Planning process

Replenishment requirements are calculated based on the configured network topology for an item. The following is a sample network topology that we use to describe various calculations involved in generating replenishment orders.

Auto Replenishment generates transfer requirements from spoke nodes to hub nodes (for example, regional DCs to the central DC), and it generates purchase requirements from hub nodes to suppliers (for example, central DC to suppliers). The following steps are involved in generating replenishment orders. These steps are repeated for each product and site combination that is in scope for replenishment planning. Requirements from downstream nodes are propagated upstream based on sourcing rules information, and the process repeats at the upstream node until it reaches the root node for that item.

• Demand processing – AWS Supply Chain prepares the historical demand or forecast data based on the replenishment plan configuration. Demand or forecasts are processed at the level of product, site, day, or week based on the replenishment plan configuration settings. Sales history or forecast data are aggregated at the product and site level if they are provided at a more detailed level, such as product, site, customer or product, site, channel. Similarly, day to week aggregation occurs if a replenishment plan is configured at the week level. In the preceding example, demand is taken from spoke nodes, which are regional DCs, and it is aggregated at the product, site, and day/week level. If consumption or demand based inventory policy is used, the last 30 days of demand (sales history) is used to calculate average consumption.

• Target inventory level – Use the demand or forecasts along with the configured inventory policy to determine target inventory level for a specific time period. Auto Replenishment supports two different replenishment models.

  • Forecast-driven replenishment

  • Consumption-based replenishment

  AWS Supply Chain generates inventory targets based on the forecast. These inventory targets are determined based on lead time and sourcing schedules to ensure inventory levels account for the variability in demand and supply lead times.

• Transfer or purchase requirements – AWS Supply Chain nets demand in each period from the supply (on-hand + on-order inventory) to project inventory into future time. AWS Supply Chain maintains the projected inventory levels at the same level as the target inventory level calculated in the previous step. The difference between projected inventory level and target inventory level is the net supply requirement or reorder quantity (RoQ). AWS Supply Chain applies minimum order quantity, or it orders multiples to generate the final transfer requirements or purchase requirement (POR). AWS Supply Chain uses the transfer or vendor lead time to determine the order by date. The default for lot size is 1.0, and the minimum order quantity is 0.

  Calculation logic

  
                      rounding=f(RoQ,MOQ,Lot_Size)
  
                      =Lot_Size×Max(RoQ,MOQ)
  
                  

  The preceding formula describes the rounding logic in Auto Replenishment. AWS Supply Chain first compares the reorder quantity RoQ and minimum order quantity MOQ, gets the final order proposal, and then multiplies by the lot size factor for the actual quantity. The lot size is configured in the sourcing rules entity with the field qty_multiple.

• Requirement propagation – For spoke nodes, AWS Supply Chain uses sourcing rules to look up parent nodes and propagate transfer requirements to the upstream node. AWS Supply Chain offsets the required delivery date by transfer lead time to determine the required date at the parent node. AWS Supply Chain only supports single sourcing. When this step is completed for all child or spoke nodes under a hub node, AWS Supply Chain repeats the previous steps on the hub node. This process is repeated until it reaches the root node in an item’s topology.

  Auto Replenishment only shows purchase order requests for vendor-facing sites. There are two kinds of vendor-facing sites:

  • Vendor-facing sites that supply other sites

  • Vendor-facing sites that don't supply other sites

  

  For vendor facing-sites that supply other sites, the reorder quantity is the reorder quantity from its child sites, plus the independent reorder quantity from its own demand. For vendor-facing sites that don't supply other sites, the reorder quantity is computed based on the demand forecast of the site. The independent reorder quantity for vendor-facing sites follows the same logic in the reorder quantity computation. The dependent demand is the summation of all the child sites. If the days of coverage is 7, the RoQ is the summation of the quantity of all orders in the covered period. The following example shows a scenario in the planning horizon where there is only one order for each site, and it explains the computation.