| The feasibility of anaerobic degradation of propylene glycol, a major component of deicing fluid, was investigated using fluidized bed reactors.; The primary objective of this research was to determine the reaction mechanisms and the responses of these reactions to various inputs. A series of batch tests and continuous experiments achieved this objective using fixed film and suspended growth microorganisms. Two identical bench-scale fluidized bed reactors were built to support this purpose incorporating control devices. Propylene glycol was degraded to methane with the intermediate products of propionic acid and acetic acid under thermodynamically favorable conditions. However, n-propanol was shown to govern the overall reaction by regulating the reducing equivalents according to the operating conditions.; The secondary objective was to develop a novel mathematical model to describe the reactions with the predetermined parameters and the experimental observations. Stoichiometry, hydrodynamics and reaction kinetics were all included in the model. A biofilm diffusion mechanism was also taken into account focusing on the spatial distribution of multi-species of microorganisms. The results illustrated how a specific group of microorganisms may dominate a layer of biofilm and what factors govern the overall performance of the reactor.; The tertiary objective was to develop an appropriate control strategy to operate the reactor safely and to attain the main goal rapidly. As the first step, a PID (Proportional-Integral-Derivative) control scheme was applied. The transfer function was obtained by linearizing the nonlinear model developed for the CSTR mode. The PID controller was tested experimentally using the parameters determined from the IAE (Integral of the Absolute Value of the Error) criterion. To identify the system characteristics prior to actual control of the reactor, as the second refinement a recursive system ID was implemented using ARX (Auto Regressive with eXternal Input) model structure. After the system was identified, the actual outputs followed the predicted values for 100 hours in the off-line experiment. The on-line adaptive control algorithm associated with the off-line system ID was then tested in a final improved control scheme, with an objective function to be satisfied. This novel idea was successfully applied to the underlying system for over 200 hours. |
