Multi-period planning models result in solutions which are feasible at the boundaries of the periods but may be infeasible within the periods. The composite algorithm presented here (i) solves coarse multi-period MILP model structure for production planning; (ii) sequences operations via a genetic algorithm to minimise switching; (iii) verifies schedule feasibility via agent-based simulation and local logical decision making; and (iv) if infeasible, re-partitions the time horizon into multi-periods and resolves from (i) until feasible. Application of the algorithm to gasoline blending illustrates its effectiveness in computing feasible plans and schedules for such systems. (c) 2012 Canadian Society for Chemical Engineering