was a 0.3-m-high supporting layer of cobblestone (the particle size
of 2–4 mm) at the bottom of the column, and the filter bed was
fixed in the tube at a height of 1 m. The flow rate was 916 mL/
min with an empty bed contact time of 8.6 min in the column.
The pilot-scale column was backwashed when the water quality
of the effluent was deteriorated or the water level reached a preset
overflow location (1.6 m above the bed layer). Back washing was
conducted via the following procedures: ① air scouring for
3 min, ② simultaneous air and water scouring for 4 min, and
finally ③ water scouring for about 3 min. The air flow intensity
and water flow intensity was 50.8–72.0 m3/(m2h) and 15.1–
36.0 m3/(m2h), respectively. With such back washing intensity,
leakage of the media would not happen. Schematic of the pilotscale
filter system is presented in Fig. 1. Raw groundwater was
pumped into a tank. The concentration of dissolved oxygen (DO)
was increased to 6.5–7.0 mg/L by aeration, and CO2 in the raw
water was stripped. Influent was added into the columns (R1 and
R2) by a booster pump.
The ripening sands were fixed into two identical columns. One
filter column, R1, was used as the comparison filter and for testing
the influence of temperature and pH. The other column, R2, was
used for the filter inactivation process and ammonium oxidation
kinetics.
2.3. Experimental set-up
2.3.1. Rapid start-up method
The virgin quartz sands (average diameter of 1 mm) were fixed
into the filter column at about 1.5 m depth. Before the in fluent
entered the filter column, 20 m of polyethylene pipe (inner diameter
of 20 mm) was used to allow for a sufficient reaction time
(about 6.8 min) with the potassium permanganate solution (about