Business requirement:
- For historical periods, Projected Inventory should be equal to the Ending Inventory for that period.
- For current and future periods, Projected Inventory should be derived using the previous period's projected inventory plus shipment forecast minus POS forecast and other consumption adjustments.
Calculation Formula:
Projected Inventory (Current Period)=Previous Period Projected Inventory + Shipment Forecast - POS Forecast - Other consumption adjustments
Calculation must be performed sequentially across forecast periods.
Challenge:
- Calculation requires period-by-period propagation of inventory values, where each future period depends on inventory values derived from prior periods.
- Past records runtime value must update current and future records.
- Despite additional nodes and aggregation logic, the final projected inventory values do not match expected sequential projections.
#Manufacturing Cloud
@Shubham Joshi DPE won't walk that chain inside one run.
A formula node only sees fields already on the row, or fields you joined in. It doesn't read the projected inventory this same execution just calculated for the prior period. Extra aggregate nodes don't change that. They still aren't a running total.
Historical periods are the easy part. Set projected inventory equal to ending inventory and stop. Those rows don't depend on a prior calculation.
Current and future periods do. The only value a join can use as 'previous projected inventory' is one that's already stored from an earlier run. In a single execution, every future period still sees the old stored number, so the sequence never moves forward.
What works is one period per execution, in period order. Write the projected inventory, then run the next period so its join reads what you just wrote. A flow can loop the periods and call the DPE in that order.
If you try to calculate every future period in one definition, the numbers won't match the sequential formula, even when the nodes look right.
