Lean versus Six Sigma - fight to the death!...?


A couple of days ago, I got in a great discussion with my colleague and fellow lean thinker Gene. Our discussion was about lean versus six sigma and its unique applications. In industry, companies seem to focus on one or the other – recently, the balance has been on the lean side because of the success of the Japanese automotive industry. The company that we work for has focused predominantly on lean and the creation of flow. Gene mentioned that he was doing a value stream map at another site and noticed that one kaizen burst was a clear candidate for a six sigma project. The problem was that with a focus mainly on lean, the site lacked the six sigma bench strength. There aren’t really tools in the lean toolkit to solve highly advanced multivariable problems. Lean focuses on eliminating waste and creating flow but does it really have the ability to reduce inherent process variation? Standard work to an extent can reduce variation, but doesn’t address those complex situations where you have many interacting and confounding inputs affecting a key output. Many argue that lean has the element of practical problem solving but really there’s only so far you can get with a fishbone and a 5-why.

Opportunities for six sigma projects can be identified during value stream maps when rework loops or yield problems are prevalent. The power of six sigma is in reducing process variation through creating controls of primary input variables. This is accomplished through the use of the DMAIC process and a rigid set of statistical tools. Hypothesis testing to determine if differences/changes are actually significant is so critical when trying to solve problems but this tool isn’t taught in many practical problem solving systems. Hypothesis testing is a fundamental tool in the six sigma package.

Lean and six sigma really have to go hand in hand. If the company focus is on lean, then there needs to be some six sigma bench strength it can draw on when the need arises – and vice versa.  You can’t create flow until you create stability and the six sigma statistical toolkit helps you create stability.

The Lean Journey - Where to begin?

All journeys begin somewhere and the lean journey is no different. The idea of lean is simple but the implementation and execution is a whole other story. For a company trying to deploy lean, the task can be quite daunting and its always difficult to figure out where to start. I often get asked this question and it always gets me thinking. Do I start with 5S? How about basic Lean training? Value Stream Mapping? I mean, really, it depends on the history of the company. Does the company have a history of lean? Did the first iteration fail? Does the company have knowledge of Six Sigma or some other process improvement system? Or does the company have no experience with process improvement at all?

Where you start depends on where you've been. If you have systems that were successful in the past, but then got dropped for some reason, bring them back. Employees who remember the system will champion it and help spread it through the rest of the business. By incorporating and adapting systems that teams are familiar with, it avoids the whole idea that lean is just another fad. At my company we had a really effective problem solving system in our old Operational Excellence days. We revitalized this system in our new Lean Enterprise and feel that it's working because so many people had success with it in the past.

Another word of advice: Think big but start small. I heard of a company who tried to do their first value stream map implementation on their most complex, biggest dollar manufacturing line. Their reasoning was that it was the line that affected sales and profit the most. Hmm, makes sense right? Well, the team failed and their lean journey ground to a halt. The site management could have selected a smaller more manageable project as their first lean implementation. Make a small model value stream, show the rest of the cells that lean systems really work and then attack the bigger projects. It's not always possible but try to pick a value stream where you know the owner will really champion the lean concepts. 

My last but perhaps most important bit of advice on launching lean implementation: develop objective measurement systems. You can't figure out whether you're winning or losing if you don't have an objective measure that tells you how you're doing. How can you really tell if you're lean systems are doing any good if you can't measure your performance? Make the measures as objective as possible - preferably out of an MRP system - this avoids any measurement bias. 

There's no magic answer to the question of "where do I start my lean journey?" but hopefully I've given you some good pointers. I'd love to hear your stories or thoughts on this matter. Where did you start and how did it work out? 

Flexibility - in the factory and on the streets

When I was driving down the highway in Vancouver with my girlfriend, I found a great example of flexibility. I got giddy with excitement and asked my girlfriend to take a picture for me. She laughed at me but then obliged. What I saw were lights that were part of their traffic system. It was a three lane road with no dividers. The lanes which you could drive in were illuminated with a green arrow. Being from LA and having to deal with the traffic on a daily basis, I thought this was a simple but brilliant system. Directional traffic volume changes on a continuous basis. In the morning, everyone might be headed north to get into the city but then in the afternoon, everyone heads south to get home. On a regular freeway, you have fixed resources - when demand changes, you can't adjust - leading to traffic. The ability to turn lanes to opposite directions allows the city to flex their capacity based on demand.

The directional traffic indicators on a Vancouver bridge

Directional traffic indicators - Close up
Does demand change in a manufacturing environment? Of course it does! Mix and volume fluctuate on a monthly, daily, and sometimes ever hourly basis in most factories. Creating flexible manufacturing cells that can react to those fluctuations is critical. Dedicating equipment and people is a great idea as long as you have demand to support that dedication.

Here's an example from my factory. We have a mixed model value stream that builds three parts at high but erratic volumes. For the purposes of this example, we'll call them Items A, B, and C. A couple of years ago, someone thought it would be a good idea to create dedicated cells to all three items. Items A, B, and C all had dedicated operators and equipment. That was fine and dandy until the demand started varying. Let's say initially the demand was 2000 A, 2000 B and 1000 C. What do you think happened when the demand went to 500 A, 4000 B and 500 C? Exact same quantity but totally different mix. Well, the B operators and equipment had already been at their maximum and overtime could only get them so far. They had the free manpower and facilities from cells A and C but they weren't trained or equipped to build Bs.

Well "what did they do?" you might be wondering. The team created 3 ABC cells that could make all three of their runner parts. They duplicated all the tools required to make all three variants and also cross-trained all the personnel. Now when demand fluctuates, they are flexible enough to react to the variation... just like those Vancouver freeways.
 

Mastering your work - a Japanese Chef

I was recently on vacation in the beautiful foodie heaven of Vancouver, BC. After a long day of sight seeing on Victoria Island, my girlfriend and I decided to have some Japanese food in the downtown area at a restaurant named Guu. I was definitely not prepared for the epic cooking mastery that I was about to witness. We got a perfect seat at the bar area facing the chefs. After watching in awe for about an hour and dining on some scrumptious food, I decided to film what looked appeared to be their head chef. Have you ever been to a factory and watched an operator just absolutely master an operation? It was like that but considering the mixed model and variable demand that a chef has to deal with, it was extra special. Enjoy the show.

The Missing Pareto line - Who cares?

Pareto Analysis is a critical part of the practical problem solving process. Paretos help you prioritize your problems so you can systematically address them. Paretos have two components:
  • Bars sorted in descending order by whatever categories you're measuring
  • Cumulative percent contribution line
A proper pareto
I made up a pareto of defects above. This graph counts defects for Acme company sorted by defect type. Most people easily grasp the whole sorted bar concept, but don't quite understand the cumulative line. I've seen many paretos where people just leave it off. They're probably thinking - "well, I know what my top bar is - I'll go work on that." Yeah, that's true - they know what their top bar is - but do they understand the affect improvements are going to have on their project? That's really what the percent contribution line does for you - it helps you estimate how much of that gap you are really going to close by improving the selected bar(s).

For example, lets say my boss at Acme Company says "You better reduce those defects by at least 75%". I look at my defect pareto and it clearly shows me that if I just address the top bar (scratches), I'm only addressing 43% of my problem. The percent contribution line immediately shows me that in order to close the entire 75% gap I have to address the top four pareto bars (this assumes I completely get rid of that defect type). Without the percent contribution line, I might have just gone off and fixed scratches then been left scratching my head trying to figure out why my boss was yelling at me!

My point here is don't forget that extra line on the pareto and don't let people tell you its not important.

SpikeTV's "Coal" and Lost Time Analysis

I've been watching this great new show on Spike called "Coal". It documents the trials and tribulations of a group of miners and supervisors at a coal mine in West Virginia. It's exciting TV because you realize just what these men are risking to support their families. It seems to be an incredibly dangerous profession, but I definitely understand why they're willing to sign up. In a state where the median household income is about $36,600 (2009), a coal miner starting salary is on average $60,000.

A recent episode really stood out in mind. The team was having average production on dayshift but was failing to meet their targets on the night shift. The mine operators were scratching their heads and wondering why they weren't hitting their targets on the off-shift. Whenever the drilling machine (continuous miner I think is what they called it), wasn't running, they weren't adding any value. A variety of things were causing the drill not to run - e.g generator failure, pump failure, conveyor belts jamming, ventilation fans shutting off, etc.

The mine's solution to their problem was to bring in a "hired gun" - a contractor to take over as the shift supervisor and solve the problems. After a couple of shifts, he quickly identified the generator failure as their biggest source of downtime. When the drill was run too fast, it would bind up and kick the generator, effectively shutting down the mine for 15 minutes. This would happen several times during the night shift due to an inexperienced drill operator, costing the mine thousands. This very observant "hired gun" was able to identify this fairly obvious root cause but it seems the mine would have benefited from a lean tool called Lost Time Analysis.

In Lost Time Analysis (LTA), you track production time (planned and unplanned) you lose to a bunch of different causes. Typical causes are like "machine down", "tooling change", "absenteeism", or "preventative maintenance". You can assign codes for each cause like 01 is machine down, 02 is tooling change, etc. Whenever the machine - whatever machine you are tracking - isn't adding value, you log the time you lost and assign it to the appropriate cause code. After a set period of time, management or engineering paretos and reviews the data with the team. The team might see that they are spending an excessive amount of time on tooling changes so maybe its time for a changeover reduction event.

Lost Time Analysis might have looked like this for the mine (simplified obviously)
For the "Coal" mine, they could have used the aforementioned causes like generator failure and conveyor belts. Whenever the drill wasn't running, they could log the cause and note the minutes they lost. Their pareto should have showed them what that "hired gun" figured out - that their generator was too small. In an industry where minutes cost you thousands of dollars in lost production, measuring and understanding your downtime seems critical.

Have any insight on what kind of systems they use in the mining industry or have any past experiences in your industry with lost time analysis? Please share.

Bubble Tea and Visual Factory

 
The line to order drinks
So a couple days ago, my girlfriend and I decided to get some bubble tea before dinner. We heard about this place called Half and Half in Rowland Heights, CA which serves a special bubble tea with honey and pudding; sounded interesting so we decided to give it a shot. We were already prepared to wait based on some reviews on Yelp. After a 15 minute wait in line, we finally ordered - one ginger tea and one honey bubble tea. We sat down and started waiting for our drink (7:34 PM).

As we were sitting there, I noticed that the store just kept filling up. They weren't calling any numbers but they kept taking orders. The seats around the room quickly filled up and the line kept getting longer. After about ten minutes, I still hadn't heard a single number called. I was sitting there mumbling about takt time to my girlfriend who was clearly looking at me like the biggest lean nerd in the world.

I get pretty annoyed by poor service and I'm thinking to myself - how can they not output a single drink in ten minutes? I decided to stand up and see what the heck was going on. How can their flow really be that crappy? Well... here's what I see when I try to see into their work area:

Giant opaque glass barricade
You couldn't see anything through this giant wall. How were they doing on our drinks? Was there actually anyone working back there? Could I expect my order shortly? How long would it be? I think this lack of transparency (no pun intended) is a failure in any business. An informed customer is a happy customer. If there was a visual signal that told me that my order was going to take an extra 15 minutes, I would probably be less annoyed. I could maybe go somewhere, do something and then come back. Even if I was able to see that there was X orders in front of me, I could estimate my wait.

A giant wall blocking your production area also kills your visual factory. A production area that isn't visible to customers and other stakeholders often hides waste. Who knows what was behind there? Could have been a monkey and a panda mixing bubble tea. When you hide your work areas, your employees may also be less hesitant to leave the area in shambles. The thought may be "well the customer can't see it, it can be messy". Well, a messy work area just creates more workplace inefficiency. Creating a visual factory is key to seeing waste, identifying bottlenecks and creating flow.

Just in case your wondering - we finally got our drinks at 7:53 PM - 19 minutes after our order. At least it was delicious.

Standard Work - ItsLean.com

Sorry I haven't been posting much recently - work has been quite busy! I'll be back in blogging mode soon.

One thing I thought I would share is James Chen's work over at ItsLean.com. He's developed an Excel based template for standard work that's incredibly user friendly and feature-rich. It's free unless you want the template customized with your company logo.

Color Me Lean

So as we speak, I'm sitting at a store called Color Me Mine watching my girlfriend paint me a Boston Celtics themed pig. For those of you that have never been, Color Me Mine is a paint your own pottery place. Basically you pick a pre molded ceramic piece (plate, cup, bowl, pig - whatever your heart desires) and they give you paints and a brush. You design your own piece and then they glaze and bake it for you.

Shelves very close to workspace
Well, since I have nothing to do for the next couple hours, I'll do some rambling about how I would make Color Me Mine leaner. The store itself has pottery pieces along the left and right walls. The area between the shelves is where all the tables are with a walkway between the two rows of tables. The layout itself causes a few problems. Before any customers sit down, they have to roam around to find the piece they want to paint. This is quite distracting for people who are already seated and painting because now they have people behind them bumping into their chair scooping out the shelves. They frequently have people waiting for a table to paint so it's in their best interest to get people in and out as quickly as possible (short leadtime). Any disruptions, like pausing when people are hovering over your table trying to see the shelf behind you, increases wait time and ultimately hurts their sales.
Lots of Inventory

They could reduce this problem and probably add more tables by eliminating half of the shelves. Their shelves are filled with multiples of the same blank piece. They could put into storage all but sample of each variety. Each piece could have a simple number on the bottom and when the customer finds the piece they want, the employee retrieves a copy from stores. These are basically the first two S's of 5S - sort and simplify.

Paint with numbering system
One lean element that Color Me Mine has implemented is basically a water spider stocking system for their paint. Each customer gets a small tray for the paints they choose. Each slot on the tray is labeled with a Sharpie and each color has an associated number. When the customer runs low on paint, the Color Me Mine employee gets the number and refills the paint from the larger bottle. The paint bottles are all numbered in a central storage area (good visual management). This drastically reduces raw material (paint) usage. People have a tendency to dispense more than they really need and once the paint is out of the main bottle, its basically gone. Refilling the trays themselves saves them a lot of money.

Leadership Standard Work and Pulse Points


I recently watched a webinar hosted by lean thinker Joe Murli. The session focused on Visual Management and leadership standard work. Joe introduced me to a concept called “Pulse Points”. After you implement and debug a Lean system like a supermarket, FIFO lane or Heijunka box, you’ll need to implement a layered audit to make sure the system stays working. The pulse points are physical spots in your value stream where you can see if the systems you have implemented are being sustained. I’m guessing they are called pulse points because they determine whether you’re value stream is still alive and kicking. The value stream leader should have a pulse point walk that he makes on some set frequency. The order of the stops on the walk doesn’t have to match up with the flow of the product – you just take the shortest path through the shop and cover all the pulse points. At each pulse point, the auditor has a question they need to ask themselves – i.e. if they are at a supermarket, “Are the min/max being adhered to?” If the answer is no, then the Act phase of PDCA needs to kick in.

Lean Video Library

The guys over at LeanAmerica.org have compiled a great library of videos on Lean. They have videos on topics ranging from shopfloor waste to Lean Burrito making. Check them out.

http://leanamerica.org/video-gallery

Starbucks - Standard Work for the Abnormal

After 3 or 4 hours of A3 Problem Solving today, I kinda started tiring. I decided that a nice cold coffee drink from Starbucks might get me re-energized. I drove to the Starbucks around the corner from the site and rolled into the drive-thru. There was an SUV in front of me and the girl was talking on the phone instead of ordering. It sounded like she was trying to take an order from a friend over the phone but she didn't have the full order. After a couple minutes she eventually hung up and ordered her drink. Somewhere between the window and the ordering kiosk, she must have gotten a call back from her friend because she asked for four additional frozen drinks at the window. So I sat behind her waiting for the barista to fill her order. When she finally sped away with her drinks in hand 15 minutes later, I pulled up to the window. I had my Starbucks gift card ready to pay with but the teller instead handed me a coupon along with my drink. He then apologized for the wait and explained that the girl ahead of me placed extra orders at the window. On the coupon itself was written an apology for the slow service.
The coupon I got from Starbucks
I think there is a lesson to be learned here from my experience today at Starbucks. To improve customer satisfaction, Starbucks has created Standard Work for the abnormal. They recognize that no matter how 'lean' they make their business, that "stuff" happens. Coffee machines break down, customers drop in surprise orders, drinks get spilled... the list goes on. Starbucks has not only acknowledged that these things happen - they have actually trained their employees on what to do in case "stuff" happens. Instead of walking away annoyed by the delayed service, I walked away happy and pleasantly surprised.

In manufacturing, I don't think we create enough standard work for the abnormal. We know that despite our Lean systems, disruptions are going to occur in our value streams and business processes. What do we do when those disruptions happen? In most cases, we just try to fight the fire whenever the bomb goes off. Processes and systems that are prone to these types of problems should have clearly defined standard work on what to do in case problems arise.

Where have you seen well-defined standard work for the abnormal, either at work or elsewhere?

Kamishibai for Sustainment

If you have a clear vision and you're persistent enough, you'll eventually have some success in your Lean journey. Your culture will start changing, managers will start buying in, value streams will start flowing and you'll feel a sense of accomplishment. The joy will quickly subside and you'll realize "oh crap, I don't know how to sustain any of this stuff!"

As dumb as it sounds, its critical that you implement systems to sustain your systems. That's what Kamishibai and Leadership Standard Work is all about. I've seen many people make mistakes with Kamishibai (Go-See) systems - particularly on the shop floor. People will often times deploy Kamishibai site-wide to sustain production enablers like standard work, visual management, 5S and autonomous maintenance. Well, what if an area hasn't deployed autonomous maintenance yet? Or what if they haven't documented standard work? What happens is that these system become implementation systems instead of sustainment systems. The audit questions on the cards are always marked as "red" and then some leader tries to run off and fix all the problems. I think it sends a bad message to the guys/gals in the shop too. If you prematurely implement these red and green audit cards, it might be a while before those audits actually turn green. It's pretty demoralizing to see a red card in your cell every day that you can't control.

Shopfloor Kamishibai systems should only be created for enablers that have some level of implementation already. Kamishibai without previous implementation just creates an unmanageable list of actions.

I've always liked how Jamie Flinchbaugh closes his Lean posts with a question to the reader so I'll pose one here: What mistakes do you commonly see with Kamishibai deployment?

Lean books I'm currently reading

Great read for Business VSM
Just thought I'd share about the book I'm currently reading. My site did a Business Excellence event last year and we found that we needed to focus a lot on business process improvements in 2011. I decided to read up on business process improvement to get an understanding about the differences between mapping strategies. After reading some reviews on Amazon.com, I decided to go with "The Complete Lean Enterprise: Value Stream Mapping for Administrative and Office Processes" by Keyte and Locher (found here).  I really think this is a great book for anyone trying to implement process improvements in the office environment.

"Learning to See" by Shook and Rather is essentially the starters guide for shop floor VSM - I think Complete Lean Enterprise is the same for Business VSM. It provides a very simple, visual explanation of how to use the standard shop-floor VSM framework and apply it to office processes.

Barcode Kanbans = Awesomeness.

After a recent value stream mapping event our team identified the need to create a supermarket. The kanban from this supermarket would have to go across to the other side of the factory. I calculated the number of Kanbans as follows:

Daily Consumption (d) = 1000 pcs/day
Std Deviation (s) = 300 pcs
Lead Time to Replenish (t) = 3 days
Kanban Quantity (q) = 48 pieces

So using the formula [(d*t)+(1.65*s)]/q with 1.65 being the z value for 95% confidence (normality assumed), I calculated that we would need 73 kanbans cards. 73!! How do you manage 73 kanban cards??

Instead of making 73 physical kanban cards, I developed a VB.net application that utilizes a networked SQL database. Basically, the customer process scans a barcode when they pull a part and it instantly pops up at the supplier process saying they need one. The tricky part was setting up the software to do a batch kanban. The kanban quantity was 48 pieces but the signal quantity was sometimes 10, 20 or 30 peices. So the software had to keep track of pieces requested until 48 was reached. The requests will queue at the supplier until he builds the item and sends it along. At this point, he'll indicate on the software that the order has been fulfilled.

I understand that Kanbans are best when they are visual and physical, but the barcode Kanban has worked great in this scenario. It eliminates the wasted travel and lag time and provides instantaneous feedback to the supplier.

The Dominos Pizza Strategy and Lean Mindset

Domino's Ad Campaign
If you watch any TV, I'm sure you've seen Domino's latest ad campaign. They basically talk about how disgusting their pizza used to be and how they dedicated themselves to improving the product. Now that they've improved it, they of course want to make sure everyone has tried the revamped product. I thought that this was a pretty brilliant marketing strategy: we admit we use to stink, here's how we improved, and now you should try us out again.

When you are trying to implement Lean into a company that's had failed continuous improvement programs in the past, the Domino's strategy might not be a bad way to go. Admitting your flaws, showing that you've learned something and demonstrating that you've rededicated yourself to the cause is quite powerful as a training tool. Our site has been through failed iterations of Lean and Operational Excellence in the past 10 years. They failed, in hindsight, for pretty obvious reasons. When we relaunched the initiative a couple years ago, we gave everyone at the site a lean overview class. One component of this class was very similar to the Domino's strategy. We admitted our past flaws (not involving all the people, trying to take action before planning properly, etc) and showed them how we've started to change (using recent successful projects). The fact that we were admitting our mistakes so honestly opened a lot of minds and made changing the culture a possibility.

Add "waste" to create flow

On a value stream I worked on once, the flow would always stop at one polishing/grinding operation. Kaizen event after Kaizen event was done in that area to make it go faster. The teams spent days upon days trying to take seconds out of the process time while ignoring the real problem. Looking at the standard work, the area was well below takt, but the flow was still jamming up - so what was going on? Well, the parts would go to the next cell and then come right back. The real problem was quality variation. Some days the rework could be 5% and other days it might be 50%. How can you really flow value around such an unstable process? You can't.

We came up with a fairly simple solution to the problem but many thought it was crazy. Add a secondary polish step to reduce the variation. No one denied that the secondary polish step improved the quality but people thought it was crazy that we wanted to increase process time by 10%. How could a Lean guy add process time!? The process was still below planned cycle time/takt time and now we created process stability. We regulated the flow and created chaos by adding a bit of extra work - well worth the price.