Operations, as Taught by Eli Goldratt · Lesson 6 of 8
Putting Drum-Buffer-Rope to work
Three practical steps, each with a trap. Who to ask about the bottleneck (and who not to), why the buffer should simply be half of today's lead time, and how to tie the rope in V, A and T plants without starting a revolt.
Daniel decides to implement DBR in the plate mill. First question: where is the constraint? He calls three people into his office and gets three answers.
Step 1: identify the constraint
Trying to identify the constraint by calculation is a recipe for confusion: the data isn’t accurate enough. The best way is to ask the people who suffer from it. But getting a straight answer isn’t easy.
Ask the manager of the suspected department, and you put them in a dilemma. To avoid the hot seat, they want to say their department isn’t a constraint. To avoid pressure for cuts, they don’t want to admit spare capacity either.
Ask people higher up, like the production manager and the materials manager, and you’ll get different answers, depending on where overtime was last approved or which department was last accused of starving another.
Ask the expeditor. The expeditor is the only person working with only the orders in mind, without caring about local efficiencies. Ask which department they have to visit more than once a day. “Every day I go to the finishing line three times. If I tell the foreman to break his set-up once more, he’ll kill me.”
At Northgate, the expeditor says, “It changes every week.” Management names three different departments. What does Daniel do?
- Run a capacity study for a month, then pick the most loaded department
- Pick the most expensive machine
- Act as if there is no bottleneck: use current market orders as the drum
If the expeditor or all of management point to one resource, accept it as the drum. If not, act as though there is no bottleneck. DBR removes the effect of efficiencies and oversized batches, the flow smooths out, and a real bottleneck, if there is one, will reveal itself.
Step 2: choose the size of the buffer
Too small a buffer means chaos: not enough inventory, everything missing. Too big a buffer also means chaos: so much inventory that nobody knows the right priorities. Between them is a wide safe zone.
Calculating the buffer is, again, a waste of time: the data isn’t good enough. Goldratt’s rule is simpler:
Buffer = half of the existing lead time.
Today’s buffer is far too big (we’re on the right-hand slope of the chaos curve), so halving it lands safely in the safe zone. Two worries usually follow:
- “For work that goes through the bottleneck, half seems too small.” DBR stops the flooding of the floor, so queue times drop drastically. Halving is very safe.
- “For work that doesn’t go through it, half seems too big.” Perhaps, but your new lead time and work in progress are still half of what they were. Not a bad start. Buffer management will refine it.
Step 3: tie the rope
Release schedule = the drum schedule minus the buffer. Never release work ahead of schedule. Where the rope must be tied depends on the plant type:
- A-plants: tie it where work is released, so parts don’t race ahead of each other.
- V-plants: because of the divergences, tie it everywhere: each work centre gets a list of release dates and quantities, and must not work ahead of it. Otherwise stealing comes straight back.
- T-plants: tie it at release and at assembly, where stealing between variants happens.
Exercise: where should the rope be tied?
- A paper mill cutting jumbo rolls into hundreds of sheet sizes
- A machine builder assembling a few large machines from many parts
- An appliance maker assembling dozens of models from shared components
The culture change
Tying the rope is where DBR clashes most with convention. The old paradigm motivated people to work as much as they can; the rope requires them to stand idle from time to time. Instructing them isn’t enough. We must explain why. People must understand the concepts: where the drum is, what the buffer is, how the rope works, and therefore how to work by the Roadrunner ethic. Goldratt used the film of The Goal to start that conversation.
Implemented well, DBR shortens lead times, increases throughput, enables on-time delivery and reduces inventory, and with that trims many of the undesirable effects. But it isn’t the end of the story. Some important questions are still open.
Check your understanding
Answer all five. Get four right to complete the lesson.
1. Who is most likely to name the real bottleneck?
- The department manager
- The cost accountant
- The expeditor
- The software
The one person who works only with the orders in mind.
2. If answers about the bottleneck conflict, you should…
- Use current market orders as the drum
- Pick one at random
- Calculate loads for a month from the routings
- Wait
A real bottleneck will reveal itself under DBR.
3. Goldratt’s starting rule for the buffer is…
- One day
- The average process time
- Twice the lead time
- Half of the existing lead time
Today’s buffer is far too big; halving lands in the safe zone.
4. In a V-plant, the rope should be tied…
- Only at release
- Everywhere, or stealing occurs
- Only at the bottleneck
- Nowhere
Divergences at every step.
5. For people to obey the rope, it is necessary to…
- Pay a bonus
- Lock the store room
- Explain why, so they understand the drum, buffer and rope
- Measure them on efficiency, so they keep working while waiting
Instructions without understanding don’t survive.