When the items are dedicated to one client for whom VMI is already in place, then there is no point in maintaining stock of these items at the plant warehouse. It could be the case that between the production and the actual shipment there is a practical need to store the items for a day or two. However, the real storage and the focus of BM should be on the client's site.

The replenishment time for VMI should include the transportation time. The characteristics of the transportation could be important because it is difficult to expedite a shipment once it has been sent. VMI is much more effective when the transportation time is much shorter than the manufacturing time. This seems to be true in the majority of cases.

Mixed (MTA and MTO) Environments

As it was already mentioned, not all items should be managed to availability, with the alternative of offering short delivery time to selected items (as MTO). It is clear that in many cases a mixed environment is advised.

There could be two different meanings to the term mixed environments of MTA and MTO. The first is that there are items that are strictly produced to availability and others that are strictly produced to order. The other possible meaning is that several items have demand both for immediate delivery and (usually for large quantities) for future delivery at specific dates.

The generic problem is how to manage an environment with two different types of buffers: time and stock. There are several sophisticated ways to do it, but we want the simple and effective way. Having the same items under two different management systems is too complicated to be truly optimal in reality. Our clear suggestion is to separate the item SKU identifier of the MTO from the item managed to MTA, even though for all practical purposes they are the same item. When the MTO orders for an item, usually managed as MTA, are treated as orders for a separate item, then there might be a case where the MTO order is expedited even though there is enough stock on-hand to deliver it. A similar case can happen when it seems we need to expedite the production orders for the MTA item, but an MTO order has been finished long before its due date and thus there is enough stock to cover the immediate demand. Our advice is to ignore those cases and simply expedite even though there may be another solution for it, or the production manager can take it on as his own responsibility to stop expediting and make the necessary shift between the two identities of the item (having two different SKUs for the same physical item).

This leaves us with managing MTA items to the side of MTO items. Before we state the solution, we recommend addressing the flawed perception of priority in reality. Put yourself in the shoes of an operator who has to choose between two orders: one is an MTO order, with a specific client willing to pay, the other is for stock, meaning we do not know when a client would buy it directly from stock. It seems that the MTO order has a clear priority because it means "Throughput now," versus "possible Throughput sometime in the future."

Suppose the MTO order is to be delivered in three weeks, while the MTA order is for a product that is currently short at the finished-goods warehouse. Would you now also prioritize the MTO order? If not, then what is the rule?

Another perspective is proposed. In MTA, the company is offering a commitment to provide availability. The same commitment is given to a client for the MTO order by specifying a date where the order would be completed. Thus, the issue of priorities means: what are the priorities to follow for the best chance to meet ALL of our commitments? BM yields the green-yellow-red priorities and the claim is that even though both have different sorts of buffers, the meaning of the green-yellow-red, and even the meaning of the buffer status, is exactly the same. When an operator is facing orders with various buffer statuses, he does not need to know which one is MTO and which one is MTA. The color code is the main priority mechanism, with the buffer status as additional, more detailed information.

There is one problem concerning the mixture of MTA and MTO: managing capacity. MTA is stricter in its requirements of maintaining protective capacity of the CCR because in MTO we have the flexibility of quoting longer lead times when necessary. When the amount of MTA relative to the MTO is relatively small, say approximately 15 percent MTA and 85 percent MTO, then reserving 15 percent of the capacity for MTA and basing the time quotation of MTO on 85 percent of the available capacity is an acceptable solution. For all other cases, we suggest taking the MTA capacity management as the overall rule. It means MTO would be handled based on time quotation of standard lead time, assuming it is always possible because there is enough capacity at hand. Having the capacity buffer in place (the ability to increase capacity easily and rapidly) is an excellent way to draw the maximum capacity from the internal resources and using the capacity buffer whenever needed. More on this approach can be found in Schragenheim, Dettmer, and Patterson (2009, Chapter 7).

Dealing with Seasonality

Seasonality poses obvious problems in managing stock in general. Chapter 11 in this Handbook deals at length with this issue, including the dilemma to choose between forecasting the demand and simply following DBM.15 In this chapter, we highlight the acute problem for the manufacturer. Managing capacity is a special problematic area in manufacturing, often for the wrong reasons, such as achieving high efficiency of every single resource. From the pragmatic viewpoint in TOC, managing capacity is still an issue-making sure there is enough capacity to meet the demand.

In most cases, when the term "seasonality" is used, the meaning is peak demand within a certain period of time. Such a peak of demand could take several months or just one or two days. There is a clear difference between these extremes; a very short peak means no replenishment could take place during the peak. This case will be dealt with as an MTS later in this chapter.

The capacity problem with seasonality16 is that within the peak the total demand might require more capacity than the CCR has. Such a situation would definitely reduce the on-hand stocks and all one could do is try to prioritize. For a short while, it could be good enough, but for a longer period of time, it would be disastrous.

Increasing the target levels before the start of the season is only a partial solution to the capacity problem. If there would be a real lack of capacity during the peak, and if the peak is not very short, then shortages will certainly occur.

Solving the capacity issue requires investment in capacity and materials before the start of the peak. The direction of the solution is to create enough inventories of several fast runners to cover the demand for those items throughout most of the whole peak. A valid way to do it is to forecast the minimum quantity to be sold through the whole peak of several fast runners and produce this amount prior to the peak. Not having to replenish those fast runners every time there is a sale would save precious capacity that will be used to replenish the other items. The reason we suggest doing this only for several of the top fast runners lies in the characteristics of fast runners to have less spread of the future sales. Even if some of the inventory is left after the peak, it will still be sold.

Problematic Environments for MTA

The replenishment solution for manufacturing is dependent on the priority mechanism in the execution phase. Not just on the priority itself, but even more on the capability to expedite. Environments with longer replenishment time and where expediting is either impossible or very difficult have to compensate for the lack of flexibility by maintaining more stock and by frequent replenishments. Even when the latter is possible, there is a real problem in achieving good availability.

Consider the case of sequence-dependent setups or even just very long setups coupled with a long list of items to be produced. For example, a production line of paint produces 12 different colors. Each of them has three to five different variations of the paint. In producing paint, the length of the setup (mainly cleaning the line from any residuals of the former paint) depends not only on the next color, but also on the previous one. When you keep the sequence that goes from light colors to the dark colors, the total setup time is much less than trying to produce according to the sequence of the real needs of the market. Hence, the production line has to stick to the preferred sequence and thus produce the whole cycle (certain slow items might be skipped from time to time). Suppose that the whole cycle takes 21 days (3 weeks). This means that the replenishment time is 21 days.17 What should the target level be? The "maximum" consumption is within 21 days, so the average sales within 32 days seems appropriate enough.

However, note that it means every item is replenished only once in 21 days. So, if the sales are higher than expected, toward the 20th day the on-hand stock might be penetrating deep into the red and there is nothing we can do about it. Deviating from the sequence might have too much impact on the capacity, maybe even turning the production line into a bottleneck.

The only remedy to this state is to hold much more stock. If the target level is twice the average of sales within the cycle time, then most of the time the on-hand stock will be in the green and toward the end of the cycle, it'll be in the yellow. This means the "too much green" parameter needs to be long enough to prevent a false recommendation to reduce the target.

Dealing with the more problematic environments should highlight how good the TOC solution is for most other environments.

MTS That Is Not MTA

There are certain cases where it makes perfect sense to use MTS, but this is not coupled with a commitment to maintain perfect availability. We can identify two categories of such cases: 1. The reasons for MTS come from capacity management and not for ensuring availability.

2. The organization is trying to provide a certain level of availability but cannot, or even does not wish to guarantee it.

Let's analyze the aspects of the two categories. The first one has already been demonstrated by the seasonality approach where sometimes high stocks, well above the target levels, are used on a few fast runners to free capacity during the season itself. Preparing for any peak of demand that requires capacity above capabilities forces the capacity planners to look for MTS even in MTO situations. Of course, if the MTO orders are all fully customized to the requirements of the clients, this is not possible but it could be that stocking some components would still relieve the pressure of capacity on the CCR.

The other category is typical of situations where the possible demand fluctuations are too high to provide excellent availability or the surplus inventory is very expensive to hold. In such situations, the marketing approach could be that we do not promise availability, so if you really want to buy, be fast! Examples of where availability cannot be guaranteed are: Launching new products, especially new innovative products.

Promotions where a peak demand is anticipated, but perfect availability should not necessarily be offered to the clients.

A short demand peak where replenishment within the peak is impossible.

Products with very short shelf life where the truly varying costs of producing them are high relative to the Throughput.

There is no point in committing to availability in such cases. Still, providing availability can be the main force behind the decision about how much to produce, but full commitment seems risky.

What should the process of managing MTS look like? Two distinct problems for MTS are not present in MTA. First, how do you decide how much to make to stock? The decision should be based on a forecast that recognizes both the reasonable minimum and maximum sales within the reliable supply time and being aware of the damage of shortages and the damage of surpluses.

Second, how do you prioritize a production order for stock and not for availability? Checking the state of the downstream inventory relative to the target level does not make sense. It seems to us that when the stock is intended for an anticipated peak of sales, setting a date for completion and treating the order as an MTO makes more sense. The date in such a case is not really artificial because the stock is targeting a certain peak demand event.

In the cases within the second category but where no special event is the trigger, the marketing approach of creating an atmosphere of being "hard-to-get" makes the replenishment technique still valid. However, the target levels are intentionally low and the expectations are to be in the red (or even in the black) most of the time. The expediting efforts might be much more restrained and most of the DBM recommendations are not going to be granted.

Implementation Issues

Some of the implementation issues have already been addressed within this chapter, especially determining the initial values of the target levels. Certainly, the process of deciding upon the initial values of the target levels should be as short as possible. There is no need to be precise-a very rough estimation is more than enough.

An important point is the initial calculation of the targets. One should assume the replenishment times to be much shorter than the current one. In many cases, cutting the current replenishment time by half is a good initial guess for setting the target levels.

Buy-in issues won't be discussed in this chapter.18 The two problematic areas discussed next are included due to the potential problems that might occur and not be fully understood.

Moving from MTS or MTO to MTA

The move toward MTA can be done from either MTO or MTS or a combination of the two. The move from MTS to MTA might be viewed as relatively easy because there should be stock in the system, both finished and in production, that is not already dedicated to customer orders. Most of the time the amount of available stock found in the system is larger than the target level. The problem comes when there is no stock that is not already dedicated to specific customer orders.

When the shop currently operates under strict MTO, then there is definitely no WIP or finished goods that are not already dedicated to customer orders. The practical action is that enough stock must be prepared. The preferred way to start the replenishment is with full buffers-the finished-goods stock buffer level is at the target. Only when the actual stock level allows for replenishment should the switch between MTO to MTA be performed. If that change were done too early, chaos would occur in the shop floor.

The problem is that building stock has to be done on top of continuing to supply to order according to the current rules and sales volumes. This means producing to stock (later it'd be for availability) on top of running the production order required for actual customer orders. The practical action is to use only excess capacity to build the required stocks. It would take time according to the available capacity. The best way to physically build finished-goods inventory buffer on top of the regular demand is to generate dummy MTO orders for the stock building using a large time buffer in finished goods so that it won't turn red too soon. The chosen date would give realistic expectations for when the availability marketing offer would be ready for launch.

Software Considerations

MTA places a lot of emphasis on the proper management of the shop floor. On top of the production side, at least two additional functions must be connected and performed well-marketing and sales (mentioned here as if they are the same function) and the purchasing of raw materials.

The marketing side needs to develop the offer to the clients, including setting the right expectations, so customers know about any limits to the quantity that can be purchased at one time. Sales needs to know when to try harder and when to find ways to restrict the demand based on the planned load.

The TOC buffers in both MTO and MTA assume perfect availability of raw materials. The generic rules of maintaining stock, explained in both this chapter and the next chapter on distribution, are applicable for managing the stock buffers of the raw materials.

It is the natural role of software to connect all the pieces to show one holistic picture. However, that natural role is seldom what one really gets. Most ERP software packages do not really display one cross-functional picture. The current situation with TOC is that while the general direction of looking at the performance of the whole organization and pointing to the weak links does exist, it has not yet been converted fully into software specifications.

Another role of software is to focus the attention of the decision makers on what really matters. It is not done currently in the vast majority of the software packages.

Software has a critical role in communicating and instituting processes and terminology. This does not mean that the current ERP packages institute the right ones. TOC challenges many of them. However, the power of the software cannot be ignored. This highlights the need for TOC software to institute the right processes and to provide the relevant information to the various levels of the decision makers.

As no current ERP package is built upon the TOC principles, and given the basic difficulty of implementing/revising such software, the practical options are either to force the TOC procedures within the existing ERP or any other information system or to link add-on software to the ERP.

Concentrating just on the MTA requirements, there are five different areas of need for TOC functionality within the information system (IT) of the organization.

1. Generating production orders based solely on replenishing to a defined target level.

2. Generating the green-yellow-red generic priorities for every single production order. The buffer status should be considered as a "bonus," which is nice to have but not a must.

3. Using DBM to recommend changes to the target levels.

4. Monitoring capacity through planned load, including being able to recommend what replenishment orders to release.

5. Providing managerial reports, including POOGI, and monitoring the number of red orders and historical behavior of the planned load.

The first area can be quite easily forced on the MRP/ERP. Still, the terminology of target levels or buffers will not be included unless a more massive development is done. It means the people handling MRP need to understand the TOC logic and the terminology well to keep the ERP updated. The real difficulty lies in forcing the MRP/ERP to determine the green-yellow-red priorities. All ERP packages assume every work order should have a date, but the TOC logic is quite different.

Dynamic buffer management is another module that cannot be easily supported within the ERP itself.

One might be able to create a variety of reports of the planned load within the ERP, but it is far from being straightforward.

All of these points are relevant for the development of the organization's approach to the software side. Every implementation should consider the options for software as an integral part of the implementation.

References

Arnold, J. R., Chapman, S. N., and Clive, L. M. 2008 Introduction to Materials Management, 6th ed., Prentice Hall.

Blackstone, J. H. 2008. APICS Dictionary. 12th ed. Alexandria, VA: APICS.

Goldratt, E. M. 2008. Strategy and Tactic tree. Consumer Goods Make-to-Stock (MTS) to Make-to-Availability (MTA) S&T, Level 5, September 2008.

Goldratt, E. M. 2009. "Standing on the Shoulders of Giants." The Manufacturer. June. http://www.themanufacturer.com/uk/content/9280/Standing_on_the_shoulders_of_giants (accessed February 4, 2010).

Goldratt, E. M. and Cox, J. 1984, 1993. The Goal. Croton-on-Hudson, NY: North River Press.

Goldratt, E. M. and Fox, R. E. 1986. The Race. Croton-on-Hudson, NY: North River Press.

Schragenheim, E., Dettmer, W. H., and Patterson, W. 2009. Supply Chain Management at Warp Speed., Boca Raton, FL: CRC Press.