Trickling Filters for Upgrading Low Technology Wastewater Plants For
Total Page:16
File Type:pdf, Size:1020Kb
Trickling filters for upgrading low technology wastewater Water Science and Technology plants for nitrogen removal P. Pearce Thames Water Utilities Ltd, Research and Development, Spencer House, Manor Farm Road, Reading RG2 0JN, UK (E-mail: [email protected]) Abstract Previous work through the 1990s in the Thames Water region in the UK has demonstrated the ability of the trickling filter process to produce fully nitrified effluents, reliably throughout the year. The original data used for the nitrification model derivations have been reanalysed, to investigate the degree of nitrogen Vol 49 No 11–12 pp 47–52 removal across the process. Removals of total nitrogen ranging from 0% to over 50% were observed across the trickling filter process and calculated total nitrogen removals of 26–63% were obtained when primary treatment was included. The degree of nitrogen removal and biological denitrification (excluding cellular assimilation) was found to be strongly influenced by BOD load, irrigation velocity and media size. Regression models were produced which gave good predictive relationships for the data ranges used. The models produced worked for filters used with and without a recirculation of effluent nitrate which suggests that a significant degree of nitrification occurred in areas of high heterotroph activity (BOD removal). The simplicity and energy efficiency of the trickling filter process, combined with its capacity for full © Thames Water Utilities Ltd 2004 nitrification and partial denitrification, make the process attractive as a combined process used with pond systems in developing countries where nitrogen removal may be required. Some of these synergies have already been developed with the PETRO® process in South Africa. Keywords Appropriate technology; nitrogen removal pond systems; trickling filters Introduction Trickling filters are low technology processes that have been proved to be capable of pro- ducing high quality, fully nitrified effluents. The process is robust and simple to operate and control. Due to the small number of moving parts (pump and rotary distributor) it requires minimal skilled maintenance and due to the efficiency of oxygen transfer, they are highly energy efficient. These attributes lend the process to application for developing countries. In recent years, Thames Water has undertaken substantial research which has signifi- cantly improved the understanding of the design and operating parameters consistent with maintaining full nitrification on trickling filters throughout the year. One of the key design parameters to achieve this is hydraulic loading (Pearce, 2002). To achieve complete nitrifi- cation throughout the year, effluent recirculation is usually required for single filtration plants. Single filtration is defined as effluent passing through a single trickling filter either as a single pass or with a degree of recirculation. Double filtration is defined as effluent passing through two filters in series with intermediate settlement with or without a degree of recirculation. In typical applications in the Thames basin stone media trickling filters are loaded below 0.15 kg BOD/m3/d. Filter depth is usually 1.5–1.8 m and so in the absence of effluent recirculation this volumetric load results in low wetting rates expressed as unit flow volume per unit plan area of trickling filter and hence, poor utilisation of filter media surface area. Incorporating effluent recirculation increases the wetted surface area and has been demon- strated to improve the degree and reliability of nitrification (Pearce and Williams, 1999; Pearce and Foster, 1999). In addition to the improved media wetting, recirculation also 47 introduces nitrate into the top of the filter where heterotroph activity, and therefore potential denitrification activity, is highest. For biological nitrogen removal reliable nitrification is an essential prerequisite, there- fore trickling filters can achieve the required degree of nitrification in a coupled process where additional denitrification can be achieved via the recirculation of trickling filter effluent via a denitrification step. Facultative pond systems have the potential for denitrifi- P. Pearce cation of the recycle stream and so in such systems where nitrification is limited in down- stream aerobic ponds, trickling filters represent a relatively low footprint upgrade for nitrification. The higher hydraulic loadings resulting from recirculation will improve the nitrification performance of the trickling filter providing the nitrate for denitrification in the recycled stream as well as potentially improving the performance of the facultative pond systems. The coupled use of pond systems and trickling filters has already been devel- oped in South Africa for the improvement of lagoon effluent BOD quality with the PETRO® process where many synergies between the two processes have been researched (Shippin et al., 1998). Work is ongoing to optimise biological nutrient removal. Results Data from the Thames trickling filters used for development of the nitrification models have been reanalysed to assess the degree of denitrification occurring. This has shown that a significant proportion of nitrogen removal can take place. Including assimilated nitrogen removal this was found to account for up to 55% total nitrogen removal from the primary effluent on the sites studied. This will reduce the amount of denitrification required on any dedicated anoxic process unit to achieve a desired effluent total nitrogen standard and so should simplify operation and control of such a sidestream process. Observed nitrogen removal in the surveyed plants is shown in Table 1 with calculated overall plant removal based on an assumed 15% removal of TKN in primary sedimentation. The operating conditions are shown in Table 2 together with calculated nitrogen removal via biological denitrification. Nitrogen loss via assimilation is based on biomass composi- tion incorporating 5% nitrogen by dry weight and a net biomass yield of 0.65 g per g BOD applied. The data are from seven sets of trickling filters on five sites and are derived from averaged 24-hour composite samples from periods of 1–6 months operation. Nitrogen removal can be seen to vary widely from zero in secondary nitrifying filters to over 50% in filters with higher BOD loadings but still achieving complete or near complete nitrification. Figure 1 shows the applied model derived from regression analysis. The analysis was Table 1 Observed and calculated nitrogen removals in trickling filters Site Effluent Effluent Effluent Observed % Calculated % N TN mg/l TON mg/l NH4-N mg/l N Removal removal from from 1o effluent inlet sewage* Arborfield Primary filters 25.3 18.6 5.5 32% Arborfield Secondary filters 24.5 23.3 0.1 3% Arborfield Overall 24.5 23.3 0.1 33% 43% Wokingham 23.6 20.1 2.3 50% 63% Wisley 23.3 21.0 1.1 37% 46% Fleet** 12–15.5 9.7–11.7 2.1–4.0 39–45% 50–53% Manor Farm Primary filters** 22–29 12–19 7.0–10.1 18–29% Manor Farm Secondary filters** 24–27 21–25 0.4–0.7 0–7% Manor Farm Overall** 24–27 21–25 0.4–0.7 12–29% 26–40% * Assumes 15% TKN removal during primary sedimentation 48 ** Ranges indicates variation of operating conditions Table 2 Operating conditions and non-assimilated N removal Site Filter Recirculation Observed % Calculated % configuration ratio r/q N Removal N removal from filter from filter feed feed via denitrification * Arborfield Primary filters Double 0 32% 28% Arborfield Secondary filters Double 0 3% 0 –1% Arborfield Overall Double 0 33% 29% P. Pearce Wokingham Single 0.47 50% 45% Wisley Single 0.42 37% 33% Fleet** Single 0–0.85 39–45% 34–40% Manor Farm Primary filters** Double 0 18–29% 25% average Manor Farm Secondary filters**Double 0 0–7% 0–1% Manor Farm Overall** Double 0 12–29% 25% average * Calculated assuming mg/l nitrogen assimilation = mg/l BOD removal × 0.0325 – see text performed on the overall averages of each data set. The results need to be interpreted with some caution, as not all of the variables are fully independent. For example TKN loading will normally increase with BOD loading, and irrigation velocity will tend to increase with recirculation ratio as will TON loading. Reasonable regressions were obtained using only truly independent variables (BOD load, irrigation velocity – including recirculation where present, and media size) and the output trends are very similar. The regressions on the larg- er data sets have therefore been presented. Several sets of loading terms were analysed and the best fit was found by the following equation: %TN removal = 0.644 + (4.3Bv)–(0.25r/q)–(0.076iv)+(4.96TONv)–(6.74TKNv)– (0.072g) where Bv = volumetric BOD load kg /m3 filter volume/day Range 0.18–0.02 r/q = recirculation flowrate/feed flowrate Range 0–0.85 iv = irrigation velocity m3total flow/m2 plan area/day Range 1.4–4.2 TONv = volumetric oxidised nitrogen loading kg TON/m3 filter Range 0–0.034 volume/day Nitrogen Removal Regression Model %TN removal = 0.644.(4.3Bv)-(0.25r/q)-(0.076iv)+(4.92TONv)-(0.0072g) 60% R2 = 0.9909 50% 40% 30% 20% Observed %TN removal 10% 0% -10% 0% 10% 20% 30% 40% 50% 60% Predicted %TN removal Figure 1 Regression for TN removal from primary effluent 49 TKNv = Total Kjeldahl nitrogen loading kg TKN/m3 filter volume/day Range 0.03–0.06 g = nominal media size, mm Range 28–63 Sensitivity analysis showed BOD loading to exert the greatest influence on the percentage removal of total nitrogen (9.35% increase in TN removal with 10% increase in BOD load), the next most sensitive parameter was media size followed by irrigation velocity (both P. Pearce inversely proportional to %TN removal). The other terms have been included as they sequentially improved the fit and reduced the residual y estimate error to 3.3%. Discussion The sensitivity analysis performed on the regression analysis helps to clarify the key mech- anisms involved.