In general, most factories today cannot react quickly enough to cost effectively do "real-time" scheduling, reacting to every production anomaly. The changeover cost and the "shock to the system" of frequent changes have ripple effects throughout the supply chain and within the plant.
if true real time is used as feedback into schedule execution, this will create a lot of nervousness in the loop. The issue that needs to be understood is at what point must deviations in schedule execution be reacted upon. This of course depends on the rate of production in most cases.
To many in the scheduling function, if a deviation does not effect a delivery schedule then it should be ignored. There are some companies that use Task time as the driver to a schedule being on time. If within the remaining time of production, lost time (as a result of production problems) can be recovered, there is no reason to act on the schedule deviations.
On another note, advanced planning systems use standard times to drive scheduling parameters. Tapping into the actual execution time of previous production runs within MES can help to ensure the standard times assigned to resources are accurate.
Every activity has its own thread of control. In other words, intelligent products get themselves produced on a network of equipment. They search for solutions and divide the effort between (trying to) adhering to the schedule/plan and looking for own solutions. The MES performs virtual execution (repeatedly) to check whether a plan is indeed feasible and to generate short-term predictions of the manufacturing activities.
The schedulers and planners generate plans to optimize their objective function within their scope (e.g. shifting bottle-neck optimization). The (intelligent) products do not necessarily share this. They are part of a flow in a supply network that has its own desires and needs. Schedule execution is more than a two-stage optimization. It is also a confrontation of concerns that coincide partially but conflict as well.