(529d) Residence Time Distribution in Reactor Networks

Networks appear in a wide variety of applications, such as communication, transportation, computers, financial systems, and the natural ecology. In recent work, we have proposed the use of interconnected reactor networks as a paradigm for robust manufacturing environments that have ultimate flexibility to self-reconfigure, when coupled with an agent-based control system, in response to external disturbances and demands. The structure and organization of a network affect the survival and performance objectives of a system. Here, we consider the behavior of bidirectional linear networks, ring networks and grid networks under the effect of a step tracer input introduced at a specific location in the network. The ?inter-connection flow rate' is a governing factor in the distribution of resources to different parts of the network. For networks of the order of two or three reactors, it is easy to predict the distribution of tracer depending on the magnitude of interconnection flow rates. However, it would be a challenging task to predict the tracer distribution for a large network or a network with random interconnection flow rates. For small scale networks, bifurcation diagrams serve as an important tool in analyzing the system configuration. Bifurcation diagrams are constructed for linear, grid and ring networks under conditions of random interaction flow rates. The bifurcation plots are evaluated using a statistical search technique called KINSOL. The propagation of tracer in a reactor network is studied by carrying out tracer experiments which consider the effect of tracer concentration in the farthest reactor. The magnitude of resource concentration in the farthest reactor and time taken for tracer to propagate can predict the performance of the system. We compare the performance of linear, ring and grid networks and observe that the grid network gives superior performance in terms of magnitude of resource and speed of response. We also consider the performance of these networks for increasing network size. A large network with random inter-connection flow rates can give rise to complex tracer distribution patterns. We can isolate the strong and weak interactions in a network and predict how a network would behave under extreme conditions. We aim at identifying sub-networks within a larger network and compare the network to extreme conditions.