Analysis of municipal wastewater sand filtration with denitrification, phosphorus removal, and suspended solids separation

QC 20241128

Abstract

Discharge of high concentrations of nutrients to seas and lakes will lead to eutrophication and thereby increased growth of algae in the water. The water bodies receive nutrient discharges from several sources e.g. agricultural activities and wastewater treatment plants (WWTP). Despite efforts to reduce emissions from WWTPs to the Baltic Sea, 16% of phosphorus and 33% of nitrogen of the total anthropogenic emissions originate from Swedish municipal WWTPs.

The Henriksdal WWTP in Stockholm introduced nitrogen removal to perform nitrification and denitrification. Following this decision, it was necessary to increase the hydraulic retention time in the biological treatment step. The volume of the aeration tanks was increased from 65 000 m3 to 204 000 m3 by increasing the depth of the tanks from 4.9 m to 12.0 m. Three more aeration tanks were also built and added to the 11 tanks, resulting in 14 tanks in total. To further increase the nitrogen removal capacity, the concentration of suspended solids (SS) in the aeration tanks was increased from 1250 mg MLSS/L to 2500 mg MLSS/L. This would overload the secondary sedimentation tanks, hence three more tanks were added to the 14 tanks. At high hydraulic loads, sludge overflow would probably occur in the secondary sedimentation tanks despite this measure, and 60 deep-bed two-media down-flow filters, with a length of 10 m and a width of 6 m, i.e. a surface area of 6o m2, as the final tertiary treatment step in the WWTP. This gave a total surface area of 3600 m2. Each filter was filled with 0.5 m sand on the bottom and above 1.0 m of crushed Leca® material, i.e. a total filter depth of 1.5 m. The effluent demands are ≤10 mg total N/L as a yearly average value, ≤3 mg NH4-N/L as an average value from July to October, and ≤0.3 mg total P/L and ≤8 mg BOD7/L as quarterly average values. The plant has an average hydraulic load of 328 500 m3/d and 884 000 persons connected.

A pilot-scale filter study was performed before the filters were built and a full-scale filter study was done after some years of operation of the full-scale filters. The aim was to study simultaneous nitrogen, phosphorus, and SS separation in the filters. Results from the full-scale study performed at 3.3 m/h showed that it was possible to achieve concentrations of approximately 0.01 – 0.17 mg PO4-P/L with Fe dosage, 0.09 – 0.70 mg total P/L, 0.5 – 6.1 mg NOx-N/L with carbon source dosage, and 2 – 7 mg SS/L in the filtrate. With a high carbon source dosage, the concentration could decrease to 0.2 mg NO3-N/L in the filtrate. At that mode of operation, however, there is a risk of having a remaining carbon source in the filtrate giving too high concentrations of BOD7.

It was important to receive long times of operation for the cycles of filter operation in order not to keep the filters in a queue for backwashing. At least two backwashing sequences during backwashing were necessary to flush out the SS caught during the filter cycle. The remaining SS would otherwise decrease the time of the subsequent filter cycle. If there was a sludge overflow in the secondary sedimentation tanks, at least three backwashing sequences would be needed to clean the filter bed. With Fe and carbon source dosages, the time of operation became around 16 – 27 hours, and without carbon source dosage, times of the filter cycles of 36 – 72 hours were found in the full-scale filter at a hydraulic load of 3.3 m/h. Without both Fe and carbon source dosage, the operation time could exceed 200 hours depending upon the Lena Jonsson TRITA-ABE-DLT-2439 vi concentration of SS in the influent to the filter. The hydraulic load to the filters influenced the operation time to a lesser degree compared with Fe and carbon source dosage. A Fe dosage of around 2 g Fe/m3 was kept on the filters during the full-scale study.

The denitrification rate in the filter bed (per m3 or m2) in the full-scale study was calculated to be 13.1 g NOx-N/(m3·h) and 19.6 g NOx-N/(m2·h). In the pilot-scale study, the denitrification rate was calculated to be 21.3 g NOx-N/(m3·h) and 30.5 g NOx-N/(m2·h). It is probably easier to watch over a pilot-scale process and thereby receive better results. These values were found in steady state i.e. after the first 1 – 2 hours of a filter cycle. Before that, the denitrification process had not yet fully started and the denitrification rate became around half of the rate in steady state, 6.4 g NOx-N/(m3·h) and 9.7 g NOx-N/(m2·h) in the full-scale filter.

The operation time of a filter cycle depends on the dynamic head loss developing in the filter during the filter cycle. The filter is mainly clogged by biological SS leaving the secondary sedimentation tanks, SS origins from primary settled wastewater (PW) bypassing the biological treatment step during high influent flows to the plant, and nitrogen gas and carbon dioxide bubbles produced in the filter during denitrification. SS is also produced during denitrification, but this SS was not found in the full-scale filter when excavating the filter bed after the study. The nitrogen gas and carbon dioxide bubbles abruptly clogged the filter in contrast to SS which slowly clogged the filter proportionally to the influent load of SS. Chemical SS mainly consisting of iron phosphate precipitated in the filter when Fe is dosed might also clog the filter, but this was not investigated in the study.

An initial head loss that slowly increased during several years of operation was detected in the filters. This head loss mainly consisted of inorganic precipitates in the slots of the nozzles and on the surface of the Leca® grains in the filter bed. To decrease this head loss, the nozzles were exchanged for new clean nozzles or cleaned by brushing. The initial head loss also developed when filter bed material accumulated in the channel below the filter bottom and clogged the nozzles from below. This partly blocked the backwashing and fluidization of the filter bed. The Leca® grains removed from the bed were estimated to exceed 0.1 m3 at an inspection. Backwashing more frequently or with more sequences also prevented the initial head loss.

urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-356921