Gemba Panta Rei

One of the most under-appreciated items in the Lean tool bag is the Table of Production Capacity by Process (TPCBP, also known as Process Capacity Table). This one-pager can define the theoretical maximum output of any process by taking into account manual time, cycle time, changeover time, etc.

The formula for capacity is simple, and intuitive once it is understood. Yet in a similar fashion to one-piece flow and takt time, there is something that seems counterintuitive or 'backwards' about it at first to most people. To review, the formula for capacity can be expressed as:

Capacity = Net Available Time / Time per Piece

where Net Available Time is time available per shift minus planned breaks and lunches, and Time per Piece is the sum of Manual Time + Auto Time + Changeover Time per piece.

The numerator (top of the line) units are seconds, and the denominator units are seconds per piece, so when seconds cancel out you end up with pieces. In other words

Capacity = Net Available Time / (Man. Time + Auto. Time) + (Changeover Time / Pieces to changeover)

This is a highly logical, fact-based way of determining the capacity of a process. For whatever reason, this is not the way most traditional companies do it.

Take the recent example of a transportation equipment manufacturer located int he mid-west U.S. The truck business has been booming in 2004, and demand has outstripped capacity by 20% or more. The company needs more throughput in order to satisfy customer demand.

Their Lean Manager, who has some TOC background, was poring over data looking for the bottlenecks so that we could target kaizen activity in the right areas. This proved tricky as on different days there were problems with material shortages, poor quality of incoming materials, absenteeism, etc. All ripe areas for kaizen. But where to start?

One of the larger constraints appeared to be at the presses. This department was under producing by 1,000 or more as compared to the best days in the assembly department (when all people and materials were available). Could we get more production out of the presses? Of course! But how much? Would it be enough or was it necessary to buy new equipment?

When asked "How many more could you produce on a perfect day?" they didn't know. As far as the press shop was concerned, they were doing pretty good already. The closest this company could get to theoretical capacity was "On a good day, we can make up to 8,500 at our bottleneck process."

Of course there were the die changes, but those were "under control". In a word, no one had done the math all the way through, using the above formula. Too busy managing the day do day. When we went to the floor, collected the data, and the math was done, the theoretical capacity of the process was nearly double what they had thought. This opened people's eyes and pointed to many areas for kaizen.

Go to Gemba, get the facts.

Posted by Jon Miller at 10:10 PM | Permalink | Comments (0)

Recent examples from clients struggling with non-Lean scheduling methods reminded me of the importance of a fundamental principle of Lean - "do today's work today". This means no late deliveries, capacity equal to demand, and no overproduction.

It's really a basic philosophy of business but it's harder than it seems. Especially when you are depending on an information system that doesn't contain the information you need.

What you need to be able to do today's work today:

1) Know what today's work is (understand daily customer demand)
2) Know how much time that work takes (understand process cycle times for all work elements)
3) Know your capacity at each process / workstation

How many companies can say they meet all three conditions? In our experience of working with or visiting over 100 companies, very few can. What then, are some remedies?

First you need a way to break down scheduling of work down to daily buckets at a minimum. Buckets of 2 to 4 hours area preferable. Within this daily schedule, there should be no changes. After all, it's today's work. Do it today.

Even if you know what today's work is, some companies choose to today's work change, today. When today's schedule can change today (schedule changes, expedites, drop-ins) it introduces complexity and waste to the system. In short, if you have less than a firm 1-day schedule, you will damage your ability to do today's work today.

The best way to avoid needing to make daily schedule changes is to increase your velocity by reducing lead-time, allowing you to respond quicker than your customer expects. If you are batching, going to one-piece flow will certainly help. Of course it helps if your capacity to supply exceeds demand, and in our experience you can recoup between 25% and 35% capacity by firming up your schedule and not introducing daily changes.

Second, go to the Gemba and measure everything you do. How many seconds does it take to do each task? Don't depend on standard times, times in the system, or estimates. Doing so is taking a costly shortcut. As you observe and time processes, look for the true value added time (transforming material or information) and make note of all the wasted time or effort. There is your opportunity for future kaizen and increased capacity.

Third, make good use of the Table of Production Capacity by Process and the formula for calculating capacity. Take into account not only actual value added time but loading time, walking time, change over time (per piece), etc. and understand what each process should take.

There will inevitably be losses of capacity, and this is where SMED, TPM, and other Lean tools come in handy. Whenever the theoretical capacity is not being met, attack the losses with kaizen.

Most MRP systems do not readily support this type of model. Several of our clients have had success customizing Microsoft Access reports and processes to give them the capacity vs. demand information they need. Understand first what you need to do today's work today, then make technology an ally of your Lean efforts and kaizen activities.

Posted by Jon Miller at 9:40 PM | Permalink | Comments (0)

During a recent Lean Office seminar, audience members were interested in the question of how to sustain office kaizen results. As in any type of kaizen, a lot of things have to go right before results achieved during one week of intense improvement activity will sustain over the long haul. These things include management support, workforce involvement & communication, rapid completion & implementation of ideas, etc.

The one thing that matters more in Lean Office implementations (more so than factory kaizen) is making the process visual. In simple terms this means... the walls must go! The cubicles must go! Bring on the open office! Improve visibility, communication, space use, 5S, enable work to flow and pull.

Why is it that we are willing to remove visual barriers in the factory in the name of Lean, but there is so much resistance to removing walls, partitions, and cubicles in the office? It has a lot to do with culture, and how people perceive 'office work' to be different from other work such as in the factory.

Letting the attitude that work in the office is somehow different from other types of value added work exist is the biggest threat to sustaining gains in through office kaizen. It is no different than removing the personal tool boxes or removing the inventory in front of the machine. Both represent comfort and safety, and require education, training and a dose of "What's in it for me?" to overcome.

We at Gemba do kaizen on our own operations. We practice what we preach. We have removed the walls and work in an open office. The more we work in a Lean Office (open room) the more we wonder how we ever managed to work inside walls.

Posted by Jon Miller at 11:58 AM | Permalink | Comments (0)

There's an often-cited Harvard Business Review statistic that goes something like "Developing a new client relationship costs between six to eight times more than maintaining an existing relationship". Spending six times more on customer retention does not sound Lean.

For review, Lean can be boiled down to three rules:

#1: The customer defines value
#2: Improvement is focused on removing waste (what customers do not value)
#3: Performing value-added work in a rapid flow maximizes profit

While there is a large body of well-developed systems and methods for achieving #3 (Just In Time, cellular manufacturing, etc.) and the entire focus of kaizen is on improving performance by doing #2, there is surprisingly little in the Lean body of knowledge on #1, defining what customers value.

Take for example the 'Lean Pathway' that the Lean Enterprise Institute uses to describe the steps in a Lean Transformation. It states "Specify Value" as step one, yet contains no tools, strategies, or methods on how to do this. At the beginning of Value Stream Mapping it is necessary to define what the customer values. While it is fairly simple to do this for a manufacturing process, it is rarely done well at the enterprise level, particularly for service companies.

To remove any doubt that customer service is an important part of Lean, let's look at how poor customer service creates waste. Doing a poor job with a client relationship requires putting in effort to repair damage (correction, defects). Having to do more sales & marketing activity to keep the pipeline full also stretches valuable resources (overproduction). Doing things we think will keeps the customer happy but may not is also a waste (processing waste). And finally, rapid growth leads to managing more customers or more projects (inventory), at a lower quality level resulting in customer turnover.

From a product design standpoint Quality Function Deployment does a good job with Voice of Customer and other tools to narrow down the design features customers want. Six Sigma tools can be applied to statistically identify the customers' spoken and unspoken needs. Both of these tend to focus on analysis of data rather than listening to how happy or unhappy the customers are about the job you are doing right now.

The rule of thumb that letting a defect pass undetected downstream and having to correct it later costs 10 times as much as correcting the problem at the source. Perhaps the "six to eight" times quoted in the Harvard Business Review is conservative. We need a "customer service at the source" as well as "quality at the source".

This begins with face to face interviews with clients, followed by defined customer satisfaction criteria, and frequent satisfaction checks throughout the process. Encourage customers to tell you sooner when they are not satisfied. Encourage people to make it right with the customer sooner. This will save time and money, and keep customers happy.

Posted by Jon Miller at 1:41 PM | Permalink | Comments (6)

And the third question on Lean applied to engineering...

3) When a process is very detailed, what is the best way to map the process so that it does not get too complicated with too much detail?

There is detail and there is complexity. Detail means that there are many small steps that need to be documented. In fact, all things are made up of small steps when you start to measure things in seconds and tenths of seconds. Measure only what you want to improve. First define if you are doing "process kaizen" or "flow kaizen".

If you are doing process kaizen in order to save seconds on incoming phone calls or shaving keystrokes, measure in detail. If you are looking for a opportunities in information flow across processes (kaizen to create flow and pull) then map the macro level first.

Another point to remember with office kaizen is that when mapping information flow it may be easier to start by asking "what is value" and map only that, leaving out much of the detail that is either waste or non-value added. This is mapping in reverse, where you start with the ideal and work your way back to current state.

Be careful not to mistake the tool for the goal. Value Stream Mapping and its hybrids are not the measuring or analysis tool for every situation. It is a waste for a kaizen team spends days Value Stream Mapping, creating a colorful wall full of information, when the problem could be solved in the time it takes to observe, test, and evaluate.

If the problem needs to be defined in more detail cause and effect diagrams and other mapping tools may be more helpful. Define the problem, measure it, and analyze the processes using whatever tool makes sense.

Posted by Jon Miller at 9:55 AM | Permalink | Comments (0) | TrackBacks (0)

Second in the series of questions on implementing Lean in engineering...

2) How do we run to a variable Takt time, and are there other ways to pace or level load work flow?

If Takt time varies due to variation in demand, there are several things you can do to maintain an even pace and even work flow.

First, running to a variable Takt time is not something unique to engineering processes. Any time demand fluctuates against a relatively fixed net available time to get the work done, Takt will vary (Takt = Net Available Time / Demand). Balancing the workload and to Takt time, and having several different work balance scenarios depending on the Takt time (longer or shorter) and the cycle time (longer or shorter) is all a basic part of creating flow in Lean. There's no need to go in to that in depth here.

Second, you can set an artificial Takt time that is faster than actual demand. Get the work done quicker and move onto something else. If something takes less time or has a long Takt time, complete the work and move on. In order to keep this person productive, cross training upstream and downstream is essential.

Third, follow the rule "when you can't flow, pull". Flowing work one-piece at a time based on downstream pull paced by Takt is the ideal for TPS. In the early stages, flowing to Takt will be difficult. Flowing at all is an accomplishment for most engineering organizations new to Lean. Connect the processes so that the customer (downstream process) can pull the work when they have time available. This will begin to set the workflow in motion while you figure out the Takt.

Posted by Jon Miller at 10:00 AM | Permalink | Comments (0) | TrackBacks (0)

We received several good questions from a manager of System and Process Improvement attempting to do kaizen in engineering. She saw the tremendous wastes (Lean opportunities) in the engineering department, but was having limited success getting the engineers to adopt Lean thinking. Some of here challenges include:

Question #1: How do we measure what is perceived "un-measurable", such as the amount of time to design or research where there are many variables?

First, there is very little that is truly "un-measurable". Design and research activities can certainly be measured by tools such as a stop watch, video, spaghetti diagram, etc.

The real question here is how to measure activities like these that are difficult to measure. These engineering processes may be iterative (happening over and over again), discontinuous, and invisible, as it is often being done inside the mind.

For iterative processes, consider each iteration to be 'rework' or the result of not having the right information, specifications, test results, etc. on the few iterations. Ask how the number of iterations can be reduced to get to a final design quicker. Measure each iteration and identify where value is being added and where it is simply being 'reworked'. Identify and kaizen the root causes of non value-added iterations.

For dis-continuous processes, capture each segment of the process when it is happening. Look for kaizen opportunities at hand-off points. Look for ways to connect the processes and implement one-piece flow of design. Identify and kaizen the reasons that process can not be flowed one at a time continuously. When you flow the work, iterations will happen quicker with better quality.

For work being done inside the mind - this is a challenge - make the work visual. Determine a Takt time, and periodically stop to make the output visual for the downstream process (the customer). This could take the form of a rough sketch, a spec sheet, a research brief, etc. By breaking the 'work in the mind' into smaller, more visible pieces it allows downstream pull by the customer - resulting in better flow.

As with all improvement efforts, it helps to define the goal first before measuring and analyzing. What is the output of the design or research process? What is the desired end product, and how much time should it take to achieve this? What are the steps involved in creating the design or research output? Once the 'ideal state' is defined, it is easier to see the waste and begin kaizen.

Posted by Jon Miller at 10:02 AM | Permalink | Comments (0) | TrackBacks (0)

During a kaizen workshop the kaizen team identified the lack of value-added content in a final inspection process. This lead to an interesting comparison of quality systems that do not practice Lean manufacturing principles, with the American criminal justice system.

In Lean thinking, final inspection is bad. You find the problems after they have happened, and if you are batching chances are you have more than one defect that needs to be scrapped, reworked, or otherwise dispositioned. As a general rule of thumb, the cost of correcting a defect increases by 10 times when it is detected each step later in the process.

Lean thinking says you check and guarantee 100% quality in-process. You feedback information about poor quality as soon as possible. You do root cause analysis and put in countermeasures, such as pokayoke devices (mistake-proofing). The focus is on defect prevention rather than catching and locking up the defects.

Our criminal justice system takes the non-Lean approach. With a few notable exceptions (such as Victim Offender Reconciliation Programs and Restorative Justice Programs) law enforcement seems to be about catching and locking up the criminals (people with defective, or unlawful behavior) rather than early detection, root cause analysis, and prevention.

Lean practitioners know how much more expensive final inspection and correction is than early detection and mistake proofing. Perhaps the cost of our criminal justice and incarceration systems would be lower if the focus was on prevention rather than detection.

Posted by Jon Miller at 9:23 PM | Permalink | Comments (0)

During a dinner meeting I had the chance to exchange views on the progress of the Lean effort at a client company with the President. They are early in the process, having trained all employees and having done two kaizens and are on their third. I felt things were progressing well both in terms of results and workforce involvement.

The President of the company was concerned that some of the team members were much less engaged during the kaizen than the others. While none of them were disruptive or resistant, they just weren't as enthusiastic or involved. Should we continue to include people who are not as eager, and limit it only to the aggressive participants?

My recommendation was, and has always been that it's important to involve those who may not be as engaged during the week, for several reasons.

First, if you only involve the "early adopters" you take the risk of charging ahead with enthusiasm only, without some of the critical thinking of the naysayers and those who know "why it won't work". The people standing by and watching may have some very valid concerns.

Second, even if some team members don't take a leading role they need to be involved the kaizen process so they have a chance to give input and understand the process that was used to make decisions during the kaizen. They are given a chance, a voice, and educated in the team decision making process of kaizen.

Third, sometimes it takes people longer to understand, accept, or learn how to apply what they are seeing during a kaizen. Repeated exposure may be necessary. We have found that people who may take longer to "get it" often have a deeper understanding. It's a matter of a different learning style.

The key thing is to keep team members from being bored during the kaizen week. This is less a matter of whether a team members is aggressive or passive and more a matter of having a team size that is right for the scope of the kaizen, and having effective facilitators or kaizen team leaders that can keep team members busy making improvements.

At least in the early stages, include team members who may seem disinterested and use the opportunity to build buy in for the process and the changes.

Posted by Jon Miller at 8:32 PM | Permalink | Comments (3)

The team leader of a kaizen project, we'll call him Tim, was very disappointed in the weeks immediately after a kaizen. Tim was the supervisor of the area, and when he checked in on the machine operators he found that there was inventory building up again and some of the changes made during the kaizen had been undone.

The team had made great progress during the kaizen week establishing standard procedures, cutting down on WIP, and identifying and eliminating wasteful steps, all in all cutting out 25% of the hands-on time. Most of these gains were achieved most of the time, but there was some slippage.

It's not rare that a kaizen team achieves a lot during the kaizen week, and fails to sustain all of the gains in the following weeks. There are many reasons for this:

1) Lack of management attention
2) Lack of buy-in or involvement from area workers
3) Lack of follow through on action items

In Tim's case it was none of the above, but a case of having simply made more good things happen during the kaizen week than they could keep an eye on and sustain. The proverbial 'water level' had been lowered to the point where the rocks were visible.

Another analogy is that it was similar to running up a mountain without driving in stakes along the way to make sure when you slipped on a loose patch of gravel you didn't slide half-way down the mountain. You need to make sure that even if you start slipping back, you lose gains only to a certain point and don't end up all the way back down the mountain.

Everyone agreed with the math and with the demonstration improved process, but not everyone had a chance to work in the new process before the kaizen week was done. Some of the parts that were not included in the top 80% analyzed during the kaizen week also caused problems.

In this case some of the 'stakes' that needed to be driven in included additional timing of the 20% items, additional time for workers to train and become comfortable in the new method, and visual controls to help the new methods become habit.

In addition to providing the time and attention needed to complete kaizen action items it is often necessary to put in visuals that help indicate how things should be run and when things begin to go awry.

What's most important is that you celebrate the gains, remain positive and be prepared to take a step back and take another run at the mountain.

Posted by Jon Miller at 5:17 PM | Permalink | Comments (2)

It's encouraging to see that as the Lean buzz expands from manufacturing to healthcare and other industries, some organizations and practitioners are beginning to recognize that there is more to Lean than kaizen events, "Lean tools" such as SMED, TPM, and kanban, or cookie-cutter production systems of the Demand Flow type.

What I'm talking about is what one of our clients coined as "Everyone, Everyday" kaizen. It's what Toyota would call the Creative Idea Suggestion System. The suggestion system died an early death in the U.S.A. but is making a comeback as part of the fabric of Lean.

There are at least two recent books on this topic, The Idea Generator by Bunji Tozawa and Norman Bodek, and Ideas Are Free by Alan G. Robinson. Both attempt to simplify and systematize the generation of small, local improvements by workers at all levels, or "everyone, everyday".

Certainly this is perhaps the most powerful Lean tool over the long term. Witness Toyota and their 700,000 (seven with five zeros) implemented improvement ideas reported in 2002, nearly one per month per employee (in Japan). Similar or better numbers are reported for the North American employees.

As part of the Lean Transformation, and as part of continuing education, awareness, and training there is nothing like requiring employees to think of and implement one improvement idea per person per month.

Posted by Jon Miller at 7:30 PM | Permalink | Comments (0)

We've found that there's a wall that's people have to get past when learning to think Lean. Teaching the 7 Wastes and 5S as eliminating searching, motion, errors, (7W) etc. by reorganizing the work area and making it more visual (5S) strikes most people as 'common sense', and they get it.

People usually get that flow is a better way to work than start-stop or batch-and-queue. However, the trouble is that pulling one piece every takt time simply does not make sense to most people. It's counterintuitive. How can doing one be more efficient than doing 10?

The basic tools of point of use and cellular layout can get you a long way, but the true breakthroughs in thinking and seeing opportunity only come when people start to pull one piece of work at takt time. Making sense of Takt-Flow-Pull starts with seeing and experiencing.

Rather than explaining using theory or examples, the best way is to demonstrate through a flow exercise (legos, paper airplanes, etc.). We have found that it's best to tell people up front that Takt-Flow-Pull will seem wrong and will seem counterintuitive, but to have faith and to give it a try.

Posted by Jon Miller at 7:12 PM | Permalink | Comments (2)