January 10, 2008
News & Opinion: Excerpt from Reaching the Goal: How Managers Improve a Services Business Using Goldratt's Theory of Constraints
The following excerpt is Chapter 3 of Reaching The Goal: How Managers Improve a Services Business Using Goldratt's Theory of Constraints by John A. Ricketts. From the publisher: "Ricketts draws on Eli Goldratt's Theory of Constraints (TOC), one of this generation's most successful management methodologies...thoroughly adapting it to the needs of today's professional, scientific, and technical services businesses. He reveals how to identify the surprising constraints that limit your organization's performance, execute more effectively within those constraints, and then loosen or even eliminate them."
Chapter 3: Theory of Constraints
Drum-Buffer-Rope (DBR) is the TOC application for operations.(8) It's often used to plan and manage discrete manufacturing, but DBR has also been used by service providers as diverse as landscapers and hospitals, even though their services are perishable.
DBR gets its name from the roles that specific elements play during scheduling and management of production. To appreciate those roles, it helps to know the problems DBR was originally intended to solve. For that we need a quick review of the state of the art in manufacturing when DBR was invented.
A simplified manufacturing process composed of just five steps is illustrated in Figure 3-1. Of course, actual manufacturing processes are often composed of many steps in complex configurations shaped like a V, A, T, I, or some combination thereof. But a simple sequence of steps is sufficient to illustrate the essential elements of DBR.
The steps are numbered 1 through 5. Assume that each step in this illustration is performed by a different machine type. RM stands for raw materials, WIP stands for work in process, and FG stands for finished goods, which are all forms of inventory.
One problem with traditional manufacturing is immediately apparent in the figure: There's a lot of inventory before, during, and after production. That's a problem because inventory is a significant investment, and it doesn't generate revenue until it's sold.
Another problem is that excess inventory impedes the production process. That is, as the shop floor becomes crowded with WIP, it gets harder to monitor due dates and ensure that the most urgent jobs are done first. The busier the shop gets, the less effective expediting becomes.
Thus, a third problem is that it's hard to predict when each job will be completed. Once jobs are released into the shop, they are hard to control. Some jobs may finish early, but too many finish late, which leads to customer dissatisfaction and missed sales. So as production slows, jobs may be started earlier, thereby further increasing WIP, slowing production, and perpetuating the push cycle.
Work also gets pushed into and through the factory by the desire for high utilization, a measure of how long each machine and each worker are actually performing tasks in the production process. The underlying assumption is that anything less than high utilization on every machine and every worker represents a lost opportunity for production.
Though appealing, that assumption is flawed. For one thing, producing goods that customers won't buy is wasteful, no matter how high it drives utilization. Yet even when customers will gladly buy what's produced, the push for universally high utilization overwhelms the constraint.
Somewhere in that production process is a step that cannot produce as many units per time period as the rest of the steps. That's the constraint. You can't see it in the figure, and neither can most managers in an actual manufacturing plant managed the traditional way. Fortunately, the constraint is hiding in plain sight, and with a little detective work it can be found.
What's far harder to do is change the perception that high utilization everywhere is a good thing. The belief that local optimizations somehow add up to global optimization is strongly held. Until this policy constraint is broken, however, the physical constraint cannot be managed.
The same five-step manufacturing process as before is illustrated in Figure 3-2, but it also includes the elements needed for DBR. Solid lines represent product flow. Dashed lines represent information flow. In an actual factory with many products and a variety of routings, there can be more than one constraint. To illustrate DBR, however, one will do.
Step 3 is the constraint. It's also called the drum because it sets the pace for the rest of the steps. That is, upstream steps will occasionally be idle--have less than full utilization--so they won't overwhelm the constraint with work. And downstream steps will likewise occasionally be idle because they're waiting for the constraint to complete its step. But if all goes well, the constraint itself will have consistently high utilization, excess WIP will disappear, and more orders will ship on time.
When this happens, the factory is producing as much as it can, subject to the current constraint. This level of production is typically much more than it ever could produce under traditional manufacturing when the constraint was invisible. So managers in a factory adopting DBR may go from wrestling with insufficient capacity to having ample capacity.
A likely place to see WIP is ahead of the constraint, because it can produce less than any other step, by definition. Therefore, that WIP is sometimes mistaken for the buffer, but the drum buffer is actually all work scheduled on the constraint, even if it's currently at an earlier step. That is, the buffer is measured in time, not physical WIP units. So a view of the true buffer is typically contained in an information system nowadays.
If all jobs ahead of the constraint are early or on schedule, the amount of work needed to keep the constraint busy is adequate, and the buffer is said to be in the green zone. However, when some jobs are behind schedule and the possibility that the constraint could run out of work becomes significant, the buffer is in the yellow zone.
Because normal variation causes some jobs to run early at the same time that others run late, it's possible that the constraint won't actually run out of work. Hence, a yellow buffer does not automatically trigger action. However, when many jobs are behind schedule and it becomes clear that the constraint will indeed run out of work without action, the buffer is in the red zone. Upstream steps then have to sprint to refill the buffer, and thereby keep the constraint busy, while downstream steps may have to sprint to finish late jobs on time.
In addition to the drum buffer, which contains WIP, the shipping buffer contains FG. Because the market is the ultimate pacesetter, the shipping buffer protects customers from late delivery, just as the drum buffer protects the constraint from overloading.
The third and final element of DBR is the ropes, which govern when gating events occur. The shipping rope governs work on the constraint needed to meet market demand and keep the shipping buffer green. The constraint rope governs the release of raw materials to start new jobs that should keep the drum buffer green.
Under DBR, jobs are released much closer to their due date than in traditional manufacturing because they will spend less time waiting between steps. Like the buffer, the length of ropes is measured in time, and the ropes are actually contained in an information system.
An information system that supports DBR leads to global optimization by optimizing the constraint rather than every step in the production process.(9) Buffer management thus keeps the process in control without requiring constant attention and fine-tuning. If a factory has more than enough capacity to meet market demand, the constraint is said to be external or in the market. In this case, the shipping buffer, not the drum buffer, then regulates when jobs are released into the shop because the internal constraint no longer limits production. If a factory has enough capacity to meet market demand during normal and slack periods but not during peak periods, its dominant constraint is in the market even though it occasionally has an internal constraint.(10) In this case, a simplified form of DBR can be implemented that makes sensible trade-offs between keeping the constraint busy and satisfying customer demand in order to protect future sales. In both cases, when the constraint is external, no step has full utilization, including the internal constraint, but this is what keeps the factory from producing excess inventory that cannot be sold. The next section covers another TOC application that specifically addresses a market constraint.
Whenever the constraint shifts (due to changes in machines, people, process, products, or demand), DBR has to be reconfigured accordingly. This is a nontrivial effort, and it's why capacity in a DBR shop is deliberately unbalanced to prevent floating constraints.
Placement of the drum, however, should be strategic, not accidental. That is, when DBR is first implemented, the constraint location may not be correctly aligned with respect to profitable market opportunities. If so, rather than implement DBR around this previously unseen constraint, it generally makes more sense to adjust capacity so that the control point represented by the drum is relocated to a position where the factory will be better able to produce goods that meet market demand profitably.
DBR is also known as Synchronous Manufacturing. In a nutshell, here's how it compares to two other widely used production management approaches:
- In their pure forms, Enterprise Resource Planning (ERP) assumes infinite capacity and schedules all steps, while DBR assumes finite capacity and schedules just the constraint. Some ERP software can schedule to finite capacity, but it does not have other essential capabilities of DBR software. For instance, ERP prohibits late release of materials, while DBR prohibits early release because it increases WIP. Moreover, ERP drives material requirements all the way through the bill of materials (BOM), regardless of stock on hand, while DBR takes existing stock and buffers into consideration. Thus, ERP and DBR are fundamentally different solutions.
- Lean/Just-in-Time (JIT) seeks to optimize individual steps, while DBR optimizes the entire process around the constraint. They are fundamentally similar, but Lean/JIT doesn't work as well in job shops as flow shops because job shops have more diverse and changeable routings.
The benefits of DBR are substantial. One literature review found the following average improvements across 82 companies:(11)
- 70 percent reduction in lead time
- 65 percent decrease in cycle time
- 44 percent improvement in due-date performance
- 49 percent reduction in inventory
- 63 percent increase in revenue
A central benefit of DBR is to change the production process from push to pull: Nothing gets produced unless there's a market for it. Market pull through the internal constraint then optimizes production while minimizing inventory.
Because the market is the key driver of DBR, how demand ripples back through the distribution chain from customers to factory affects DBR. This connection leads to the next TOC application.
8. Kelvyn Youngman, "A Guide to Implementing the Theory of Constraints (TOC)," http://www.dbrmfg.co.nz, 2005.
9. Eliyahu Goldratt, Eli Schragenheim, and Carol Ptak, Necessary But Not Sufficient, North River Press, 2000.
10. Eli Schragenheim and H. William Dettmer, Manufacturing at Warp Speed: Optimizing Supply Chain Financial Performance, St. Lucie Press, 2000, pp. 151–152.
11. Victoria Mabin and Steven Balderstone, The World of the Theory of Constraints, St. Lucie Press, 2000, pp. 11–12.
Reaching the Goal: How Managers Improve a Services Business Using Goldratt's Theory of Constraints
By John A. Ricketts, published by IBM Press, November, 2007
Copyright 2008 by International Business Machines Corporation. All rights reserved.