Experiment utförda på en nitrifierande biobädd 1995

Report, 1996

A number of experiments was conducted on a pilot scale nitrifying trickling filter during 1995. The experiments aimed for better knowledge of the flow-dependent fast dynamics, and the slow bacterial dynamics. The residence time distribution was investigated by a number of impulse response experiments where dissolved LiCl was added to the influent. The experiments showed that the amount of water in the trickling filter is almost independent of the flow through the plant, and corresponds to a liquid film thickness of approximately 0.5 mm. The residence time distribution can be approximated by a model with four or five identical and ideally continuously stirred tanks. Based on the results of the impulse response experiments the plant is modelled by four continuously stirred tank reactors in series, where the nitrification in each tank is described by a physically derived nonlinear expression. Data from a few step response experiments, where the ammonium concentration in the influent was raised from a low constant level to a high constant level at a constant flow through the plant, was compared to model simulations. The comparisons showed that the fast dynamics in the biofilm can be neglected in comparison to the dynamics caused by the mixing in the bulk and the residence time distribution. Implicitly, this means that the response time for the active nitrifying bacteria to changes in ammonium concentration is less than a few minutes, also when the ammonium load has been very low for a long time. An experiment, where the flow was stochastically varied around an operating point during one day, showed that the simple model derived sufficiently well describe the fast dynamics of nitrifying trickling filter also when the flow changes. When the ammonium concentration in the effluent is low, a model where the nitrification rate is assumed constant is not sufficient. The slow dynamics that depend on the growth and decay of the active nitrifying bacteria was investigated by a three months long step response experiment, where the ammonium concentration in the influent first was held at a high level (not full nitrification) for approximately one month and then at a low level (approximately 50% of the nitrifying capacity) for one month, and finally at the same high level as before for one more month. In spite of several practical problems, the experiment indicated that it takes one to two weeks for the concentration of active bacteria in the biofilm to increase to a new higher concentration after the raise in influent ammonium concentration. The corresponding increase in nitrification rate is approximately 20%. The two periods of the same high influent ammonium concentration was during periods with different water temperature. Comparisons of the nitrification rate between the two periods indicated a stronger dependency on the temperature than has earlier been observed. The standard temperature dependency of the maximum growth rate for nitrifying bacteria that are used for laboratory scale experiments may well apply also for this large scale process. During periods of the experiment the ammonium sensors were not working. Therefore the possibility to determine the influent ammonium concentration based on the flow into the plant was investigated. Both black box models and a physically based model was fitted to data. The investigation showed that with a good model of the influent flow to the plant it may be possible to predict the ammonium concentration with quite good accuracy. The trickling filter was flooded weekly for a couple of hours for predator control. An investigation of the nitrification rate before and after the floodings showed no short term effects of the flooding. When the pilot plant was taken out of operation at the end of the year the uppermost meter of the plant was investigated. It was observed that the biofilm thickness was approximately 0.5mm and no bare surfaces without biofilm could be observed.