Monday, March 6, 2017

Deterministic planning and safety stock

Last week I visited three manufacturing companies. One from the medical device industry, the other one producing fresh and frozen foods and the third in the business of musical instruments and electronic components. they are of very different size, one being a huge, global manufacturer, present almost everywhere in the world, the other one still in the billion dollar revenue range. the third is a mid-size corporation that is privately owned.

Even though they seem so different, the one thing they all have in common is that they're planning deterministically. I am not sure, but I seriously doubt that they a) are fully aware that they plan the demand only and b) that they comprehend all the unintended consequences of their actions and behavior in supply and demand planning.



First, let's discuss what it means to plan deterministically. My definition is, that you plan for the demand that you know (this could be an actual demand or a forecast that you desperately try to get right) and react to every change that comes along until fulfillment of the actual customer order. That is a heavily noise-laden planning process, especially if you propagate the changes throughout the Bill of Material or to feeder plants and distribution centers in a manufacturing and distribution network. As we all know, the 'bullwhip effect' does its damage on top of it and increases not only variability but also noise throughout the supply chain.

I strongly advice against practices like this and almost always recommend to switch to buffering practices where the supply chain is de-coupled and through use of the three buffers inventory, capacity and time, the process absorbs variability and takes the noise out of the planning procedures. There is an entire methodology behind the design and implementation of a buffering planning process to replace the deterministic process, but any more detail would exhaust the purpose of this post. (I simply want to raise awareness about what's potentially happening in your environment too)

Now most planners would argue that they use safety stocks to buffer variability. But - at least in the case of the three customers I worked with last week - the safety stock that was used was static and did not help at all. If you are using SAP software to run your supply chain and you are using the field safety stock in the MRP2 screen you are doomed by the static behavior of such a static safety stock which does absolutely NOT behave like a buffer.

Assume, as an example, that you are holding a safety stock of 100 pieces and you have a demand of 200 pieces due in 2 months from now. The MRP run will generate a purchase requisition of 200 pieces to add to the 100 pieces in safety stock, so that just before the due date you have 300 pieces in inventory (after delivery of 200 you would still have the safety stock of 100). After you converted the requisition into a PO and sent it to your supplier, the customer increases the order from 200 to 250 pieces. As safety stock (at least in SAP) is taken out of the net requirements calculation, the MRP run will generate a new requisition for 50 additional pieces. This way you'll end up with an inventory of 350 pieces even though you need only 250. The safety stock would have covered the additional demand and that is what it is actually meant to do: to buffer variability... but it didn't and your safety stock ends up in unused dead stock. this situation gets only worse if you have minimu lot sizes that you need to order.

You might say: "there is the possibility to customized the safety stock so a portion of it is used as a buffer". Correct, I say, but who does that? the three companies from last week didn't... why? because they were not aware what actually is happening.

Take a look in your own organization. I can tell you it will be worth your while if you use a coverage profile with a dynamic safety stock instead (make sure the minimum safety target is always set to 1 day). This is what we call "picking up low hanging fruit"