A hybrid process structure, a continuous flow followed by a batch process, is used in the primary aluminum production. The continuous flow provides a steady stream of Work-In-Process (WIP) to the process; but the WIP is usually delayed before it can move to the batch process. This leads to increased flow times, increased inventories, and inefficient use of resources. In such hybrid structures, process synchronization can be achieved by carefully scheduling the production and allocating the resources. We study the performances of two alternative production scheduling policies and two alternative resource allocation mechanisms in the primary aluminum production process, using simulation. First, we build a simulation model that can simulate a nearly-realistic primary aluminum production under various design settings. Second, we show that the production scheduling policy led by the latter stage of the production significantly improves the process performance compared to a decentralized policy, in which each stage makes its own schedule. Third, we show that the amount of resources at the buffer zone in between the two stages of the process is very critical for the process performance, and we quantify its benefits through numerical work. Finally, we suggest that dedicating common resources, such as trucks, to individual stages of the process slightly improves the process performance as opposed to pooling those resources.