The Goal: A Process of Ongoing Improvement, Third Revised Edition

by Eliyahu M. Goldratt

  skinny fingers through his thinning black hair.

  "This isn't the way it's supposed to be," he's saying to Stacey

  and Bob.

  "Ahh, perfect timing," says Ralph when he sees me. "Do you

  know what we just did?"

  "You found Herbie?" I say.

  Ralph says, "No, we just spent two and a half hours calculat-

  ing the demand for machines that don't exist."

  "Why'd you do that?"

  Ralph starts to sputter. Then Bob stops him.

  "Wait, wait, wait a minute. Let me explain," says Bob. "What happened was they came across some routings which still listed

  some of the old milling machines as being part of the processing.

  We don't use them—"

  "Not only don't we use them, just found out we sold them a

  year ago," says Ralph.

  "Everybody down in that department knows those machines

  aren't there anymore, so it's never been a problem," says Bob.

  So it goes. We're trying to calculate demand for every re-

  source, every piece of equipment, in the plant. Jonah had said a

  bottleneck is any resource which is equal to or less than the mar-

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  148

  ket demand placed on it. To find out if we've got one then, we

  concluded we first would have to know the total market demand

  for products coming out of this plant. And, second, we would

  have to find out how much time each resource has to contribute

  toward filling the demand. If the number of available hours for

  production (discounting maintenance time for machines, lunch

  and breaks for people, and so on) for the resource is equal to or

  less than the hours demanded, then we know we've found our

  Herbie.

  Getting a fix on the total market demand is a matter of pull-

  ing together data which we have on hand anyway—the existing

  backlog of customer orders, and the forecast for new product and

  spare parts. It's the complete product mix for the entire plant,

  including what we "sell" to other plants and divisions in the company.

  Having done that, we're now in the process of calculating the

  hours each "work center" has to contribute. We're defining a

  work center as any group of the same resources. Ten welders with

  the same skills constitute a work center. Four identical machines

  constitute another. The four machinists who set up and run the

  machines are still another, and so on. Dividing the total of work

  center hours needed, by the number of resources in it, gives us

  the relative effort per resource, a standard we can use for com-

  parison.

  Yesterday, for instance, we found the demand for injection

  molding machines is about 260 hours a month for all the injec-

  tion molded parts that they have to process. The available time

  for those machines is about 280 hours per month, per resource.

  So that means we still have reserve capacity on those machines.

  But the more we get into this, the more we're finding that

  the accuracy of our data is less than perfect. We're coming up

  with bills of material that don't match the routings, routings that

  don't have the current run-times—or the correct machines, as we

  just found out—and so on.

  "The problem is, we've been under the gun so much that a

  lot of the updating has just fallen by the wayside," says Stacey.

  "Hell, with engineering changes, shifting labor around, and

  all that happening all the time, it's just plain tough to keep up

  with it no matter what," says Bob.

  Ralph shakes his head. "To double-check and update every

  piece of data relevant to this plant could take months!"

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  149

  "Or years," mumbles Bob.

  I sit down and close my eyes for a second. When I open my

  eyes, they're all looking at me.

  "Obviously, we're not g°ing to have time for that," I say.

  "We've only got ten weeks now to make something happen be-

  fore Peach blows the whistle. I know we're on the right track, but

  we're still just limping along here. We've got to accept the fact

  we're not going to have perfect data to work with."

  Ralph says, "Then I have to remind you of the old data

  processing aphorism: Garbage in, garbage out."

  "Wait a minute," I say. "Maybe we're being a little too

  methodical. Searching a data base isn't the only way to find an-

  swers. Can't we come up with some other faster way to isolate the

  bottleneck—or at least identify the candidates? When I think

  back to the model of the boys on the hike, it was obvious who the

  slower kids were on the trail. Doesn't anybody have any hunches

  where the Herbie might be in the plant?"

  "But we don't even know if we've got one yet," says Stacey.

  Bob has his hands on his hips. His mouth is half open as if he

  might say something. Finally, he does.

  "Hell, I've been at this plant for more than twenty years.

  After that much time, I know where the problems usually seem to

  start," he says. "I think I could put together a list of areas where we might be short on capacity; at least that would narrow the

  focus for us. It might save some time."

  Stacey turns to him. "You know, you just gave me an idea. If

  we talk to the expediters. They could probably tell us which parts

  they're missing most of the time, and in which departments they

  usually go to look for'them."

  "What good is that going to do?" asks Ralph.

  "The parts most frequently in short supply are probably the

  ones that would pass through a bottleneck," she says. "And the

  department where the expeditors go to look for them is probably

  where we'll find our Herbie."

  I sit up in my seat. "Yeah, that makes a lot of sense."

  I stand up and start to pace.

  "And I'll tell you something 7 just thought of," I say. "Out on the trail, you could tell the slower kids by the gaps in the line.

  The slower the kid, the greater the distance between him and the

  kid in front of him. In terms of the analogy, those gaps were

  inventory."

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  150

  Bob, Ralph, and Stacey stare at me.

  "Don't you see?" I ask them. "If we've got a Herbie, it's

  probably going to have a huge pile of work-in-process sitting in

  front of it."

  "Yeah, but we got huge piles all over the place out there,"

  says Bob.

  "Then we find the biggest one," I say.

  "Right! That's got to be another sure sign," says Stacey.

  I turn and ask, "What do you think, Ralph?"

  "Well, it all sounds worth a try," says Ralph. "Once you've narrowed the field to maybe three of four work centers, it won't

  take long for us to check your findings against the historical data

  just to be sure."

  Bob looks at Ralph and says in a kidding voice, "Yeah, well,

  we've all seen how good that is."

  But Ralph doesn't take it in a kidding way. He looks embar-

/>   rassed.

  "Hey, I can only work with what I've got," he says. "What do you want me to do?"

  "Okay, the important thing is that we have new methods to

  try," I say. "Let's not waste time pinning the blame on bad data.

  Let's get to work."

  Fueled by the energy of new ideas, we go to work, and the

  search goes quickly ... so quickly, in fact, that what we discover

  makes me feel as though we've run ourselves straight into a wall.

  "This is it. Hello, Herbie," says Bob.

  In front of us is the NCX-10.

  "Are you sure this is a bottleneck?" I ask.

  "There's some of the proof," he says as he points to the

  stacks of work-in-process inventory nearby—weeks of backlog ac-

  cording to the report Ralph and Stacey put together and which

  we reviewed about an hour ago.

  "We talked to the expeditors," says Bob. "They say we're

  always waiting for parts from this machine. Supervisors say the

  same. And the guy who runs this area got himself a set of ear-

  plugs to keep him from going deaf from all the bitching he gets

  from everyone."

  "But this is supposed to be one of our most efficient pieces of

  equipment," I say.

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  151

  "It is," says Bob. "It's the lowest-cost, highest-rate means we have of producing these particular parts."

  "So why is this a bottleneck?"

  "This is the only one like it we've got," he says.

  "Yes, I know that," I say, and I stare at him until he explains.

  "See, this machine here is only about two years old. Before

  we installed it, we used other machines to do what it does. But

  this machine can do all the operations that used to take three

  different machines," says Bob.

  He tells me about how they used to process these parts using

  the three separate types of machines. In one typical instance, the

  process times per part were something like two minutes on the

  first machine, eight minutes on the second, and four minutes on

  the third—a grand total of fourteen minutes per part. But the

  new NCX-10 machine can do all three processes in ten minutes

  per part.

  I say, "You're telling me we're saving four minutes per part.

  Doesn't that mean we're producing more parts per hour than we

  were? How come we've got so much inventory stacked up for this

  thing?"

  "With the old way, we had more machines," he says. "We

  had two of the first type, five of the second type, and three of the

  third type."

  I nod, understanding now. "So you could do more parts,

  en though it took you longer per part. Then why did we buy

  e NCX-10?"

  "Each of the other machines had to have a machinist to run

  Bob says. "The NCX-10 only needs two guys on it for setups.

  ,e I said, it's the lowest cost way for us to produce these parts."

  I take a slow walk all the way around the machine.

  "We do run this thing three shifts, don't we?" I ask Bob.

  "Well, we just started to again. It took a while to find a re-

  placement for Tony, the setup guy on third shift who quit."

  "Oh, yeah ..." I say. Man, Peach really did it to us that day.

  I ask, "Bob, how long does it take to train new people on this

  machine?"

  "About six months," he says.

  I shake my head.

  "That's a big part of the problem, Al. We train somebody and

  after a couple of years they can go elsewhere and make a few

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  152

  dollars more with somebody else," says Bob. "And we can't seem

  to attract anybody good with the wages we offer."

  "Well why don't we pay more for people on this equipment?"

  "The union," says Bob. "We'd get complaints, and the union

  would want us to up the pay-grade for all the setup people."

  I take a last look.

  "Okay, so much for this," I say.

  But that isn't all. The two of us walk to the other side of the

  plant where Bob gives me a second introduction.

  "Meet Herbie Number Two: the heat-treat department," says

  Bob.

  This one looks more like what you might think of in terms of

  an industrial Herbie. It's dirty. It's hot. It's ugly. It's dull. And it's indispensable.

  Heat-treat basically is a pair of furnaces ... a couple of

  grimy, dingy, steel boxes, the insides of which are lined with ce-

  ramic blocks. Gas burners raise the internal temperatures to the

  1500-degree-Fahrenheit range.

  Certain parts, after they've been machined or cold-worked

  or whatever at ordinary temperatures, can't be worked on any-

  more until they've been treated with heat for an extended period

  of time. Most often, we need to soften the metal, which becomes

  very hard and brittle during processing, so it can have more

  machining done to it.

  So the furnace operators put in the parts, from a dozen or

  less to a couple of hundred, then they fire up the thing and cook

  the parts in there for a long time—anywhere from six hours to

  sixteen hours. And afterwards, the parts always have to go

  through a further cool-down to air temperature outside the fur-

  nace. We lose a lot of time on this process.

  "What's the problem here—we need bigger furnaces?" I ask.

  Bob says, "Well . . . yes and no. Most of the time these fur-

  naces are running half empty."

  "How come?"

  "It's the expediters who seem to cause the problem," he says.

  "They're always running over here and having us run five of this

  part or a dozen of that part just so they can have enough to

  assemble a shipment. So we end up having fifty parts wait while

  we heat-treat a handful. I mean, this operation is run like a bar-

  bershop—take a number and stand in line."

  "So we're not running full batches."

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  153

  "Yeah, sometimes we are. But sometimes even if we do a full

  batch in number, it's not enough to fill the furnace."

  "The batches are too small?"

  "Or too big in size, and we have to run a second heat to

  handle the pieces that wouldn't fit in the first. It just never seems

  to work out," says Bob. "You know, a couple of years ago, there

  was a proposal to add a third furnace, on account of the prob-

  lems."

  "What happened to it?"

  "It was killed at the division level. They wouldn't authorize

  the funds because of low efficiencies. They told us to use the

  capacity we've got. Then maybe they'd talk expansion. Besides,

  there was all kinds of noise about how we've got to save energy

  and how another furnace would burn twice as much fuel and all

  that."

  "Okay, but if we filled the furnace every time, would we have

  enough capacity to meet demand?" I ask.

  Bob laughs.

  "I don't know. We've never done it that way before."

  Once upon a time, I had an ide
a for doing to the plant essen-

  tially what I did with the boys on the hike. I thought the best

  thing to do would be to reorganize everything so the resource

  with the least capacity would be first in the routings. All other

  resources would have gradual increases in capacity to make up

  for the statistical fluctuations passed on through dependency.

  Well, the staff and I meet right after Bob and I get back to

  the office, and it's pretty obvious, awfully damn quick, that my

  grand plan for the perfect un balanced plant with Herbie in front is just not going to fly.

  "From a production standpoint, we can't do it," says Stacey.

  "There is just no way we can move even one Herbie—let

  alone two—to the front of production," Bob says. "The sequence

  of operations has to stay the way it is. There's nothing we can do

  about it."

  "Okay, I already can see that," I say.

  "We're stuck with a set of dependent events," says Lou.

  As I listen to them, I get that old familiar feeling which

  j anes whenever a lot of work and energy are about to go down

  the tubes. It's kind of like watching a tire go flat.

  I say, "Okay, if we can't do anything to change their position

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  154

  in the sequence, then maybe we can increase their capacities.

  We'll make them into non-bottlenecks."

  Stacey asks, "But what about the step-up in capacity from

  beginning to end?"

  "We'll reorganize . . . we'll decrease capacity at the head of

  production and increase it each stage on through," I suggest.

  "Al, we're not just talking about moving people around. How

  can we add capacity without adding equipment?" asks Bob. "And

  if we're talking about equipment, we're getting ourselves into

  some major capital. A second furnace on heat-treat, and possibly

  a second n/c machine . . . brother, you're talking megabucks."

  "The bottom line," says Lou, "is that we don't have the

  money. If we think we can go to Peach and ask him for excess

  capacity for a plant that currently isn't making money in the mid-

  dle of one of the worst years in the company's history . . . well,

  excuse my French, but we're out of our goddamned minds."

  E.M. Goldratt

  The Goal: A Process of Ongoing Improvement

  Captured by Plamen T.

  155

  19

  My mother and the kids and I are having dinner that eve-

  ning when Mom says to me, "Aren't you going to eat your peas,