Recently, a client of ours was taking steps toward Lean RfS (Repetitive Flexible Supply) as we had suggested in our report to this client. (Read more about Lean RfS here and here.)

LEAN RFS Buffer Sizing example.jpg

The client was stuck trying to figure out a way to make the transition to Lean RfS when he saw high levels of demand volatility in some of the firm’s green stream SKUs. To be fair, this client has some other limiting factors in its supply chain and production:

      • Short product shelf-life (they produce refrigerated food products with relatively short times to expiration)
      • Some limitations around batch sizing not driven necessarily by policy, but physics


Nevertheless, we are finding a way to help them make the transition.


Given demand variability, which is driven in part by the mode in which this firm’s customers order, this client wanted to know how they could possibly move to the “flow” of RfS and not end up with huge overstocks on weeks when demand was well below normal.


The following is based on our response to this client’s inquiries:

------------------------------------------------------------------------------------------------

We are pleased that your are, in fact, taking steps toward Lean RfS, as we suggested in our report.


With the specific issue you are asking about, however, we think you might be leaping to a weekly RfS schedule with too much haste or, perhaps, not giving due consideration to all of the factors. Allow us to elaborate.


The first question we would ask is this: When you performed your Glenday sieve analysis, how many items were in your green stream? That is, what number of SKUs account for 50 percent of your volume activity? This tells us how many SKUs should be targeted—ultimately—for the potentially weekly cycle.


Most organizations do not attempt—and we do not necessarily recommend—leaping from today’s non-RfS mode of scheduling immediately to a weekly RfS schedule. Because the ability to slash changeover or set-up times comes with repetitive practice, it generally makes some sense to begin with, say, a monthly or bi-weekly RfS schedule first. Then, after some months of practice, and a POOGI (process of ongoing improvement) focused on improvements in changeover/set-up times, it may be practical to move to a weekly RfS schedule. This new schedule may be predicated on shorter runs of each product than may be practical under today’s scenario because changeover times have been dramatically or, at least, significantly reduced.


Another factor to you should take into consideration is the “tank” or buffer size you establish for each SKU in your green stream. The buffer size should be determined by product demand variability. The greater the demand variability (as a percent of average demand), the larger the buffer required to accommodate that variability. How have you gone about determining your proper buffer size? You should be able to use Excel, for example, to do a simulation of buffer sizes and production schedules for a SKU (using the last 12 months of actual demand) to see if buffer sizes and production schedules are achieving your desired end for RfS.


Establishing an initial buffer size that is likely to cover your demand (in light of the replenishment cycle under the new regime) may come from relatively simple analytics around average weekly demand and standard deviations. In the accompanying figure, for example, establishing an initial buffer size somewhere between the average demand an one standard deviation above the average probably makes sense. After the initial buffer size has been established, however, the buffers should be dynamically managed by rules (see below).


It is important to note also that there may be demand variability that drives other decisions relative to green stream SKUs. For example, let us say that you have established a weekly RfS schedule (at some time in your future), and this particular SKU you are considering is part of the green stream. You would still monitor buffer penetration and dynamically manage the buffer size. Much of the dynamic buffer management can be automated with relative simplicity.


Rules governing dynamic buffer management are typically along these lines: if a buffer penetration is between one-third and two-thirds at each replenishment point, then no action need be taken. The buffer is properly sized to accommodate demand variability.


If, however, buffer penetration is found above two-thirds (67% or greater) at two or more consecutive RfS replenishment cycles, then you should consider increasing the buffer size by one-third. Similarly, if the buffer penetration is in the top one-third of the planned buffer (less than 33% or overstocked) on three or more consecutive RfS cycles, then we suggest that you reduce the buffer sized by one-third.


As part of dynamic buffer management, it is necessary to also institute a “cooling-off” period after each change in the buffer size. For increases in the buffer size, allow a cooling-off period of at least one full replenishment cycle. For decreases in buffer size, do not start counting buffer penetrations for further adjustment until the buffer has fallen from above (overstock position) into the top one-third of the buffer during a replenishment cycle.


Whenever dynamic buffer management rules trigger a change in the buffer size, it is wise also to determine if the cause is excess variation in demand or unexpected variation in supply. The RfS plan should be changed if there is an apparent persistent risk of service disruptions (out-of-stocks) or distressed stock (obsolescence in overstocks) from the combination of the buffer size and the present R¦S plan.


Also, if demand variation (or a disruption by Murphy) indicates a severe potential out-of-stock condition in the near future, then you should consider running the SKU on the red line, in addition to its normal R¦S scheduled run on the green line. On the other hand, if your monitoring shows that you are likely to create an overstock condition of more than one-third of the present buffer size, then you should consider canceling the next RfS-scheduled run for that SKU.


Of course, known variations in demand (that is, policy-induced demand variances) should be accommodated in your RfS schedule and dynamic buffer management. For example, if you know in advance that demand is likely to leap 30 percent over a two-week period due to promotions being carried out by your sales channel, then adjustments to the buffer size and/or RfS schedule should be made at least one replenishment cycle prior to the known change in demand. Typically, these RfS schedule changes simply mean a longer run, or supplemental runs of green stream SKUs on the red line, in addition to the normal green stream RfS routine.


If you need further assistance in working through any of these matters, please do not hesitate to let me know. RfS and dynamic buffer management take some practice and it is a culture shift. These changes should be introduced with executive-level support, of course, but also by carefully and thoroughly training those who will be affected by the program so that they fully understand why these changes will make their life better in the long-run.


We find that the three leading constraints in almost every business we touch are these:

  1. Policies – written and unwritten
  2. Training – whether people are trained to “think,” or just to “do”
  3. Metrics – how people and work are measured and, especially, when metrics for one department lead to actions that conflict with the metrics in some other part of the organization

 

The good news about these top issues is that these cost nothing to change. There is virtually no capital investment required to eliminate bad policies or metrics or change the way employees are trained.


Please note that R¦S should also drive improvements in changeover and set-up times. Many times we have found that holding POOGI breakthrough events to innovate around how improvements can be made is well worth the time and effort. Toyota has taken some setups and changeovers that used to take as long as a full shift and, over a period of several years, reduced these changeover times to a matter of minutes (sometimes less than ten minutes). This, too, takes more than just a top-down demand to “get this done.” It takes a change in culture, so that the workers see the work as always reinventing itself. No incremental improvement is too small to be considered.


We’re here to help, if you need us. Thanks for contacting us on this question.

----------------------------------------------------------------------


We hope this information helped you, as well. If you have further questions or concerns regard this approach to production, inventory and supply chain management, please post your comments or questions below. You may also feel free to contact us directly.