I’m a big fan of Eliyahu M. Goldratt, especially his Theory of Constraints (TOC) and his work on bottleneck management. Goldratt revolutionized the manufacturing world with a critical insight into the overall operation of the plant floor. Before Goldratt, local efficiency was the key metric used to measure performance on the shop floor.

Eliyahu M. Goldratt-blog

Imagine that a factory had 10 workstations, each with a specific job in the manufacturing process. At one time, management believed that if each machine was running close to 100 percent of the time then the factory would be as efficient as possible. However, Goldratt figured out that the factory as a whole could only run as fast as its slowest machine. By running machines at disparate production rates to full capacity, the manufacturer was, in fact, just creating work in process (WIP) and jamming up the factory shop floor.

One simple example in Goldratt’s book, The Goal, involves a Boy Scouts’ hiking trip. If everyone is going to stay together, the troop can only walk as fast as its slowest scout. By offloading the slowest scout’s backpack to the fastest kid, Goldratt simultaneously increased the speed of the slowest hiker and decreased the speed of the fastest hiker. As a unit, the troop moved more quickly, even though the fastest hikers were not walking as fast as they could.

By increasing the production rate of the slowest machine in the factory, you increase the manufacturing capacity of the entire factory. One of the most elegant aspects of TOC is that with each bottleneck or constraint you find and resolve in the system, you discover a new one. The inefficiency you resolved unveils an inefficiency that was, heretofore, unnoticed. It is a process of continuous improvement, where fun and profit is found in discovering and resolving problems.

  • Tony Welsh

    The above example is spurious, which is typical of Goldratt’s woolly thinking. The scouts’ walking happens in parallel, whereas the factory processes generally happen in series.

    The whole area of manufacturing scheduling had been well explored decades before Goldratt, and more abstractly the central tenet of TOC – that there is exactly one constraint – was known to be wrong since the time of Pascal. He only lacked a computer, but his work was the basis of the modern discipline of mathematical programming pioneered by Danzig and others. The number of constraints is equal to the number of degrees of freedom, and this applies to systems as diverse as car suspensions and factories.

    As for keeping machines 100% occupied for efficiency, it depends what you mean by “efficient.” There is a well-known conflict between maximizing through-put and minimizing production time – as exemplified by queuing theory – so it depends what you want. This seems to be denied by some Goldratt adherents, of not by the man himself.

    Imho Goldratt’s pseudo-science has added nothing to our understanding
    of anything.

    • http://visualizetheplan.com/ Tim Mather

      Thanks for your comment Tony.

      I’m not familiar with the author your are referring to but I’m going to add it to my book list and circle back with you.

    • Roger Berntsen

      Dear Tony, I’m not a religious TOC man and I do not look at Goldratt as the Messiah of modern industry , but pay attention to what he says. He argues that the TOC and bottleneck theory only complies to organizations that want to make money.

      So, he only talks about those who want to make money. To make money is easy to express in mathematically terms, Increase Throughput (Sales), ROI and Cash flow simultaneously. And you make money.

      Any other “queuing theory” may have other goals and you can’t compare it with Eli’s theory. Thats also what you saying… with your thoughts about “efficiency”

      • Tony Welsh

        Roger, all companies want to make money, or at least that is the assumption behind most management science. Indeed, it is normally assumed that they want to maximize the NPV of their profit subject to certain legal and other constraints.

        TOC concentrates on removing or relaxing those constraints within management control, but makes the erroneous assertion that at any time there is only one of these. This is true only if management has only one decision variable, which is generally not true. Pascal showed that the number of effective constraints was equal to the number of decision variables, and the advent of computers has made it possible for Dantzig and others to exploit this in practice using a technique called linear programming. This not only tells you how to maximize profit within the existing constraints, but can list _all_ active constraints together with the marginal benefit of relaxing each one.

        My point is that Goldratt ignored all this prior and much superior work. All adherents of TOC should read Dantzig.

        As for queuing theory, this is a side issue, but it is a fact that there is a conflict between throughput and turn-around time. Depending upon the nature of ones business, maximizing profit may require maximizing throughput or minimizing turn-around time, or most likely a combination of the two. My point was merely that I have heard TOC protagonists claiming they can maximize both at once.