3. Increase the frequency of replenishment.

4. Manage the flow of inventories using buffers and buffer penetration.

5. Use Dynamic Buffer Management (DBM).

6. Set manufacturing priorities according to urgency in the PWH stock buffers.

Each step is discussed in the following sections.

Aggregate Stock at the Highest Level in the Supply Chain: The Plant/Central Warehouse (PWH/CWH)

The first step of the proposed TOC solution is to keep larger buffer stocks at the divergent point-where the stocks can be used to serve many different destinations-and use a pull replenishment mechanism triggered by sales at the end of the chain-the consumption point. This method guarantees we keep the lowest stock level possible to support the demand (what, when, where) of the various consumption points (the shops).

In order to have the product available at different locations, it is recommended to aggregate the inventory at the supplying source and, when necessary, build a PWH/CWH for that purpose. When the organization is a manufacturer, the entity is called plant warehouse (PWH) as this is the finished goods warehouse9 of the plant. When the organization is a distributor, the entity is called a central warehouse (CWH) and is the distribution hub.

We keep most of the stock (see Fig. 11-3) at the PWH/CWH by setting the buffer stock size high. According to the principles of statistics, this aggregation of inventory guarantees a more stable and responsive system than a system of keeping large inventories at the different consumption points (shops). In the TOC solution, the amount of stock (buffer stock size) at the consumption point is very low for each SKU. When a given consumption point sells a unit, the consumed unit will be replenished as soon as possible from the PWH/CWH.

When the transportation time from the PWH/CWH to the consumption points is relatively long, a regional warehouse (RWH) might be needed between the PWH/CWH and the consumption points to reduce lead times. This is the case in most global supply chains and large companies where customer responsiveness is crucial to sales. An RWH pulls inventory from the PWH/CWH and ships it to the consumption points it is serving. This is just an extension of the TOC solution and all other assumptions and considerations remain the same; the idea is still to pull from the PWH/CWH based only on consumption from the consumer.

FIGURE 11-3 The push versus pull distribution supply chain model.

Determine Stock Buffer Sizes for All Chain Locations Based on Demand, Supply, and Replenishment Lead Time

The stock buffer size is the maximum amount or quantity of inventory of an item held at a location in the supply chain to protect Throughput (T). The stock buffer size (limit) is dependent upon two different factors: 1. Demand rate-demand is the need for an item while the demand rate represents the amount demanded per time period (day, week, month, etc.).

2. Supply responsiveness10-how quickly the consumed units can be replenished. The main factor here is the TOC replenishment (lead) time (RLT,), which is defined in the TOCICO Dictionary (Sullivan et al., 2007, 41) as "(t)he time it takes from when a product is sold until a replacement is available at the point of sale/use." ( TOCICO 2007, used by permission, all rights reserved.).

A significant difference exists in the definitions of RLT in TOC and traditional push systems, and this difference and its impact should be noted before proceeding. In the APICS Dictionary (Blackstone, 2008, 117), the traditional definition of replenishment lead time (RLT) is "(t) he total period of time that elapses from the moment it is determined that a product should be reordered until the product is back on the shelf available for use." APICS 2008, used by permission, all rights reserved. TOC Defines this period to start from the moment the unit is consumed and not from the moment it is determined to be reordered. Looking closely at these two definitions, you notice that TOC replenishes when an item is sold or consumed versus traditional push replenishes when the quantity remaining in inventory is reduced to the reorder point (either the reorder point in the economic order quantity model or the minimum level in the min-max inventory system). This difference is significant!11 Similar to the traditional RLT, the TOC RLT is comprised of three components: 1. Order lead time-the time it takes from the moment a unit is consumed until an order is issued to replenish it.

2. Production lead time-the time it takes the manufacturer/supplier from the moment he issues the order until he finishes producing it and puts it in inventory or ships it.

3. Transportation lead time-the time it takes to actually ship the finished product from the supplying point to the stocking location.

For example, take a regular person in his private life managing his refrigerator content. Once a week (on Monday morning), he calls the grocery store to send him two bottles of milk and some vegetables. The grocery store takes two hours to prepare the order and then another hour to send it. The order lead time in this example is one week (a whole week can pass before consumption and replenishment). The production lead time is two hours, and the transportation lead time is one hour.

Figure 11-4 depicts the traditional saw tooth diagram. The APICS Dictionary (Blackstone, 2008, 122) defines this as "(a) quantity-versus-time graphic representation of the order point/order quantity inventory system showing inventory being received and then used up and reordered." ( APICS 2008, used by permission, all rights reserved.) This diagram represents the reorder point/economic order quantity ROP/EOQ model and the similar min-max inventory model;12 these models are the standard inventory models taught in many schools for managing stock levels. Figure 11-4 also shows the replenishment lead time components (note that in the case of a distributor the production lead time is zero and therefore the PLT in the figure contains only the transportation lead time).

If the RLT can be reduced, then numerous desirable effects materialize: The amount of stock to cover demand during lead time can also be reduced.

The amount of safety stock (to cover for uncertainty) associated with this shorter lead time is reduced.

The forecast for new products is for a shorter time interval; hence, it is more accurate.13 The responsiveness to actual demand is increased.

FIGURE 11-4 A typical sawtooth diagram of ROP/EOQ or min-max at the retailer or central warehouse.

These benefits (or desirable effects) make it worthwhile for you to study RLT. Try to apply these general guidelines for each component of lead time in your own supply chain.

Order lead time-if possible, cut the order lead time to zero. For example, if you replenish each consumption point daily based on the previous day's consumption, then the maximum buffer size at the CWH for each SKU should be a few days' demand for downstream consumption points. Note that if you have an adequate CWH buffer, this becomes a reality. The only reason not to cut the order lead time to zero is if the Operating Expense (OE) goes up; this topic is discussed later. The magnitude of cutting the order lead time is demonstrated in Fig. 11-4, where it is evident that just by cutting it to (almost) zero more than half of the replenishment time is saved.

Production lead time-Simplified Drum-Buffer-Rope14 should be implemented and the priority of the manufactured parts should be tied to the actual buffer stock level at the PWH (this topic is discussed later). The PWH inventory buffer decouples production from distribution. S-DBR significantly reduces production lead time because of Buffer Management (BM) and the use of stock buffers on the shop floor to respond rapidly to the next demands not covered by finished stock in the PWH buffer.

Transportation lead time-try to look for faster alternatives for transportation; for example, reduce the shipping interval by using trains or ships daily instead of weekly or using air shipments for some parts. Finding closer suppliers for raw materials (RMs) or purchased parts is also a possibility in many cases. Usually, this is the part of the RLT that one can do the least about, so every possibility needs to be checked. A simple test should be conducted, comparing on the one hand the extra cost of operating by faster means of transportation and on the other hand the cost saved by keeping less inventory and the extra T generated by not having shortages. In some industries (such as the fashion industry), such a simple calculation shows it is very beneficial to gain T by using an expedited means of transportation. For example, Item A is a fashion item that is sold at a T of 80 percent from the selling price. Totally variable costs15 (TVCs) are 15 percent for RMs and 5 percent for transportation by sea. Shipping by sea has a lead time of three months. Shipping by air has a lead time of two weeks; air costs double that of sea shipment. Therefore, the T for air shipment is 75 percent. In the example, it is quite clear that shipping by air is preferable-losing a sale of one unit because of a shortage compensates for selling 15 items for a lower margin, not even counting higher inventory investment and carrying costs which are much higher when using sea shipment.

Increase the Frequency of Replenishment

When applying the TOC distribution/replenishment solution, some factors are relevant when determining the frequency of delivery. The traditional purchasing practice for managing within a supply chain encourages purchasing in large quantities. The main reasons are as follows: 1. Time and effort are required in listing all available inventories and issuing frequent orders even for a small quantity. Economies of scale exist in processing a large order versus several small ones for the buyer. However, the extra cost of managing small quantities is usually quite small and involves, at a maximum, hiring some low-salary staff to help. Sometimes a pick, pack, and ship area is required to respond to small-quantity orders.

2. Some items can only be shipped in bulk because of transportation issues: fragile items sometimes can be better protected if shipped in a whole container; small items are stacked in boxes; and for sea and semi-trailer shipments, the minimum transportation volume is by container, making it economically beneficial to fill the container. Economies of scale exist in shipping a large order versus several small ones. New packaging might be required for the shipped items; instead of shipping a case of 48 of the same item, a mixed case of 6 units of 8 different items might be more useful. Sometimes, it is possible to use half-size containers instead of the full-size containers, so these problems can be dealt with as well.

3. Frequently, a volume discount is offered to purchase a large quantity of the same item. Additionally, a discount is given to a purchase order above a certain dollar amount. Free handling and transportation might also be given as an incentive for a larger purchase. Economies of scale exist in processing a large order versus several small ones for the seller. These discounts might be negotiated to be offered for large dollar quantities ordered over a year's period. In this way, one can order frequently and still enjoy the discount. Based on Cost World thinking (a focus on saving money everywhere), the additional shipping cost one might incur in increasing the frequency of shipments is seen as a big deterrent by most supply chain links. However, this cost is dwarfed by the increased T.

TOC takes a very different perspective, that of Throughput World thinking (a focus on making money now and in the future), in determining the direction and frequency of replenishment. It focuses on the additional T and return on inventory investment. There is a tradeoff between the additional cost one might invest in raising the frequency of shipments and the cost of having lower availability-by making the frequency of delivery higher, a better availability is created whereas the cost of shipments increases. By making the frequency lower, one will have to pay with either lower availability or with much higher inventory levels kept at the consumption points in order to cover for variations in demand. Note, in many cases the frequent transportation will not cost more than the current large-batch transportation. While transportation costs may go up, inventory investment decreases significantly. This frees up cash that can be used to purchase product variety from the same supplier. For example, instead of having large quantities of four products from one supplier, one can invest in having smaller quantities of 10 products from the same vendor. These different products each represent opportunities for sales. In the traditional approach, the shop has four opportunities to sell to a customer; in the TOC approach, the shop has 10 opportunities to sell to the same customer. In most cases, the additional revenue produced will dwarf the extra cost. Using Throughput Accounting (TA) (classifying accounting numbers into T, Inventory [I], and OEs) to get estimates of the impact of increasing replenishment frequency using mixed orders (replenishing all stock buffers with each order) is an easy calculation and a profitable exercise.

For example, a manufacturer owns a fleet of cars to distribute its goods. He currently makes weekly shipments to the end points. In this example, moving to a frequency of once per day instead of once per week will generate the following: Increase in shipment costs-since he owns the cars, only the TVCs are added, meaning the cost of fuel and maybe hiring more drivers to fill the shifts.

Decrease in inventory costs-instead of having a weeks' worth of inventory covering for extreme cases, he will move to keeping a daily amount of inventory. Inventory costs are effectively down by 80 percent, while the chance of running out of stock is decreased.

Manage the Flow of Inventories Using Buffers and Buffer Penetration

The TOC logic is to define the required safety and constantly monitor how the safety is being used. This safety is called a buffer. In a distribution environment, the quantity of an SKU we would like to keep at the stock locations (including the PWH/CWH and RWHs) is defined as stock buffer size. The buffer size or limit for this SKU depends on the three questions of what, where, and when to ensure high availability to support T and low inventory investment with low associated OE. For example, if the stock buffer size is 100 units for a given SKU and 40 units are currently on hand, then we expect that 60 units are on order or need to be placed on order to the supplying location. If those 60 units are not on order or on the way, a replenishment order of 60 units should be issued immediately. Note that each SKU represents an item at a location; therefore, each SKU stock buffer size may be and probably is different.

Buffer penetration is defined as the number of missing units from the buffer divided by the stock buffer size expressed as a percentage. The number of units missing from the buffer is the stock buffer size minus what is on hand and already ordered. For the previous example, the buffer penetration for the stock at this site is 60 percent ((100 40)/100). The buffer size is divided into three equal zones.16 The buffer penetration sets the color of the buffer according to the different zones: Less than 33 percent buffer penetration: Green Between 33 and 67 percent buffer penetration: Yellow Between 67 and 100 percent buffer penetration: Red 100 percent buffer penetration (being stocked out): Black The buffer penetration color gives an indication of the urgency of replenishing this stock.17 Green-the inventory at the consumption point is high, providing more than enough protection for now. Action required: ORDER a replenishment amount (in case of replenishing from a plant, prioritize depending on whether there is enough capacity to produce this order versus more urgent orders).

Yellow-the inventory at the consumption point is adequate. There is a need to order more units from the upstream supply chain. Action required: ORDER the replenishment amount (in the case of replenishing from a plant, order even if lacking in capacity as otherwise it might be too late. The capacity problem will be dealt with on the floor if it exists).

Red-the inventory at the consumption point is at risk of stocking out. Units in transport/manufacturing (depending on the entity that is in charge of replenishing that stock) should be considered for expediting efforts and an urgent replenishment order must be put to the supplying source if nothing is available on the way to the consumption point. Action required: INVESTIGATE, ORDER, AND POSSIBLY EXPEDITE.

Black-the stock has run out at the consumption point; every hour that passes at this stage means (potential) lost sales opportunities. This situation must be resolved as soon as possible because it represents real damage, especially at the most downstream links in the supply chain (for upstream links, it means the ability to respond to replenishment and to buffer changes is diminished). Action required: EXPEDITE AND ORDER IMMEDIATELY.

Figure 11-5 illustrates how the buffers are placed and how the region colors are used for prioritization. It shows the modeled network of a pull distribution system shown in Fig. 11-3. The same item has different buffers, one at each location. These buffers are managed separately. The buffer in the PWH/CWH is in the size of 600 units and is currently in a buffer penetration of 20 percent (it has 480 units out of the total 600). Therefore, the priority color of this buffer is green. Likewise, in shop 1, for example, this item has a buffer of 60, out of which there are currently only 24, making the buffer penetration for this buffer 60 percent and the priority color yellow. This is how the buffers are placed and how their replenishment is prioritized at the upstream link. However, this priority is not enough, as the same buffer can have some stock in the location and some on the way.

FIGURE 11-5 Item stock buffer sizes (limits) and buffer penetrations across the pull supply chain.

Several views of the same buffer are possible and of value. Inherent Simplicity18 developed the concept of the Virtual Buffer Penetration (VBP), which defines the priority at any stock point as the status of the stock at the downstream links in the supply chain. This concept is valid only until the next stocking point, meaning that the VBP for an SKU in the PWH/CWH will take into account only the physical stock at the PWH/CWH, while the VBP for a shipment will take into account the stock at previous shipments and the physical stock at the target. Figure 11-6 demonstrates this concept for managing across the supply chain.

In Fig. 11-6, the retailer stock buffer size for the SKU is 100 with 25 units currently available and a shipment of 25 units on the way from the PWH/CWH to the shop. The Virtual Buffer figures appear on top of each stock on the way to the retailer. The VBP takes into account the aggregated stock of in-transit and downstream stocking points. The SKU priority is determined by the Virtual Buffer Penetration of the next downstream stock location (shown above it in Fig. 11-6). The VBP provides a very powerful tool-full visibility across the supply chain, coupled with a clear and simple priority mechanism for the various stock point decision makers involved in the supply chain. The translation of the current information for various supply chain links for this example is: FIGURE 11-6 Virtual buffer concept applied to a shop item and in-transit shipments to this shop.

The warehouse manager at the stock location (shop manager in Fig. 11-6) can see clearly that the priority of this SKU is red at 75 percent buffer penetration. The buffer size is 100 units and 25 units reside at the shop, meaning 75 are missing. The shop needs to find out how to get more stock for this SKU as soon as possible.

The transportation manager can get the priority of the shipments, for example, what shipments need to be expedited. In this case, the shipment of 25 units of this SKU needs to be expedited based on a 75 percent buffer penetration (this is the same VBP as the plant warehouse manager sees). The virtual buffer for the PWH/CWH manager is computed as the shop buffer plus the transportation shipments. If the virtual buffer status was red, then the transportation manager should investigate to determine when the order will arrive at the shop. If there is some delay, he should expedite it.

The PWH/CWH manager can get the replenishment priority of this SKU. This virtual buffer takes into account all stocks on the way and at the shop for this SKU. In this case, he needs to replenish 50 percent of the buffer size of this item in the PWH/CWH (50 units) and the priority of this replenishment shipment is yellow based on a 50 percent buffer penetration (the buffer size is 100 units while 25 reside on the site and 25 are on their way to the site, meaning 50 are missing).

Use Dynamic Buffer Management

TOC aims at very simple, straightforward methods so that understanding and use come easily. The concepts of stock buffer size, buffer sizing, and buffer penetration replace the need for an understanding and use of sophisticated forecasting techniques. Variations exist in reality. Therefore, Dr. Goldratt provided a mechanism to manage buffers in a dynamic environment, thus eliminating the need for these complex forecasting models. The TOC logic dynamically measures the actual usage of the stocks and readjusts the stock buffer sizes (maximum target for replenishment) accordingly. This method is referred to in TOC literature as Dynamic Buffer Management (DBM).

By monitoring the SKU buffer penetration (i.e., item at each stock location for each product), we can identify whether the buffer size that we set for this SKU is about right. The essence of the idea is to monitor the combined impact of both the supply flowing in and the demand flowing out of the stocking point, where forecasting looks just on the demand side. The DBM approach argues that by monitoring and adjusting the buffer sizes, one can easily come to the "real" stock buffer level one needs to keep at the site in order to cover for the demand, taking into consideration the supply side (how fast one can deliver to the stock location).

The DBM mechanism is designed to alert the manager with two different warnings-one is when the buffer size is too large and the other is when the buffer size is too small.

When trying to measure whether the buffer size is too high, the indication is when actual stock of the relevant SKU compared to the target is too high for too long (e.g., staying in the green region for three consecutive replenishment periods). In other words, the stock buffer limit for that SKU should be adjusted downward, when the buffer penetration of the SKU has remained in the green zone for too long. This condition is designated as Too Much Green (TMG). This means that the stock buffer level is set too high for current demand. Remaining in the green zone for too long19 can be caused by the following: The demand rate has decreased (demand has gone down).

The supply responsiveness has increased (the supply side has improved).

The initial buffer size was too high.

Demand fluctuates severely and is currently low. This is usually quite a rare statistical fluctuation. In these cases, accepting the recommendation for a stock buffer limit decrease will not reflect the reality, and therefore, soon the DBM algorithm will suggest increasing the buffer again. This condition can be caused by a downstream link offering volume discounts on a specific item or downstream links using traditional ordering models (ROP/EOQ and min-max).

The default recommendation for remaining in the green zone too long is to decrease the buffer size. The basic guideline is to decrease the buffer size by 33 percent, but this depends on several factors: The speed desired to lower inventories.

The risk/importance placed on this SKU.

The risk/importance of this stock location.

A very similar mechanism is used for determining whether the buffer size is set too low. Determine whether this SKU inventory, after replenishment, stays in the red zone. This condition is called Too Much Red (TMR). In other words, based on the stock buffer size the actual stock amount remains after sequential replenishments in the red zone. These algorithm parameters are different from TMG, as here the risk we are trying to avoid is a stockout, while in the TMG we are trying to avoid overstocking. The basic algorithm for the TMR condition is to determine whether an SKU is in the red for several days (usually using the replenishment time as the parameter of the number of days). The more advanced algorithms also take into consideration how deep into the red the inventory at the site dropped.

The reasons for being in the TMR condition are: The demand rate has increased (demand has gone up).

The supply rate has decreased (the supply side has deteriorated).

The initial buffer size was too low.

Demand fluctuates severely.

The guideline for relieving the TMR condition is to increase the buffer level by 33 percent. Both the definition of too long in a zone and the definitions of how much to decrease or increase the stock buffer level for each SKU are dependent on location, item, etc., and may differ across SKUs. These parameters are just good rules of thumb to establish the system.

After adjusting the buffer, the SKU needs to go through a "cooling period" in which no buffer changes are suggested until the system adjusts to the revised buffer size. This cooling period should be long enough to let the adjustment take place (the new quantities ordered should arrive at the stock location), yet short enough so that a sudden real change in the market demand will not occur without someone noticing. For the TMR, the cooling period is usually a full replenishment time, and for the TMG, the cooling period is usually letting the inventory at the location cross-over to the green from above (since lowering the buffer size probably caused the current inventory at the site to be above the buffer size level).

Set Manufacturing Priorities According to Urgency in the PWH Stock Buffers

Many manufacturers make products to customer order. This means that each work order on the shop floor is for a specific customer with a given due date. For that environment, TOC prioritizes the production orders based on their due dates (for more details, please refer to Chapter 9, which covers the make-to-order environment).

When manufacturers embrace the TOC replenishment/distribution solution, another source of demand has to be dealt with-consumption from the PWH back through the manufacturing process. For these PWH orders, the right priority for manufacturing should be set (not according to time) based on the priority of the SKU. (Recall SKU means location and we know what, where, and when given by the buffer status.) The best priority mechanism is to take the buffer penetration for the item at the PWH location (the VBP representing the physical stock at the PWH versus the buffer stock limit) as the priority for the replenishment manufacturing order, since the stock status at the PWH reflects the consumption from all downstream locations, and thus the total status of this item in the supply chain, eliminating the need for forecast. If there is more than one production order for the same SKU, the best priority mechanism is to use the VBP as illustrated in Fig. 11-7.

As shown in Fig. 11-7, every production order looks at the VBP of the previous production order in production (the one that was released before it) to get its current manufacturing priority.

In this example, we see that stock in PWH for item A is 25 units versus a buffer size of 100 units. VBP is 75 percent and in the red zone. In the plant, WO1 is for 25 units, bringing the buffer to 50 units and to the middle of the yellow zone with 50 percent VBP. WO2 is for 40 units, bringing the buffer to 90 units and to almost the top of the green zone with 10 percent VBP.

FIGURE 11-7 Virtual buffer concept applied to prioritizing work orders (WO).

This penetration measure shows that manufacturing is synchronized to the actual usage of the stock. If the stock is depleted fast, the manufacturing order will be expedited through manufacturing. Otherwise, it will follow its normal processing sequence. Using this VBP concept provides a holistic system measure that fully aligns and synchronizes the chain links with the goal of the system-to be responsive to the actual consumption of stocks by the consumer across the chain.

Why Does a Pull Supply Chain Work Better?