Drum Buffer Rope Part 6

In my last post we discussed the various types of Drum Buffer Rope systems with both single drums and multiple drums.  In this final post in this series I will discuss one additional usage of multiple Drum Buffer Rope and then present the total view of Drum Buffer Rope. 

An additional variant of this same concept can also exist in a production/manufacturing situation where the buffer supports different lines, but the buffered product is the same for each “I” line.  An example of this situation might be a maintenance and repair organization (MRO) for aircraft.  The lines in this case would be equivalent to hangar space.  The drum would be the length of time required to repair each aircraft, and the buffer is equal to the number of aircraft waiting to enter the repair system.  Each hangar can and does repair the same type of aircraft.  The difference becomes the duration of the repair cycle time.  Some will be faster and some will be slower.  It’s the total repair time duration that determines the drumbeat in each “I” line.  Aircraft waiting in the buffer would be required to have the repair problem isolated (to some reasonable level), and fully ready to enter with the necessary repair parts, the ability to move aircraft in and out of the hangar and a crew waiting and ready to perform the repair.

The Total View

Even with all the respectable improvements that can be achieved with a synchronized flow using traditional DBR, S-DBR or even M-DBR, there can also be some problems associated with achievement, especially with traditional Drum Buffer Rope.  It’s not a bad problem, just one you need to be aware of.  When you follow Goldratt’s Five  Focusing Steps, it is possible during Step 2 (the exploitation step) that a constraint can be improved to the point that it is no longer the constraint, and at times this can happen very quickly.  When it does happen, you have effectively “rolled” the constraint to a new location, which means you only finished Step 2 before it is now time to go back to Step 1 again.  The original system process that was considered to be the constraint today is no longer the constraint tomorrow.  These types of rapid system improvements can obviously cause some problems.

When a new constraint is identified in the system, then the system effectively has a new drumbeat.  When that happens, you also have to move the buffer location(s) to reside in front of the new constraint(s), and you have to move the rope signal from this new location back to the release point at the front of the line.  In some systems it might be possible to roll the constraint several times to several different locations before an acceptable level of system stability is achieved.  This fast action of fixing and rolling the constraint can and does cause a certain amount of chaos in a system.  Workers will quickly become confused about “Who is the constraint today?” 

Improvements can happen so fast that the negative effects of change will outweigh the positive effects of improvement.  This was a problem recognized early on by some implementers of the Theory Of Constraints and Drum Buffer Rope concepts, and there are some simple and robust solutions to overcome this phenomenon.

This completes my blog post series on Drum Buffer Rope. In my next post, I will begin a new series of posts on a new subject.

Bob Sproull

Drum Buffer Rope Part 5

In this post we will discuss different types of constraints as they relate to Drum Buffer Rope, These constraints are basically internal and external constraints.

Different Types of Constraints

Constraints can exist in one of two different types.  The first type is the internal constraint— which means that the market demand for your product is higher than the capacity of the system to produce it.  Customers want much more of what you have to  offer them, but you lack the capacity to produce more.  It’s a good situation to be in, but only up to a certain point.  If you can’t figure out a way to meet market demand, then your competitors will usually figure out a way to do it for you.  In other words, they will search for other sources.  This situation is ideal for implementing traditional DBR to meet the demand and capture more market.

The second type of constraint is an external constraint.  In this case the market demand is less than your ability to produce.  The market is buying less, and in some cases much less than you can produce.  This is a less desirable situation, but one that nonetheless can exist.  This situation usually means that there is not an internal constraint to contend with. If this is the case, then it is somewhat improbable that traditional DBR will provide an acceptable solution to this conundrum.  Instead, in this situation, a modified or simplified form of DBR might be more practical.  Consider S-DBR.

Drum Buffer Rope Variants

Simplified Drum-Buffer-Rope (S-DBR)

The concept of S-DBR was developed by H.W. Dettmer and E. Schragenheim is defined in their book Manufactruing at Warp Speed¹.  The S-DBR concept assumes that the constraint is external to the system and resides in the market segment.  Customers aren’t buying as much product as you can make, or there is significant variation in market demand, which can cause the constraint to float back and forth between internal to external locations.  In this situation, the constraint becomes interactive by moving between the market constraint (external) and the production constraint (internal).  This oscillating cycle between internal and external constraints can cause its own brand of chaos in deciding which market segments should be pursued and which ones might be better left alone.  Either way it is a decision that must be dealt with.

In the scenario of an external constraint, the drum is determined and activated only when the system has firm orders in place.  The rope is now determined by the orders that actually exist, which are released based on due dates.  If the orders exceed the capacity of the system, then the constraint has become internal and different actions must be taken.  This also assumes that the internal constraint will exist only for short periods of time and can be overcome by actions like implementing additional shifts or short-term overtime.  Dettmer and Schagenheim have argued, quite successfully, that the market is the true constraint of any system.  There is much more reading available about this concept at Dettmer’s Goal Systems web site.

Multiple Drum-Buffer-Rope (M-DBR)

There is another unique situation that can require the implementation of a third type of Drum Buffer Rope, known as Multiple-Drum-Buffer-Rope (M-DBR). The situation for M-DBR is created when a single buffer location is required to supply products to more than one assembly line, and each assembly has its own drum that is keeping pace at a different rhythm. Figure 2 shows an example of an M-DBR configuration.

 Figure 2.  The M-DBR Model.

An example might be a surface mount technology (SMT) machine that is required to make different types of circuit cards for different assemblies, and each circuit card assembly flows down different assembly lines.  It’s similar to moving from the apex of an “A” plant into the base of a “V” plant, with different parts now required for a number of different “I” lines (see Appendix 8 for discussion on plant types).  The demand (drumbeat) for each “I” line will vary, but each line requires its input from a single SMT machine or a series of SMT machines.

In the configuration of multiple drums, there are also multiple ropes, and each has different requirements.  There are two signal points for the rope.  The first would be back to the buffer to release more work for that particular line.  The second rope goes back to the SMT machine to support the needs of the buffer, which in turn releases work at the front of the SMT line.

The advantage of this concept is to reduce the tendency for economic batch size quantity at the SMT machine.  Many organizations believe when they set up a machine to make parts, they should make as many parts as they can, especially if the machine is expensive, or the setup times seem especially long.  The thought that this economic batch size quantity somehow saves money is, at best, absurd.  The economic batch size only serves to slow down throughput in the system.  No money has been saved at all! In fact, it will cost additional amounts of money because throughput will have been damaged (revenue lost) and the dollars will have been spent to buy those raw materials and parts that aren’t needed yet.  Instead, you should manage the constraint, conduct the setups in the sequence and frequency required from the drumbeat in the lines and solve the problem of shorter setup times as they occur using Single Minute Exchange of Dies (SMED) techniques.  The action of the machine should be to support the buffer for the various “I” line drums, not maintain “high efficiency” at the expense of making money (remember the goal of a company?).

Drum Buffer Rope Part 4

In this post we continue with our discussion of TOC's scheduling system known as Drum Buffer Rope, focusing on the key elements of buffers.

The systems inventory not only includes the work located at the buffer, but also the cumulative total of inventory (work) at other process locations as well.  It is possible, and recommended, that you establish an additional buffer at the shipping location.  The shipping buffer can be used to help control any system variation that occurs after the constraint.  Bad things can and do happen after processing at the constraint.  The constraint buffer provides the necessary protection in front of the constraint, and the shipping buffer provides protection after the constraint.  The shipping buffer is just a mechanism to absorb and manage the inevitable variation that will occur.  Buffer sizing at these two locations is a variable, but you do need to start with something.  Consider, as a starting point for the buffer at the drum (the constraint) location to be about one and a half for whatever units of time you are measuring.

For example, if your constraints can produce ten units in one day, then the buffer should be set at fifteen units (or 10 x 1.5 = 15).  You may decide in time that the buffer is too large or too small, so you can adjust it either up or down depending on the need.  The shipping buffer could be three or four days or less depending on the speed of product through the system.  It doesn’t need to be necessarily large in quantity or long in time.  It just needs to be sufficient to protect against variation after the constraint.  It’s also important to consider your shipping buffer time in your scheduling calculation to determine the correct release date into the system for on-time delivery.  If you watch your buffer locations carefully, you can make good decisions to increase or decrease them based on some supportive historical data.  If the buffer is constantly on the high end, then reduce it.  If it is constantly on the low end, then increase it.  Apply the rule of common sense to determine the correct buffer.

When you know and understand the constraint location, and you buffer the work activity, and you send the correct release signal to the front of the line to release more work, then you have in essence implemented a system of synchronized flow.  Figure 1 defines the DBR steps and integration.

 

 Figure 1. DBR Steps and Flow.

But wait!  With a synchronized flow, and actively implementing system subordination, there is a very high probability that the performance metric of efficiency will deteriorate quickly, at least for some period of time.  It will manifest an unacceptable efficiency performance metric that is considered undesirable by most companies.  The new mantra will be to “stop the synchronization nonsense and improve the efficiency.”  Be careful what you consider to be nonsense.  In this case, the real nonsense is the efficiency metric.  When the synchronized flow is implemented, then excess capacity at non-constraints will be quickly exposed, at least for some period of time.  Based on the efficiency metrics it will appear that everything is falling apart, and you are headed in the wrong direction.  But through time, the new system reality and thinking will expose new evidence about what is actually happening in the system.  The new reality is this:

·       [PJ2] Throughput rates will increase.

·       Lead times through the system will be reduced.

·       Work-in-process inventory will go down.

·       On-time delivery will improve.

If you consider these results to be nonsense, then think of the possible consequences if you return to the “high efficiency” metric:

·       [PJ3] Throughput rates will decrease.

·       Lead times through the system will become longer.

·       Work-in process inventory will go up.

·       On-time delivery will go down.

So think carefully when answering the question about which one of these methods is really the nonsensical approach.

Drum Buffer Rope Part 3

In my last post we began our discussion on managing variation especially as it applies to Drum Buffer Rope. In this post we will continue our discussion by continuing on with Goldratt's 5 Focusing Steps.  You will recall that Goldratt's first step tells us to identify our system constraint.

Once the systems constraint is identified, it must be subjected to the red carpet treatment.  Nothing in your system is more important than the constraint— nothing!  Once you have this information, you must decide how to best manage the constraint.  If the output from your entire system depends solely on the output of the constraint, then it certainly merits special considerations.  One of those considerations is to exploit the constraint, which is Step 2 of the Five Focusing Steps.  Exploitation means that you evaluate the process to get the most out the constraint activity. 

Rarely is a constraint being utilized at, or near, the maximum that it can do.  The exploitation effort means looking for things that the constraints can stop doing.  This could be an excellent opportunity to employ the Interference Diagram (ID) to define the interferences that stop you from getting more from your constraint. You may want to implement Lean concepts to reduce waste or Six Sigma to better control variation and quality.  It might also mean taking actions as simple as keeping the machine, or process, busy during break time and lunch time, or perhaps implementing a second shift or a third shift, or even off-loading work to non-constrained processes or resources.

Exploitation does not mean buying a new machine or adding more resources, at least not yet.  It simply means finding ways to get more out of the current process than you are currently getting.  There is a very high probability that during the exploitation exercise, the constraint capacity could be improved above and beyond the capacity of the next constraint in the system.  If such is the case, then go back to Step 1 and redefine the constraint.  In a normal improvement effort, this repeating cycle between Steps 1 and 2 might be completed many times before the system is stabilized.  When the system becomes stable, then go to Step 3 in the Five Focusing Steps and ratify the subordination rule to synchronize the product flow. The end result is to stabilize and synchronize the system, and then focus on the constraint.  Let the non-constraints work as required to produce sufficient quantities to keep the constraint busy.

The second consideration is to make sure the constraints are busy all the time.  Never let the constraint run out of work to do.  If the constraint stops or slows down, then the entire system will stop or slow down.    The best way to accomplish this objective is to make sure there is always work in the queue in front of the constraint.  In other words, create a buffer of work in front of the constraint.  The entire system output has total dependency on the constraint output, and constraint output is directly proportional to system output.  Think in terms of the right amount of work, in the right location and at the right time.

The system constraint not only determines the amount of throughput you can achieve, but it also determines the correct amount of work-in-process (WIP) inventory that should be maintained in the system.  The correct inventory level will be reached almost by default when system subordination is actively pursued and implemented.

The rope is actually a mechanism that controls two different functions.  First, it is the mechanism that determines how much and when to release inventory into the system.  The most common practice is to tie an artificial “rope” from the constraint operation back to the front of the line.  When the constraint produces and completes one unit of work and passes it to the next operation, then the rope is pulled to signal the front end of the line to release one more unit of work into the system. Rope signaling systems can vary.  The rope signal is equal to the output of the constraint operation, no more and no less.  This release mechanism, tied to the drumbeat of the constraint, will allow for a synchronized work flow and a smooth transition of work through the system. 

The second function of the rope is to initiate and maintain subordination for all other processes in the line.  By default, following the cadence of the rope release signal causes subordination to the remaining non-constraint processes to be executed.  The non-constraints processes can only work on what has been released to work on.  By releasing work only to the drumbeat, all other operations are held in check to the rule of subordination.  Even if the non-constraints can do more work, they are restricted by subordination and only allowed to work on parts or products required by the constraint.

Drum Buffer Rope Part 2

In Part 2 of my blog series on Drum Buffer Rope, we will dive into variation.  That is, how variation impacts your system and how DBR manages it.

Variation

For years, if not decades, people and organizations have dedicated considerable time and effort to remove variation from systems.  The utopian goal is to remove as much variation as possible from the system.  No matter how much planning is employed, no matter how much effort is extended, the bottom line is that variation will still exist. If you were asked how long it takes you to get to work every day, your response might be something like, “about thirty minutes.”  The instant you answer with the word “about,” you have introduced variation into the system.  You  know that historically speaking, some days you get to work in twenty-five minutes and yet others days it can take thirty-five or forty minutes.  In your “get to work” system, things can happen that will either speed up the process or slow it down.

Variation exists in everything, especially within a system. You understand that some processes will produce at a faster or slower rate than others, and this is the premise behind variation. Because of variation, the output from a system will not be linear,but rather it will operate within a range that changes.  This variable range is known as statistical fluctuation, and it exists in every system.  It’s important to understand that you cannot make variation go away.  The theory and practice of Six Sigma has pioneered the race to reduce variation.  But even with the most valiant efforts of time and money, not all variation can be removed.  You can reduce the amount and severity of variation, but it will still exist.  Once you understand that variation is a constant variable in any system, it’s easier to understand that at some point you will reach the minimum variation that is controllable in the system, and any efforts to reduce variation beyond that point are typically fruitless.  Perhaps, instead of spending so much time and effort on techniques to remove variation, the focus should really be on techniques to manage variation.    

Managing Variation with Drum Buffer Rope (DBR)

When viewing a system through the eyes of DBR, it becomes quickly apparent that improving every step in the process is not required, nor will the sum total of all of those discrete system improvements equal an improved overall system.  When conducting a full systems analysis, with the intent of implementing DBR, an important consideration to know and understand is the location of the system constraint, or the slowest operation.  In Goldratt’s Five Focusing Steps, this is Step 1—Find the constraint.  Once you know where the slowest operation resides, you now have the information necessary to know where to focus your attention within the system.  Why is it important to understand where the slowest operation is?  Because this is the location that controls and determines the output for your entire system.  In essence, the entire system will produce no faster than the slowest operation can produce.  (The system can produce less, but it won’t produce more.)

With the constraining operation identified, you have collectively quarantined the “drum beat" for your system.  Knowing the drumbeat is of strategic importance to implement and gain any system improvements.  The drum provides you with the necessary information of knowing where to focus your improvement efforts.  Historically, many organizations can and do conduct many improvement projects on a yearly basis.  The mantra seems to be that every organization and every process should strive for improvement.

The thought is that each organization is improving at some level of frequency to make the whole system better.  However, the sum of many efforts does not always equal what is good for the whole. The problem with this type of thinking is it is a totally unfocused shotgun approach to solve the problem.  In effect, it presents an improve-ment policy that states: "If I select a wide enough range, then I should hit the target, or at least come close to the target."  When you take the shotgun approach you might hit everything a little bit, but miss the full impact required to make real change and improvements.  If your shotgun approach includes trying to improve non-constraints, and many do, then the system as a whole gains nothing!

The improvement of non-constraints in isolation of the entire system, without a comprehensive analysis, is just a way of dealing with symptoms and not the real issue (constraint).  Without the ability and the accurate information necessary to focus on the real issues, the disease goes merrily on.  Improvement of non-constraints is a noble gesture, but one that yields little, if any, real improvements.  Every process within a system does not need to be improved at the same time!  Some system processes are more important than others.  Without knowing where your constraint resides, your efforts to improve will be unfocused and many times worthless, serving only to consume large amounts of money, resources, and time.

In my next post we will continue with our discussion of managing variation especially as it applies to Drum Buffer Rope.