Nitrification in moving bed and fixed bed biofilters treating effluent water from a large commercial outdoor rainbow trout RAS

The nitrification performance of two fixed bed (FB) biofilters and two moving bed (MB) biofilters was evaluated. They received the same cold (8 °C) influent water from a commercial outdoor RAS facility producing rainbow trout (average density 32 kg m−3). The filters were constructed as four identical 5.5 m3 tanks with different filter media inside and tested simultaneously in duplicate. FB filters were filled with 4.2 m3 polyethylene netshaped cylinders (Bioblok®, 200 m2 m−3), and MB filters with 2 m3 polypropylene carriers (Biomedia, 850 m2 m−3). Nitrification rates were measured 31/2 months after start-up, and inlet water was supplemented with ammonium chloride in order to determine maximum nitrification rates (0-order kinetics). The filters were conditioned at an inlet TAN concentration of 2.89 ± 0.1 mg l−1 and water in-flow ranging from 1 l s−1 to 4 l s−1. Expressed as volumetric total ammonia-N (TAN) removal rate, the MB filters had significantly higher removal rate (231 ± 17 g N−1 m−3 d−1) compared to the FB filters (92 ± 2 g N−1 m−3 d−1). Expressed as surface specific TAN removal rate MB filters had significantly lower removal (0.27 ± 0.02 g m−2 d−1) than FB filters (0.46 ± 0.01 g m−2 d−1). When conditioned to a higher inlet TAN concentration (6.27 ± 0.39 mg l−1) for 2 weeks, the FB filters increased the removal rate (146 ± 3 g m−3 d−1 or 0.73 ± 0.01 g m−2 d−1) while the MB filters had unaltered performance. The results indicate, that the more heterogeneous and stratified biofilm to be expected in FB, can react more flexibly when challenged with changes such as differences in TAN loading. The effect of dissolved oxygen level on FB filter nitrification rates was additionally tested at TAN 5.35 ± 0.06 mg l−1. Below approximately 60% saturation (7.1 mg O2 l−1) measured at the filter outlet, nitrification rates started decreasing rapidly. An exponential expression (y = −10.05 + 10.48(1 − e−0.0798x), R2=0.96) was found to model the whole data range from 40% to 80% DO saturation well.