Last week I got engaged with a client to work on a model, so they can perform effective materials planning. The client did great preparations in defining guidelines for policy setting in an ABC / XYZ matrix. This kind of thinking drives SAP software to support the achievement of targets and goals: Periodic policy setting based on changing situations is imperative, so that it automates ongoing balancing between service levels and optimized inventories.
One part of the solution was dealing with production scheduling. The question at hand: how do we level the production program MTS (X and Y items) and let MTO (Z items) orders flow in ?
The answer: use different policies to drive automation for different situations. Below you can see the grid we came up with.
The idea behind setting policies in the various classes is to determine an ideal mix of production quantities. This in turn drives great service levels and minimum inventory holdings, even though variability causes a deviation of actual order quantities and dates from the forecast…
For all X items we assume a high forecast accuracy (90 to 95%), and therefore a Make To Stock strategy is applied in order to manufacture finished product to a forecast into inventory. Because of the high degree of predictability, we can assume that any forecast error can be covered with safety stock. That is why we want to use strategy 10 which creates an LSF requirements type. LSF requirements do not consume itself against customer orders and therefore additional, actual demand must be covered by safety stock. This should be in order, since the forecast error is minimal for X items.
as for safety stocks our policies suggest to use a dynamic safety with a coverage profile. This is because the dynamic safety stock goes up and down with future, forecasted demand and would therefore adjust itself up and down for important A items. For less important B items we would use an automatically calculated static safety stock, whereas we would use a static safety stock manually for the least valuable C items.
For Y items there is slightly more variability and the forecast does not necessarily provide as much comfort as it does for X items. That is why our policy calls for strategy 40 and the associated requirements type VSF. A VSF is consumed by incoming customer orders and therefore – if actual customer orders exceed the forecast – additional demand is directly transferred into the production program. In this case we do not need a safety stock, since we satisfy demand over and above the forecast, directly from production.
A fixed lot size is used, so that the MRP run generates supply in fixed quantities, which a scheduling system can level, sequence and distribute within available capacity on the line. The production program for XA products is generated weekly wheras for XB and XC products the program is created bi-weekly and monthly respectively. This is due to the fact that A items need more attention and need to be checked more frequently, due to their nature of being of high consumption value.
Z items are set up to be Made to Order. However, there is a forecast placed on the finished good, so that it is possible to procure raw materials and reserve capacity on the lines. This is achieved using strategy 52, which generates a VSE requirements with statistical planned orders which reserve capacity and allow the transfer of requirements to raw materials. Z items do not have safety stocks and a production program is not necessary, since actual orders flow into the free capacity spots that were reserved.
In this sense it is possible to generate a level production plan considering all products running on the ‘mixed model lines’. Class A materials build the basis with a weekly, undisturbed schedule of fixed order quantities that build inventory to the forecast plus a safety stock. Class B materials are then stacked on top, using about 30% of the capacity (for an approximate total of A and B items of 80%). Orders exceeding the forecast will be placed into the bi-weekly production program and reduce the remaining available capacity for C products.
This kind of thinking is nothing new, but unfortunately it’s not applied very often. In many cases, people think that MRP should generate the production program directly. That is not how it works in an MRP type planning system. MRP will generate supply based on your basic data setup, but under no consideration of available capacity on the lines. MRP also does not create a sequence of orders and it does not fix the schedule. Because of that, you should introduce that extra step in between the MRP run and handing over a schedule to production: Sequencing, Leveling and Scheduling !