"Pull" is a simple concept but it can be hard to get one's head around if you start by looking at your own complex operations.  So let's look at a simple example.  I can't put an illustration directly on this post, so I'm going to ask to you download a PPT file.   Simple Pull Illustration Diagram  So, download that, maybe print it out, and we'll go from there.  (I would have put it right onto this post but I don't think I can do it with this blogging software.)

You see you have a very simple process in which only one product is made that has two steps and four components.  A and B are assembled together in Step One.  That assembly moves along and components D and E are added.  Now it's a finished product.

Let's set up various scenarios with our illustration.  Suppose any or all of the following is true:

1. Any combination of Components A, B, C, and D are purchased in large quantities because of price discounts.
2. Steps 1 and 2 are far apart.
3. Large numbers of the A/B assembly (Step 1) are put together without regard to whether Step 2 needs them or not.
4. Scheduling of Step 2 is separate from the scheduling of Step 1.  In other words, Step 1 makes as much stuff as it can.  Then Step 2 makes as much stuff as it can.
5. The distances between where A, B, C, and D are stored and where Steps 1 and 2 are long.
6. A, B, C, and D are replenished in large quantities without regard to production levels. 

Traditional manufacturing allows any of these scenarios to take place.  In fact, traditional manufacturing and cost accounting actually support some of these tactics.  In pull systems, none of these scenarios are allowed to exist.  Instead, we work to set up the following conditions:

1. Steps 1 and 2 are moved close together in a cell.
2. Components A, B, C, and D are purchased and replenished in small amounts.
3. Step 1 makes only what Step 2 needs. 
4. Step 1 moves small batches to Step 2.
5. Step 2 is making only what the customer needs. 

You can see that this links Steps 1 and 2 very closely.  Perhaps too closely, especially if the two steps have very different cycle times or the components have very different lead times.  So, we might think of putting some inventory between the two steps.  We'll call this our "supermarket".  It will serve as a buffer between the two steps.

Let's say the company gets an order for 100 of whatever product that our simple line makes.  Step 2 takes, say, 10 sub-assemblies from the supermarket and starts to work on them.  Then, keeps taking ten at a time until the order is filled.  Step 1, seeing that there's a "hole" on the supermarket starts working to fill it.  Seeing the second "hole" of 10, then schedules to fill it and so on until the "hole" quits appearing. 

So, Step 2 "pulls" from the supermarket and Step 2 "pushes" to the supermarket.

You probably have a number of questions at this point:

• How much product should go into the supermarket?
• How did Step 2 decide to take ten sub-assemblies?  Why not 20? Or 2?
• What if an order for 300 of the product comes in?  Does it all work the same way?

There's no way of answering these questions without have a lot more information.  This is why developing pull systems is challenging.  More on all this in our next post.