Key principle: variation in processing and material transfer times should be taken every bit as seriously as variation in critical-to-quality characteristics.

• A production line cannot work faster than the slowest workstation. If it tries, inventory will accumulate. Constrictor snakes such as the one shown at the right are designed to process meals in the indicated manner, and they often sleep for several months while they digest their meals. This is not how we want a factory to operate; pig-swallowing should be left to systems that are designed to handle it.
• Standard, Charles, and Davis, Dale. 1999. Running Today's Factory: A Proven Strategy for Lean Manufacturing. Cincinnati, OH, Hanser Gardner Publications (1999, 111-112) uses the phrase "pig in a python" to describe large inventory bubbles that move through a factory. "If smaller orders are released more often, the factory resources are loaded much more easily. …This is analogous to the python swallowing dozens of little piglets instead of one large pig. …Surprisingly, many factories prefer to 'stretch the python' so it can swallow an even larger hog!" The reference mentions an explorer who saw an anaconda swallow an adult hog in a Peruvian rain forest. (Dale Davis studied anthropology and specialized in Maya culture.) However, Levinson (editor), Leading the Way to Competitive Excellence (1998, ASQ Quality Press) coined the phrase "pig-swallowing," and even included an actual photo of a recently-fed python from The Serpent's Den in Milford, PA.
• Goldratt example: a line of soldiers or Boy Scouts cannot march faster than the slowest person (the constraint).
• If they try, the line will extend in length as the faster people get ahead of the constraint. The distance between them represents inventory.
• The only way to get more capacity (or go faster) is to elevate the constraint, or improve its performance.


The Theory of Constraints introduces some very interesting (and important) economic considerations. All of these result from the fact that time lost at the constraint is lost forever: it can never be made up. The cost accounting system does not reflect the true effect of lost time at the constraint accurately. (Remember that accounting statements whose purpose is to satisfy the requirements of the Internal Revenue Service and the Securities and Exchange Commission are rarely suitable for making decisions in the factory!)
 

• The true loss due to any scrap, rework, or stoppage in the constraint is the marginal profit on the lost unit of production.
• Rework in the constraint can therefore be worse than pre-constraint scrap!
• Post-constraint scrap is irreplaceable.
• Theory of Constraints uses only three performance measurements:
1. Throughput: finished goods that are shipped to customers. More is better.
• The marginal profit (sale price minus raw materials) on each transaction is a good measurement.
2. Inventory: Money that is invested in items the factory intends to sell. Less is better.
3. Operating Expense: Money spent to convert inventory into throughput. Less is better.


Note the total absence of considerations like "absorption of overhead," "labor per unit," "equipment time per unit," and other things that might interest the accounting department but which are largely irrelevent to the true economics of running the factory.

The capacity-constraining resource may change with the product mixture. Linear programming is very useful for assessing these situations.

• LP Simplex can identify the constraint(s) and quantify the slack capacity (excess capacity) in the non-constraints.
• It can optimize the product mixture for maximum profit.
• It can account for market constraints (lack of demand) and contractual obligations as well as capacity constraints.
• Shadow prices show the marginal effect of increasing a constrained resource, i.e. elevating a constraint.

SFM is a pull manufacturing system (like kanban) but it offers some apparent advantages over kanban. In a kanban system, each downstream operation pulls work from the next one upstream. In SFM, the only information transfer is between the capacity-constraining resource (CCR) and production starts. The CCR pulls work into the factory in the form of production starts. It is always desirable to keep a buffer of work in the line upstream of the CCR because a shortage at the CCR means an irrecoverable loss of production time.

• The constraint operation limits the overall rate of production. This operation "beats the drum" to control the factory's speed.
• This is achievable by tying the constraint to production releases by a rope.
• No need for any other production controls, such as kanban!
• An inventory buffer at (or heading toward) the constraint protects it from upstream stoppages.
• This gives us Drum-Buffer-Rope (DBR) production control

The matchsticks-and-dice experiment from Goldratt and Cox, The Goal, shows why it is seemingly impossible to achieve 100% utilization in a balanced factory. As a matter of fact, when you try the simulation (below), you'll end up with substantial work-in-process (WIP). Average production will be substantially below 3.5 units per turn even though this is the factory's theoretical capacity.

Henry Ford said he wanted every employee to have "all the time he needed but not a second more" to perform his or her task. Given the simulation from The Goal and the fact that cycle time in queue is proportional to u/(1-u) where u is utilization, this sounds like a formula for a deranged nightmare in which inventory buries the entire factory.

The equation (Standard, Charles, and Davis, Dale. 1999. Running Today's Factory: A Proven Strategy for Lean Manufacturing. Cincinnati: Hanser Gardner Publications) allows reconstruction of the technique that Ford used to do the seemingly impossible. The cycle time in queue is also proportional to the variation coefficients for arrivals at the workstation (a function of material transportation in the factory) and effective processing time.


• Variation in arrival rate at the workstation
• Batch-and-queue operations aggravate this problem substantially by necessitating the transportation of batches of products. Ford achieved single-unit flow (recognized as highly desirable today).
• Ford Highland Park plant: Work slides allowed immediate transfer of finished parts to the next workstation.
• Ford also used work cells (or their equivalent), with all the machines necessary to make a part close together and in sequence of operation.
• Variation in processing time
• Automation helped remove human-induced variation from each operation. Consider the following example:
• Worker using screwdriver ==> considerable variation in assembly times, noting that there is variation in the time it takes the worker to turn his or her wrist. (Also very inefficient, and an invitation to repetitive-motion injury).
• Worker using powered screwdriver ==> Now most of the variation comes from setup (in the form of placing the screwdriver in the screw's slot) because the screw is driven at a constant rate.
• Automated multiple-spindle screwdriver (actually used by Ford) ==> Essentially no variation in assembly time.
• Unitary machines performed many operations between loading the part and unloading it, thus eliminating the variation associated with load/unload.
• Subdivision of tasks suppressed human-induced variation as well. Example from Rudyard Kipling's Captains Courageous: cleaning cod on a commercial fishing boat
• Inefficient: One worker does the entire task: pick up knife, slit the fish open, cut off the fish's head, put the knife down, pull out and discard the insides except for the liver, which goes in a separate container, pick the knife up and cut out the fish's backbone. There will obviously be substantial variation in processing time (although there is no downstream operation that suffers as a result, as the cleaned fish goes straight into a tub of salt water). Also, note the waste motions: "pick up knife" and "put down knife."
• Efficient: Three men working together: the first man, who never puts his knife down, slits the fish and makes cuts at its neck. The second man pulls off and discards the head, removes the entrails, and puts the liver in the cod liver container. The third man cuts the fish's backbone out and tosses the fish into a tub of salt water. Furthermore, tool maintenance is done offline (externalization of setup, per SMED). A fourth person, often a boy or apprentice fisherman, sharpened the knives and exchanged them for the cleaners' dull ones.

Henry Ford (who repaired watches as a boy) designed his factory to run like a clock. The operations had to keep time with each other almost perfectly to prevent inventory buildup at slower workstations and shortages at faster ones. This brings to mind the concept of a pendulum or metronome, the swing of a conductor's baton, and the beat of a drum.

Goldratt and Cox's Theory of Constraints suggests that some operations will always have excess capacity, and that the capacity-constraining resource should set the pace for the entire operation. This is absolutely true but three familiar performing arts: ballet, orchestras and bands, and marching bands show that you can successfully run an activity in which no "operation" has any excess capacity. In these performing arts, people (with their built-in variation) constitute the "equipment" or "workstations."


• Ballet: individual dancers or separate groups will often perform different sets of motions, all of which must coordinate perfectly. In many cases, no one stands around doing nothing (i.e. no one has "excess capacity") while waiting for others to catch up. The music from the orchestra, of course, sets the timing for all the dance movements: a concept very similar to the "drum" in drum-buffer-rope and also takt time.
• Orchestra or band: This is a little easier than ballet because, although musicians play different instruments, they are using the same set of notes. Again, however, the swing of the conductor's baton or a memorized beat keeps everyone together.
• Marching bands: Think of a football halftime performance in which a marching band plays music while turning itself into different shapes on the football field. This is much harder than marching in step (which soldiers have done for thousands of years) because everyone is going in the same direction. Elements of the marching band must go off in different directions and then rendezvous at prescribed intervals. The music again sets the pace but the length of the steps and the directions must be carefully prescribed.
  
Now go back to factory equipment which, if properly designed, has far less inherent timing variation than even a skilled ballet dancer or member of a marching band. The fact that performing arts can achieve 100 percent utilization of all the performers (who are performing different activities, as opposed to soldiers who are simply marching in a single formation) suggests that a factory also is capable of achieving 100 percent utilization of all of its equipment.

The matchsticks-and-dice exercise from The Goal is an excellent way to show the effects of variation in processing and material transfer times. One should not, however, assume from this exercise that the variation is unavoidable and nothing can be done about it. As shown above, it is possible to suppress both forms of variation. The key message, though, is that variation in processing and material transfer times should be taken just as seriously as variation in critical-to-quality characteristics.

Wall Clock for the Theory of Constraints $13.99 Benjamin Franklin and Henry Ford (who cited Franklin as an influence on his thinking) remind your visitors that time lost at the constraint is lost forever. The Franklin quote is from The Way to Wealth and the Ford quote is from Ford Ideals (1922, in the public domain due to age).

Buttons and Magnets: $1.75, 10 for $15.00, 100 for $89.50 2.25 inch diameter, red and blue lettering on white background.