Rapid development and execution of complex agent-based energy system simulations: the FAME framework

Abstract

Agent-based models and simulations (ABMS) are capable of capturing complex behavior of heterogeneous actors and their emerging dynamics. Yet, in energy systems modelling, ABMS are clearly a minority. To foster rapid development of ABMS in this field of research we introduce the open software framework FAME. It provides powerful components for common tasks in agent-based simulation models addressing, e.g., data input, data output, scheduling agent actions and communication between agents. Java models based on FAME are highly performant, portable and parallel-computing compatible. Using FAME reduces programming efforts for modelers, enabling them to focus on developing model logic instead of writing supplemental code. With FAME's additional Python tools and graphical user interface, users can conveniently manage, create, and edit model configurations. FAME was successfully tested on various computer and operation systems, ranging from small desktop computers to large-scale high-performance computing clusters.