In real-life environment, 80% of business processes are dynamic whereby each process is dependent on individual conditions of execution and at the same time contains a large amount of parameters that makes them difficult to model. A self-Adaptive, agent-based simulation model for dynamic processes enables reduction of costs, resources and efforts in designing new models. This paper presents a workflow for modelling dynamic processes that consist of key parameters needed for the design and refinement of the simulation model, which are data collection and data analysis. Three dynamic processes are chosen as case studies; crime investigation, new student registration, and transportation requests processes. The workflow of each case study is analyzed using cross-case analysis, directed approach, and grounded theory. The findings showed similarity of key parameters shared by three dynamic processes and thus required to refine the self-Adaptive agent-based simulation model.