A comprehensive approach to the incorporation of adaptive behavior into complex systems is provided through attentive combination of a wide... Show moreA comprehensive approach to the incorporation of adaptive behavior into complex systems is provided through attentive combination of a wide variety of methods from di↵erent fields. Two examples of complex systems were subjected to this approach, which involves modeling, analysis and control of the processes. The first system of interest is an autocatalytic reactor with multiple resources and multiple species. A continuous stirred-tank reactor model was developed, and subsequently analyzed using bifurcation analyses. The multiplicity character of the system was investigated, and emergent features, such as stable coexistence of autocatalysts, were discovered. Next, an agent-based control system was implemented on the process model using finite-state machines with simple rules as local control elements. The control system is capable of transitioning the process between the numerous steady states evidenced by the bifurcation analysis, as well as exploring the state space for a feasible steady state, given arbitrary set points for the autocatalysts. Eventually, Equation-Free coarse graining methods were applied to this model to investigate the local stability of the agent-controlled system. It was shown that this approach indeed was able predict the correct stability behavior for a given steady state, using only input-output information from short simulations with the closed-loop system. The second system considered was the activated sludge process, which was modeled based on data from an existing wastewater treatment plant. This realistic and complex model was then subjected to in-depth bifurcation analyses using various model parameters. The bifurcation analyses revealed many interesting phenomena about the process, including steady state multiplicity and conditional stability. In addition, potential risks of operation under certain conditions, such as the loss of biomass, were outlined. Furthermore, the aeration requirement of the process was investigated and the possibility for 55% savings in aeration with the same e✏uent quality was illustrated. Finally, in order to enhance the potential of savings in aeration, an agentbased control system was implemented on the activated sludge process. This control system optimized the local airflow rates in real time, and provided a decrease of up to 65% in the aeration rate. In addition, the process was shown to endure severe storm events and aeration failures thanks to the adaptive features of the agent-based control system. Ph.D. in Chemical Engineering, July 2015 Show less
