An investigation of the fast dynamics in a nitrifying trickling filter

Rapport, 1996

The fast dynamics of a large pilot scale nitrifying trickling filter (NTF) using effluent waste water from Rya Waste Water Treatment Plant in Göteborg, Sweden, has been investigated experimentally and by simulations. The plant is 7.2 m high, have a diameter of 2.7 m, and was filled with a cross-flow media with a specific surface area of 226 m^2/m^3. Continuous ammonium meters connected to the NTF measured influent and effluent ammonium concentrations. Two different phenomena may affect the fast dynamics in this type of biofilm reactors: The mixing in the bulk flow and the dynamics within the biofilm, i.e. the mixing in the biofilm and the response time for the bacteria to changes in substrate concentrations. Pulse response experiments conducted at two different flows (7.3 l/s and 14.5 l/s), where dissolved LiCl was used as trace substance, showed that the flow through the NTF could not be characterized as laminar, but significant mixing occurs inside the plant. The residence time distributions are well approximated by four identical continuously stirred tanks in series, and from the mean residence time the liquid film thickness was estimated to be 0.5 mm. Two experiments, where the influent ammonium concentration was rapidly increased while the flow was held constant at 14 l/s, were also conducted as well as an experiment where the flow was stochastically varied around 12 l/s. Comparisons are made between the measured effluent concentrations in these experiments and simulations of a model of the plant, where the NTF is divided into a series of continuously stirred tank reactors (CSTRs) having the total volume estimated from the pulse response experiments. The nitrification rate in each CSTR is modelled as a first order dynamic system driven by a physically derived expression of the stationary nitrification rate. Simulations for different values of the time constant of this dynamic system, and comparisons with the experimental data, shows that the dynamics within the biofilm are much faster than the dynamics of the mixing in the bulk. Implicitly, this means that the response times for the nitrifying bacteria to changes in ammonium bulk concentration were less than a few minutes. Hence, the fast dynamics in the biofilm can in many cases be neglected, which greatly facilitates simulations of many nitrifying biofilm reactors when more complex biofilm models are used.