Realistic agent-based epidemic simulations usually involve a large scale social network containing individual details. The co-evolution of epidemic dynamics and human behavior requires the simulation systems to compute complex real-world interventions. Calls from public health policy makers for executing such simulation studies during a pandemic typically have tight deadlines. It is highly desirable to implement new interventions in existing high-performance epidemic simulations, with minimum development effort and limited performance degradation. Indemics is a database supported high-performance epidemic simulation framework, which enables complex intervention studies to be designed and executed within a short time. Unlike earlier approaches that implement new interventions inside the simulation engine, Indemics utilizes DBMS and reduces implementation effort from weeks to days. In this paper, we propose a methodology for modeling and predicting performance of Indemics-supported intervention studies. We demonstrate our methodology with experimental results.