The Goal

active, the flow of parts to X must be greater than the flow of parts leaving X. Which means...'

He walks over to the work-in-process mountain and makes a sweeping gesture.

'You end up with all this in front of the X machine,' he says. 'And when you're pushing in more material than the system can convert into throughput, what are you getting?'

'Excess inventory,' says Stacey.

'Exactly,' says Jonah. 'But what about another combina- tion? What happens when X is feeding parts to Y?'

Jonah writes that on the floor with the chalk like this...

X -' Y

'How much of Y's 600 hours can be used productively here?' asks Jonah.

'Only 450 hours again,' says Stacey.

212

'That's right,' says Jonah. 'If Y is depending exclusively upon X to feed it inventory, the maximum number of hours it can work is determined by the output of X. And 600 hours from X equates to 450 hours for Y. After working those hours, Y will be starved for inventory to process. Which, by the way, is quite acceptable.'

'Wait a minute,' I say. 'We have bottlenecks feeding non- bottlenecks here in the plant. For instance, whatever leaves the NCX-10 will be processed by a non-bottleneck.'

'From other non-bottlenecks you mean. And do you know what happens when you keep Y active that way?' asks Jonah. 'Look at this.'

He draws a third diagram on the floor with the chalk.

In this case, Jonah explains, some parts do not flow through a bottleneck; their processing is done only by a non-bottleneck and the flow is directly from Y to assembly. The other parts do flow through a bottleneck, and they are on the X route to assem- bly where they are mated to the Y parts into a finished product.

In a real situation, the Y route probably would consist of one non-bottleneck feeding another non-bottleneck, feeding yet an- other non-bottleneck, and so on, to final assembly. The X route might have a series of non- botjtlenecks feeding a bottleneck, which in turn feeds a chain of more non-bottlenecks. In our case, Jonah says, we've got a group of non-bottleneck machines down- stream from X which can process parts from either the X or the Y route.

'But to keep it simple, I've diagrammed the combination with the fewest number of elements-one X and one Y. No mat- ter how many non-bottlenecks are in the system, the result of

213

activating Y just to keep it busy is the same. So let's say you keep both X and Y working continuously for every available hour. How efficient would the system be?'

'Super efficient,' says Bob.

'No, you're wrong,' says Jonah. 'Because what happens when all this inventory from Y reaches final assembly?'

Bob shrugs and says, 'We build the orders and ship them.'

'How can you?' asks Jonah. 'Eighty percent of your prod- ucts require at least one part from a bottleneck. What are you going to substitute for the bottleneck part that hasn't shown up yet?'

Bob scratches his head and says, 'Oh, yeah... I forgot.' 'So if we can't assemble,' says Stacey, 'we get piles of inven- tory again. Only this time the excess inventory doesn't accumu- late in front of a bottleneck; it stacks up in front of final assem- bly.'

'Yeah,' says Lou, 'and another million bucks sits still just to keep the wheels turning.'

And Jonah says, 'You see? Once more, the non-bottleneck does not determine throughput, even if it works twenty-hour hours a day.'

Bob asks, 'Okay, but what about that twenty percent of products without any bottleneck parts? We can still get high effi- ciencies with them.'

'You think so?' asks Jonah.

On the floor he diagrams it like this...

This time, he says, the X and Y operate independently of one another. They are each filling separate marketing demands.

'How much of Y's 600 hours can the system use here?' asks Jonah.

'All of 'em,' says Bob.

'Absolutely not,' says Jonah. 'Sure, at first glance it looks as if we can use one hundred percent of Y, but think again.'

'We can only use as much as the market demand can ab- sorb,' I say.

'Correct. By definition, Y has excess capacity,' says Jonah. 'So if you work Y to the maximum, you once again get excess

214

inventory. And this time you end up, not with excess work-in- process, but with excess finished goods. The constraint here is not in production. The constraint is marketing's ability to sell.'

As he says this, I'm thinking to myself about the finished goods we've got crammed into warehouses. At least two-thirds of those inventories are products made entirely with non-bottleneck parts. By running non-bottlenecks for 'efficiency,' we've built inventories far in excess of demand. And what about the remain- ing third of our finished goods? They have bottleneck parts, but most of those products have been sitting on the shelf now for a couple of years. They're obsolete. Out of 1,500 or so units in stock, we're lucky if we can sell ten a month. Just about all of the competitive products with bottleneck parts are sold virtually as soon as they come out of final assembly. A few of them sit in the warehouse a day or two before they go to the customer, but due to the backlog, not many.

I look at Jonah. To the four diagrams on the floor, he has now added numbers so that together they look like this...

Jonah says, 'We've examined four linear combinations in- volving X and Y. Now, of course, we can create endless combina- tions of X and Y. But the four in front of us are fundamental enough that we don't have to go any further. Because if we use these like building blocks, we can represent any manufacturing situation. We don't have to look at trillions of combinations of X and Y to find what is universally true in all of them; we can generalize the truth simply by identifying what happens in each of these four cases. Can you tell me what you have noticed to be similar in all of them?'

Stacey points out immediately that in no case does Y ever determine throughput for the system. Whenever it's possible to