The current industrial standard of production planning is to use MRP-II type of logic with higher level Master Production Schedule (MPS) and Material Requirements Planning (MRP). This system works well in many industries, but there are some restrictions in this logic. I have been working on some production planning and control project recently and found some challenges for the future systems:
(1) Engineer-to-Order - In some cases you need to promise a delivery date before you have any Bill of Materials. Some suggest using project management for these cases, but how about situation when the factory has tens of thousands "mini-projects" each including some engineering. The solution could be some type of generic-BOM or descriptive-BOM, which would then get more precise attributes during the design and engineering phase. Variant configurators and phantom products may give a partial solution, but there is still lot of work to do to fully support "white spots".
(2) Capacity management. Eli Goldratt introduced the Theory of Constraints (TOC) a long time ago and some scheduling principles for capacity management. Still most of shop floor control systems use traditional FIFO, SPT, LPT, Due dates type of scheduling and do not care too much about key resource capacity planning.
(3) From sales order to production orders. In the era of mass customization and built-to-order type of philosophies the sales orders should be booked to production automatically and real-time. This part of the system should not need any manual production planning work.
(4) Integration between factories is about smart syncronisation. The most challenging is the case of mid-volume manufacturing with high-mix products. The dependencies are not obvious and perceiving "a customer project" or "sales order" from several items coming from different factories/production lines is not easy. Anyhow, we expect to get everything at the same time and avoid too much merge-in-transit type of inventory.
Probably this list could be continued with several bullets, but the point is that we need to pay attention to assumptions of MRP-II logic. The current state-of-the-art ERP systems do not always support todays requirements. The industry gives good examples of the needs of future production planning systems. This is good input for the research community in logistics modelling and software engineers who are designing data models to support all the complex products and routings.