviation directly over sample sizes larger than 8 (no need to use SD estimation tables based on subgroup size).

@@TomMentink Hi Tom, thanks for the detailed explanation. In response to what you have said above. i thought if you collected x50 pcs off a run (1 part after another), then this would be classed as a subgroup size of 1 and then if you done the same an hour later (collected another x50 pcs) that would then be subgroup 2 or have i misunderstood would you have said

@@deanopenn these 50 pieces would make up 1 subgroup of size 50. Subgroup size means how many pieces/parts are within 1 sample. And again - when calculating Cmk, don't worry about subgroups, since there is just one large sample (of 30+ pieces), this has nothing to do with how many pieces you'll put into each sample subgroup for later tests.

Happy to see you liked it, hope it will help you in your professional journey.

Great to hear you enjoy this and other videos I made, thanks for sharing that. And yes, it is super important to lead the way in 5S by managers organising their own offices, but it can also be a great way to discuss processes, waste, material and information flows, etc.

The most common reason for this difference is that the averages of the samples vary quite a bit over time as well. So the short term variance (as seen inside each sample group) is pretty small, but centreline of your whole process is shifting between samples. This method of averaging variances assumes that the centreline of the process is stable. When calculating the SD over the whole set of all individual measurements, this change will affect the result. In general, you'd prefer to use the direct calculation of SD for describing the process. Using subgroup SD's is used for process control (where you are interested in spotting that shift of the centreline more than saying something about the whole production run). For reference: the effect you observed is very similar to something in Process Capability: when you see a higher Cpk than Ppk, this shifting centreline is also what's happening. By the way: sometimes (quite rare) you'll see a lower SD from direct calculation (or a higher Ppk than Cpk), that's when the subgroup averages are very similar (more similar than you'd estimate based on the subgroup SD's). But that's a bit of an edge case,, so more for interest than practical use.

@@TomMentink Thank you so much!

Haha, so recognisable. So take up the challenge - clean that thing out and unscrew it from you desk - your legs will also thank you for the extra space.

Happy you like the video and the template, I hope you get great use out of it. I know, maybe I should set up a small store of useful templates, but I'd also be happy if people like and share the videos and maybe purchase one of my courses from time to time - that allows me to keep providing free content for the whole community.

Thanks for sharing that and glad to see you're enjoying many of my videos ☺

When dealing with cell configurations, probably best to focus each AM team on a single cell. Although even then, maybe one of the AM team members is very good at setting up the replenishment system, for instance, and would be a great candidate to set that up similarly for all cells...

Thank you, that's so nice to hear. Hope you'll be enjoying many more of my videos and that they may help you professionally.

Hope you enjoy it and that it will help you teach Lean to your colleagues/students

Hi Karin, how great to hear, I'm guessing your program has had some good early successes, since you're implementing these pillars that are generally launched a bit later in the program. I'll pull my thoughts together and think about how to summarise those pillars into a video - they might be a bit less popular than the main TPM pillars, but they deserve some attention and explanation too 😉

Hi David, that's great to hear - hope you have great fun and learnings from your thesis study. Do note that the OLE calculations I describe here are based of machine OEE x crew size. This is a way to take machine effectiveness a level further and make it way more valuable than the usual manhours/units produced. If you're dealing more with people capacity and the machines are just there to help them (no really bottleneck machines or machine efficiency to deal with), then it's better to look at your Labor Effectiveness in much the same way as your usual OEE - that is, just the one-dimensional scale. Take away off-time first, then planned breaks etc, and then split real stoppages from overall lower speed and constant minor issues. To do this, you will need to have a standardised unit for measuring output per hour - that's a problem for many job shops, but I think it shouldn't be a problem in most agricultural companies, as it's high volumes with very comparable tasks day to day (at least while you're handling the same produce) - just like in most manufacturing.

Hi Paul, thanks for you reply, those are very valid points indeed. For anyone interested in this great video by Paul: th-cam.com/video/vH-1deoUIZw/w-d-xo.htmlsi=f3ZiBfe7v36B2IBp) I agree that continuous reduction of process variability will lower the loss to society (and is almost always beneficial for the producing company too). The issue I have with Process Capability (variation divided by specs) is that this ratio does not inherently benefit society, because the specification limits are more or less arbitrary. For a quality characteristic that applies to the design or direct function of the part (like the size of a nut, the positioning of a label, etc.) it's best to be in the centre, yes. I would still argue that you should consider transferring this decreased process variability downstream by promising tighter specifications (especially if the customer is willing to pay a premium for it - something Ford's competitors did, by the way, and that drove market for Mazda). When we're talking about a quality characteristic that is more like unilateral tolerance (like weight, break strain, etc. that usually have a hard minimum and a maximum that's more to have product categories), I would say it's often wasteful to stay more than 4-5 SD's away from that limit. In fact, especially in filling weights/volumes, you'll often see companies choose to stay much closer to that minimum, reducing the average when they successfully reduce process variation.

Good point about the specs - The loss function ignores the tolerances, as they are as you say arbitary often, when they should include an element of the voice of the customer. Good stuff as always Tom...

@@paulallen5321 and thank you for the challenge - it’s always good to add these extra nuances and the different angle of approach/thought 👍🙏

Yes, of course you may still use the Fishbone, that's also the premise of this video - just try and get that 5 Why part outside of the Fishbone's visual. I would advice limiting how many 5 Why's you do - my general advice is to first check which assumptions in the Fishbone are true and then pick the 1 or 2 your team feels are most impactful to the problem -> start by diving into those with 5 Why before spending a lot of effort on deep diving all those minor causal factors. It's not a hard and fast rule, just advice to keep your Root Cause Analysis efforts more effective in 90+% of the cases.

Did you have to get a degree?

I usually start with a roll every 30 seconds, just one plane ordered of the colour that was rolled. You might want to hint to the teams that they could negotiate with the client about higher or lower sales volumes (which translates to rolling every 20 sec, or every 40 or 45 sec.), but they will have to deliver on that promise of course 😉 Round one will be interesting in any case - they will make a mess of it because everything is new. Do mention that while they're allowed to build up stock, any inventory in the factory will be counted and cost them money 'for keeping raw materials and WIP'. A general bonus tip (if you can handle the extra workload as the game master 😅): ask the team members how they feel after each round, how smooth were things, how's their stress level, etc. At first, you'll likely hear that some stations are super busy and can't handle the load while others are relaxed, almost bored (and probably laughing at those busy colleagues, since it's all a game anyway). If you do rounds with good kanban, and definitely if you do a full silent round, you'll see that stress levels go down a lot (the silent round with kanbans in place is the most productive and relaxed 'factory floor' anyone has ever experienced).

That is so great to hear - it brings me joy to know that my content helped you directly. Hope you’ll enjoy many more of my videos and wish you the best with your next Kaizen events!

Thanks for those nice words, happy you like the video - feel free to share it far and wide 😉

Good to hear you liked the video, hope you enjoy many more of my videos. If you ever have a question or video request, don’t hesitate to ask.

@@TomMentink , Thanks Tom and from the next view will try and contribute whatever I can for the growth of this channel.

Thanks Roby. Statistics is a strange beast: some people just get it immediately, that was not the case for me either. I found that very practical examples were the key to understanding the concepts of statistics for me, and that’s why I always try to explain it in as simple terms as possible - hope that helps others. Glad to hear that it works for you 😃

Thanks for the kind feedback, hope you enjoy many more of my videos. Since you’re not really producing a product, quality and product/service specifications will be less important (although I can imagine OTIF is really important and the market expects immediate reactions to demand changes 😅). That means you use the Planned Maintenance side of CI much more: deeply get to know your machinery, how wear and tear affects parts and make preventive maintenance plans accordingly. Many of your larger spare parts are ridiculously expensive, and also have very long leads times when ordered new, so you benefit even more from conditions based maintenance than most other industries. Of course the basics will work for you just as for any other: organise workplaces for optimal efficiency, safety and comfort with proper 5S, control standards on the 4 M’s, foster a Kaizen mindset of making and maintaining small improvements every day.

Thanks for the kind feedback, hope you enjoy many more of my videos. Since you’re not really producing a product, quality and product/service specifications will be less important (although I can imagine OTIF is really important and the market expects immediate reactions to demand changes 😅). That means you use the Planned Maintenance side of CI much more: deeply get to know your machinery, how wear and tear affects parts and make preventive maintenance plans accordingly. Many of your larger spare parts are ridiculously expensive, and also have very long leads times when ordered new, so you benefit even more from conditions based maintenance than most other industries. Of course the basics will work for you just as for any other: organise workplaces for optimal efficiency, safety and comfort with proper 5S, control standards on the 4 M’s, foster a Kaizen mindset of making and maintaining small improvements every day.

Glad you liked it, hope it can also help you explain our field to friends and colleagues 😉

I'm happy to hear you liked the video and got value from it! If it raises any other CI questions or suggestions for future videos, don't hesitate to ask - I'm happy to make videos 'on demand'

If they have different items in the 3PL than in your main warehouse, just treat them the same as any other product in your warehouses - determine the replenishment rates and stock levels and set up Kanban from there. If they are a sort of middle warehouse between your suppliers and you, then they will have a much more specific role to play: they will be your buffer between 'shop floor replenishment' and actual ordering from your supplier. Things will depend on what flexibility that 3PL warehouse has; say for example you're having daily freights and have to provide a picking list 16:00 the day prior, any Kanban card or empty box seen during a check round that comes in before 16:00 will be delivered the next day, boxes that are emptied later will have to wait 2 days before they're replaced. So make sure the remaining number of boxes will last you at least 2 days. Also, make the boxes big enough that it's easy to handle for your 3PL (if they only want to send you full pallets, than your 'box size' becomes a fully pallet...).