Investigation of a Biofilm Reactor Model with Suspended Biomass

Abstract: Biofilms are compact, sessile microbial communities that attach to surfaces in aqueous environments. In wastewater treatment, they are especially important for removal of phosphorus and nitrogen, which, if released into a receiving water body, can cause severe eutrophication. Mathematical models of biofilms in wastewater are used to understand the underlying processes and to describe and analyze biofilm development. Although biofilm reactors always contain an amount of suspended biomass, this biomass is mostly neglected in mathematical models of biofilm reactors. This thesis is based on four papers which investigate the role of suspended biomass in biofilm reactors. A one-dimensional mathematical model of biofilm and suspended biomass in a continuous stirred tank reactor is presented and analyzed in the first paper. The underlying model is a hybrid model of chemostat-like mass balances for the substrate and biomass in the reactor, coupled with a free boundary value problem for the substrate in the biofilm. In a single species single substrate setting, stability conditions for washout and persistence are given. It is found that biofilm and suspended biomass are either both present in the reactor or completely washed out. Numerical simulations show that biofilm dominates over suspended biomass in the longterm reactor performance, but that suspended biomass is relatively more efficient at substrate removal. The model is extended to a microbially and algebraically more complex multi-species multi-substrate model in the third paper, describing two-step nitrification in a Moving Bed Biofilm Reactor (MBBR). Nitrogen enters the reactor in the form of ammonium and leaves as nitrate after an intermediate conversion to nitrite. Numerical simulations show that suspended biomass does not contribute significantly to the overall reactor performance, but is substantial in the intermediate processes. In the second paper, the biofilm model is numerically validated against microelectrode measurements of oxygen gradients across the biofilm depth of a nitrifying biofilm attached to a suspended carrier harvested from an MBBR. Finally, a single species single substrate case with a limited amount of substrate and treatment time is considered as a two-objective optimization problem. With the bulk flow velocity as the control, different classes of admissible functions are investigated. It is found that, given the uncertainties in the initial data, none of the other functions perform better than the constant flow rate, i.e. the uncontrolled reactor.