Hello, I have a problem which is fairly simple and I have not been able to find a way to solve it. We only have IBP SOP, no other modules, so we work with the infinite heuristic.
In our supply network, we have a 'distribution center' that sends products to +100 'warehouses' (both are modelled as locations). Each warehouse stores 8 products. We use Dyinamic Periods of Supply with Safety Stock for both the DC and the warehouses (i.e. LOTSIZEPOLICY=5 ) because:
- we need the algorithm to schedule shipments every X weeks to each 'warehouse' instead of everyweek (for this we use SUBPERIODSOFSUPPLY)
- we need the algorithm to keep a safetystock for each location (for this we use SUBPERIODSOFSUPPLYSAFETYSTOCK).
However, this results in the following problem. Since each Warehouse can start with any level of inventory (at INITIALINVENTORY), this can lead to a plan which looks like this: the warehouse receives Product 1 on Week1, product 2 on Week2 and product 3 on Week3. This means that at the DC will ship product 1 on Week1 (assume lead time =0 for simplicity), product 2 on week2 and so on. In addition, this pattern will repeat through the entire planning horizon.
This is highly inefficient.
So far we have analyzed these alternatives:
- Defining a KF that aggregates the TRANSPORTDS keyfigure for all products from the DC to the warehouse for X periods. Where X is equal to SUBPERIODSOFSUPPLY. However, this can result in a plan that tries to ship products that have not yet arrived to the DC. For example, sending Products 1, 2 and 3 on Week 1 may not be possible, because the planning algorithm planned product 3 to arrive to the DC on week 2. On the other hand, sending products 1, 2 and 3 on week 3 may cause a stock-out of product 1 at the warehouse on week2.
- Creating an inbound calendar. We could define an inbound calendar that only lets a warehouse receive products every X periods, where X is equal to SUBPERIODSOFSUPPLY. Then we could have another warehouse to be able to receive products the following week and so on. Warehouses would be grouped in these cycling weeks in such a way that every week, the number of products shipped (or the number of shipments) are reasonablly stable. This way, the shipping capacity of the DC would not be overwhelmed. Yet, this would be particularly difficult, because the lead time and SUBPERIODSOFSUPPLY is usually different. So this groupping would have to account for the different lead times. (An 'outbound calendar' would be ideal, but it does not exist). To make matters worse, every month nearly 5 of these warehouses are shutdown while another 5 are opened, so the fine-tuning required for this to work would have to be reapeated every month.
As you may have noticed, these solutions have... much room for improvement.
I will be very grateful for all your suggestions and ideas!
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.