An Agent-based Framework for Control of Reactor Networks with Autocatalytic Replicators

Several recent studies on autocatalytic reactions in single and coupled continuous stirred-tank reactor (CSTR) networks have demonstrated a rich spectrum of complex behavior. From a control systems perspective, the operating regime of a CSTR network can be manipulated by changing the flow rates between the reactors. Systems of more than one CSTR require multiple controllers and may need transients through several operating regimes to achieve the desired operation. This may require a hierarchical control structure whereby local control objectives can change dynamically in order to achieve the global control objective of the system. An agent-based control system is used to observe and control various aspects of a CSTR network. The primary focus of analyzing complex emergent behavior is demonstrating methods or combinations of methods that influence the behavior and capabilities of the agent-based control system. Simulation studies illustrate scenarios of interest, especially conditions that lead to emergent behavior.