Blog Archives

Don’t Over Complicate the Formula

People are enamored with kanban systems.  This can be a good thing, but all too often they don’t understand kanban systems are there to help highlight make problems visual.

The first thing almost everyone jumps to is the calculation for the minimum and maximum levels for the kanban.  I have seen some formulas that would make a mathematician with 3 PhDs blush.  I don’t understand the need to have a complex formula.  For years now, I have used what I see as a basic quick and easy formula to calculate the min and the max.

Min = Lead Time + Safety Stock

Max = Min + (Min/2)

Lead time is the time it takes from the moment the component is ordered until it is received and ready to be used.

Safety Stock is the amount of stock to hold because of something that could occur to delay the lead time.  Base this on where you are getting the parts from, how often does something go wrong, etc…  For example you might hold a little more safety stock for something you purchase from a company 300 miles away versus a component that is made in-house.

If the process is working smoothly, you will receive the component you ordered right as you get into the safety stock.  When the minimum level is set properly, you will feel freaked out because you believe you will run out and right about that time the components will arrive.  It is a weird feeling that you will adjust to, but makes you heartbeat fast the first few times until you get used to it and trust the process.

The maximum is something a friend and I completely made up several years ago.  There is no reason it has to be this.  I continue to use it because so far it has worked well for me over the last decade.  I always round up to the nearest full day.


Min = 2 day lead time + 1 day of safety stock = 3 days

Max = 3 + (3/2) = 4.5 round up to 5 days

The only other number that is needed is the quantity of the product used per day.  This is used to translate the number of days to a quantity of the component.

1 day usage equal 500 parts

Min = 3 days x 500 parts = 1500 part

Max = 5 days x 500 parts = 2500 parts

The point of the kanban min/max levels are to get you in the ballpark.  It shouldn’t be an exact science because you will probably round to nearest full carton or order quantity anyway.  Plus, min/max levels should NEVER stay static.  They are dynamic and change.

I wold recommend on having what you might think is a little too much inventory to start.  You can always adjust your kanban min/max levels down as you understand your process.  If you start with too little of inventory, you will run out of parts and people will not have faith in the new process and give up early on before it has a chance to work.

Get rid of the waste in your kanban calculation and go and see your process to understand if your kanban min/max are appropriate.


Supermarket Pull System

A couple of years ago, I was working with a group to create a complex supermarket pull system.  The supermarket was centered around a component that was manufactured inside the assembly facility and feed several assembly lines.  A timed delivery system was going to be used to deliver components to the each assembly line every two hours.  The deliveries were based on the orders the lines placed during the previous two hours.  Only four hours worth of components would be stored at the assembly line and the rest would be stored in the centralized supermarket.

Below is a quick presentation I gave to the leadership to help them understand how the system would work.  Slides 5 -15 show the specifics for this supermarket delivery system.

This is a specific example of how to use this supermarket system.  However, the concept is the same no matter what supermarket you use.  There should be:

  • Centralized storage location for the components
  • A small amount of inventory at the point of use
  • A signal to replenish the point of use inventory
  • A signal to replenish the supermarket inventory
  • Replenishment triggers should be based on lead time to receive inventory from supplier

A well designed supermarket system can be a very powerful tool to help reduce inventory and simplify operations.

The Supermarket is the Customer

When an organization is focused on implementing lean, one of the first things that everyone talks about is focusing on the customer.  It could be the external customer or the internal customer.

Who the external customer is seems to be clear in most cases.  But when dealing with internal customers it isn’t always as straight forward.  In my experiences, the person says their internal customer is the next physical process that changes the product (or service).

What if you have implemented a pull system with a supermarket in between two physical processes?

The functioning of a supermarket is a process.  Product enters, waits, and then exits to the next physical process.  The product does not physically change, but the supermarket is a process.

Even in lean notation we over look this.  If you are familiar with Value Stream Mapping, the is a specific inventory icon for a supermarket.  It does not get notated with a process box.

Now if the mindset is the supermarket is a process, then it is the customer of the physical process upstream and the supplier of the physical process downstream.

This changes how questions are asked around creating a pull system.  Instead of asking, “How much does process 2 produce?  Or what is does process 2 need?” , the questions should be, “How much leaves the supermarket in X timeframe?  When does the supermarket tell me to produce?”

It is a slight shift in thinking, but it can cause a lot of different questions to be asked.


Continuum of Flow – It does not stop at Supermarkets

Organizations are very happy when they are able to implement a pull system using a kanban.  The success should be one to celebrate, because it is not easy and it is a step in the direction of moving from Traditional Batch & Queue to the Ideal State of Continuous Flow.  As the chart below shows, it is just the first step in moving towards the ideal state.  This chart is one of my favorites from “The Toyota Way” by Jeffrey Liker.

The chart shows the stages of moving from a traditional batch and queue system to continuous flow.  The immediate realization is that pull using a kanban is just one step better than traditional batch and queue.  The difficulty of implementing a pull system and using a kanban is great enough that companies feel they are successful when this is achieved.  The misunderstanding is people believe this is the end state, so this is where a lot of companies stop.  For some cases, this may be what is best for their business.  For many others, they can push further towards the ideal state of continuous flow.  Many companies need help in seeing how they can move further down the continuum.

First, we should clarify two words that I hear used interchangeably, pull and kanban.  There seems to be a misunderstanding of what each really means.  The definitions below were taken from Wikipedia.  They are short and concise and seem to describe each nicely.

Pull – a system where the consumer requests the product and “pulls” it through the delivery channel

Kanban – a signaling system to trigger action

Supermarket Pull – Just like in a supermarket, there is a variety of product to choose from.  The customer (next process) chooses what it needs and pulls the product.  Once so much of the product is pulled out a kanban signal is sent to the supplier that more product is needed.  The supplier will then produce a batch of parts to fill back into the supermarket.  The difference between batch and queue and supermarket pull is the customer is telling the supplier when to produce, not the supplier deciding on their own when to produce.

Sequenced Pull – When I worked in the auto industry we would get orders from our customers for car grills.  For certain programs, the plant would sequence the grills in the packaging in the order they would be used on the assembly line.  So as the cars came down the line, the operators were not pulling from a supermarket, but rather a bin that had the grills in order, so they didn’t have to search a supermarket for the right part.  Two other benefits were a decrease in floor space needed to store the parts and it highlighted production issues immediately.  If something was wrong with the part, there was nothing there to replace it with, so the line had to stop and a team leader notified.

FIFO Sequenced Pull – This method is like the Sequenced Pull with the addition of having standard WIP between the unlinked processes.  The standard WIP is the pull signal for the previous process to produce more work and it is placed into the flow on a First-In-First-Out basis.  When the standard WIP is full, stop producing parts.  This is similar to the Supermarket Pull, but the next process does not schedule the work.  The only way they know what to produce is from the FIFO lane.  I toured a facility a few years ago where they did this very well.  The molding presses were lined up on one side of the small building.  They produced product and filled the lanes in the middle of the building using FIFO.  The assembly cells were on the opposite side of the FIFO lanes.  They would produce whatever was next in the lane.  They did not use a schedule for the assembly cells.  They pulled just pulled the next cart of work.

Continuous Flow – This is the ideal state.  Processes are linked together so there is no inventory between them.  The work is passed from process to process with no interruptions in the flow.  Don’t get hung up on the definition of 1-pc.  I see many people get caught up that 1-pc means one individual item at a time.  1-pc could mean one bin of parts at time.  The bin is moving through the process with no interruptions and there is only one single bin of parts being worked on and no bins waiting.  I know the thought of one bin as 1-pc can be controversial.  This does not mean that 1 bin = 1-pc is the right choice all the time.  Understanding the business and the situation makes this a case-by-case call.

There may be times where you have more than one of these types of flow in your facility.  Different value streams progressing along at different rates along the continuum of flow.  This is OK.  In fact, it would be surprise if you could implement a type of flow all the way through your value stream at once.  Use this as an opportunity to create an experiment and see what works best for you.