Face milling is widely used in machining processes, aimed at providing workpieces with high surface quality. The chatter generated in face milling could lead to tremendous damage to machine tools, poor machined surface quality, and loss of processing efficiency. Most related researches have been focused on the modeling of spindle dynamics and discretization algorithms for chatter prediction. However, few published articles have taken the geometric characteristics of workpieces into consideration, especially for workpieces with discontinuous surfaces in face milling, which leads to poor accuracy of chatter prediction as well as the waste of processing efficiency. To overcome this shortage, a novel dynamic model for the face milling process is built in this paper, considering the cutting insert engagement based on the geometric characteristics of the workpieces and the tool path. The stability lobe diagrams (SLDs) applicable to workpieces with discontinuous surfaces are constructed. A process parameter optimization model is developed to maximize the chatter-free processing efficiency of the face milling process. The sensitivity analysis is utilized to simplify the objective function, and the genetic algorithm is employed to solve the optimization model. The proposed approach is validated by an experimental case study of an engine block, improving the chatter-free material removal rate by 53.3% in comparison to the classic approach.