Developments of information technology provide a foundation for implementing shop floor control in high variety production environments. A dynamic flexible assembly job shop is abstracted from a mold industry, we suggest the use of a hierarchical production control framework - dynamic flexible assembly job shop control (DFAJSC). DFAJSC consists of three sub-decisions, which are release decision, routing decision and sequencing decision, these decisions are made dynamically to react to disturbances (e.g. machine breakdown, rush order). As no study has applied controlled release in flexible assembly job shop, we would improve state-of-the-art controlled release methods for flexible assembly job shop. Building on this, we explore when and how routing decision is incorporated. Simulation experiments are designed to examine the performances of DFAJSC methods, coupling effects of our improved controlled release methods and routing decision rules are evaluated in different levels of disturbances. Results reveal that input/output control (IOC) release methods are overall best performed in all levels of disturbances, Dynamic Process Planning (DPP) decision is not significantly advantageous over Non-Linear Process Planning (NLPP) decision combined with IOC release methods. Thereof, IOC release integrated with NLPP is suggested for practical applicability.