IWA Publishing 2012 Water Practice & Technology Vol 7 No 3 Comparison of denitrification-nitrification and step-feed activated sludge processes with dynamic simulation K. Sahlstedt a, H. Haimi b and J. Yli-Kuivila c a Pöyry Finland Ltd, PO Box 50, FI-01621 Vantaa, Finland. E-mail: kristian.sahlstedt@poyry.com b Dept. of Civil and Environmental Engineering, Aalto University, P.O. Box 15200, FI-00076 Aalto, Finland. E-mail: henri.haimi@aalto.fi c HSY Water, PO Box 100, FI-00066 HSY Finland. E-mail: jukka.yli-kuivila@hsy.fi Abstract In conjunction with choosing the treatment process for the new wastewater treatment plant of Espoo, Finland (400,000 P.E.), Denitrification-Nitrification (DN) and Step-Feed activated sludge processes were compared in terms of required basin volume and consumption of aeration air and methanol. The comparison was made using dynamic process simulation. The advantages of the step-feed process reported in literature smaller volume required to treat an equal load or ability to treat a higher load in an equal volume were questioned. In terms of consumables, the two processes were found practically equal. This is, to the best of our knowledge, the first comparison of these process configurations with dynamic simulation. Key words: DN, modelling, nitrogen removal, simulation, step-feed, wastewater treatment INTRODUCTION A new wastewater treatment plant will be constructed for the city of Espoo, Finland by the end of 2020. The population equivalent of the new plant will be 400,000. The old wastewater treatment plant, a conventional activated sludge plant based on denitrification-nitrification (DN) process (P.E. 250,000) will be in use until the commissioning of the new plant. In conjunction with a feasibility study made for the new plant, the step-feed process was identified as a promising alternative both for the new WWTP and for optimizing the old one. In order to quantify the differences between the two processes as realistically as possible, taking into account the hourly variations of influent load, they were compared using dynamic simulation. The objective was to identify the most important factors for investment and operational costs and operational requirements. DN process The DN or pre-denitrification process (DN) is a well-established solution for total nitrogen removal from municipal wastewaters. In this configuration, all wastewater (after primary treatment) is introduced to the beginning of the aeration basin, together with return sludge. The beginning of the basin consists of anoxic compartments, in which nitrate is denitrified to nitrogen gas. Subsequently, there are aerobic zones, in which ammonium is nitrified into nitrate. At the end of the aeration line, there is usually a small gas removal compartment, in which free oxygen is removed from the mixed liquor. Nitrate is recycled from the gas removal zone to the beginning of the aeration line by pumping. This configuration is a popular solution for nitrogen removal in the Nordic countries and Northern Europe. Schematic representation of the DN process is given in Figure 1.
Figure 1 Schematic representation of the DN process as simulated in Phase 1. Grey ¼ anoxic; white ¼ aerobic. Above: T ¼þ8 C. Below: T ¼þ14 C. Step-feed process In the step-feed process, wastewater is introduced (after primary treatment) to several points along the aeration basin. Usually, two, three or four steps are applied. When nitrogen removal is required, anoxic and aerobic zones alternate, and wastewater is introduced to the beginning of the anoxic zones. Typical feeding ratios are 42:32:26 for three steps and 15:35:30:20 for fours steps (Johnson et al. 2005). Separate nitrate recycle pumping is usually not applied. Reported advantages of the step-feed process as compared to traditional activated sludge processes include (i) treatment of up to 20% higher loads in equal aeration basin volume and settler area; (ii) alternatively, treating an equal load in 25% smaller volume; (iii) operation with higher sludge retention time (SRT) without increasing the solids load to the secondary clarifier; (iv) more stable nitrification; (v) reliable and flexible operation (Daigger & Parker 2000; Johnson et al. 2005; Zhu et al. 2007). The step-feed process is widely applied for removal of BOD and nitrogen e.g. in the United States and Canada (debarbadillo et al. 2002; Johnson et al. 2005). Promising results for enhanced biological phosphorus removal have also been reported (Daigger & Parker 2000). Schematic representation of a step-feed process with three steps is given in Figure 2. Figure 2 Schematic representation of the step-feed process as simulated in Phase 1. Grey ¼ anoxic; white ¼ aerobic. Above: T ¼þ8 C. Below: T ¼þ14 C.
MATERIALS AND METHODS Simulation model The simulations were based on a previously constructed, calibrated and validated mathematic model of the existing Espoo WWTP. The model was developed by the primary author of this paper in 2006 2007 and it has been described in Phillips et al. 2010. The model was edited as necessary to enable the simulations described below. Wastewater characterization and parameters of the biokinetic model were not changed from the values of the original, validated model. In all simulations, a static steady-state with average values of influent flow and concentrations was first simulated. After that, a 48-h time series of actual influent flows and concentrations was simulated repeatedly until the effluent concentration profile did not change (dynamic steady-state). The applied simulation software was Hydromantis GPS-X 5.0. Simulations, Phase 1 First, the necessary aeration basin volume was determined for both configurations. The dimensioning temperature was þ8 C. The total volumes of the basins were adjusted until the required treatment results were achieved. Second, the dimensioned processes were operated in the temperature of þ14 C, which is the yearly average temperature recorded at the existing Espoo WWTP. A maximum number of zones were kept anoxic without violating the requirements for effluent ammonium nitrogen. DN-process consisted of eight functional zones, measuring respectively 14, 13, 13, 9.5, 9.5, 19, 19 and 3% of total basin volume. This configuration was equal to the existing process at the Espoo WWTP. The step-feed process consisted of three steps, each of which included three zones measuring 8.3, 8.3 and 16.8% of the total volume, respectively. These configurations and their operational modes in different temperatures are presented graphically in Figures 1 and 2. The boundary conditions of the simulations are presented in Table 1. Both configurations were equipped with fine-bubble bottom diffusers, with equal diffuser density and specific oxygen transfer efficiency. Table 1 Boundary conditions for the simulations of Phase 1 Parameter Unit DN Step-feed Influent division ratio 42:32:26 Methanol dosing to 2 nd zone 34:33:33 Return sludge pumping % of influent flow 150 150 Nitrate recycle pumping % of influent flow 150 Sludge volume index ml/g 120 120 DO setpoint in aerated zones go 2 /m 3 2.0 2.0 MLSS to settling, max kgmlss/m 3 5.0 5.0 Effluent ammonium nitrogen, max gnh 4 -N/m 3 2.0 2.0 Total nitrogen removal rate, min % 70 70 Simulations, Phase 2 In the simulations of Phase 1, operating parameters of the step-feed process were based on typical values found in the literature and assumptions. However, the number of steps, the division of influent wastewater and the division of methanol affect the performance of the step-feed process (debarbadillo et al. 2002; Tang et al. 2007). Therefore, the configuration and influent and methanol feeding ratios of
the step-feed process were optimized by engineering judgement, and the dimensioning and consumable requirements were assessed again. A four-step process was introduced as an additional option. A gas removal zone measuring 1% of the total volume was added after each step of the three- and fourstep processes except for the last step. The configuration and settings of the DN process applied in Phase 1 were considered optimal as such, because they were as close as possible to the actual conditions at the existing Espoo WWTP. With this operational mode, the existing process has achieved e.g. yearly average nitrogen removal rates of more than 80%, although the nitrogen load of the plant is currently almost 50% above the dimensioning value. RESULTS AND DISCUSSION Simulations, Phase 1 The results of dimensioning and first comparison of the studied processes are presented in Table 2. The required total basin volume is about 5% smaller for the step-feed process than for DN process. In the dimensioning temperature, SRT of the step-feed process is 1.5 d longer than that of DN process, while the MLSS concentrations at the end of the basin are equal. Step-feed consumes approximately 15 30% more air than DN process. This is due to at least three reasons. First, in step-feed process, dissolved oxygen (DO) must be elevated from zero to setpoint level after each anoxic zone. In the DN process this is done only once, and all aerobic zones except the first one receive free oxygen with the flow of mixed liquor from the preceding zone. Second, less ammonium nitrogen is removed by assimilation and more by nitrification in the step-feed process because of the longer SRT. Third, the amount of total biomass is approximately 10% higher in the step-feed process, whereby the amount of endogenous respiration is also higher. Table 2 Results of the simulations of Phase 1 (dimensioning and first comparison) Dimensioning 8 C Comparison 14 C Parameter Unit DN 3-step DN 3-step Volume of basin m 3 24 200 23 000 24 200 23 000 Division of wastewater 42:33:25 42:33:25 Division of methanol 33:33:33 33:33:33 Aerated of total % 71 67 48 50 SRT, total d 13.5 15 12.1 12.7 SRT, aerobic d 9.6 10.0 5.8 6.4 Methanol dosage mg/l 80 65 15 22 Air flow, average Nm 3 /h 11 940 13 020 10 360 12 380 MLSS at beginning of line mg/l 4.8 6.3 3.8 4.6 MLSS at end of line mg/l 4.8 4.8 3.8 3.7 Effluent ammonium nitrogen, average mg/l 1.4 1.5 1.4 1.6 Total nitrogen removal rate % 71 71 71 71 DN process required 25% more methanol than step-feed process to reach equal total nitrogen removal rate in T ¼þ8 C. This is because of at least two reasons. First, the anoxic volume is, in this case, slightly smaller in DN than in the step-feed process. Second, because of nitrate recycle pumping, immediate retention time in the anoxic zones is shorter in DN than in step feed, whereby
the use of methanol is less efficient in DN. However, in the temperature of 14 C, the difference is reversed. The main reason is probably that in the step-feed process, a part of methanol is consumed aerobically due to the free oxygen transported to the second and third anoxic zone from the preceding aerobic zones. In the DN process, the recycle flows contain virtually no oxygen at all, and therefore all methanol is used for denitrification. In contrast with literature, step-feed and DN needed a practically equal total volume to reach full nitrification and given total nitrogen removal performance, when treating an equal load. Moreover, the step-feed process consumed more air and methanol in the average operating temperature than DN process did. To investigate why higher amount of biomass in the first zones of the step-feed process did not produce higher substrate conversion, the conversion rates of ammonium nitrogen in the aerobic zones of the respective processes are plotted in Figures 3 and 4. These plotted rates represent the total net conversion rate of ammonium, which includes the combined effect of nitrification and assimilation (decrease of ammonium concentration) and dissolution from hydrolysis of organic matter and endogenous respiration of biomass (increase of concentration). Figure 3 Net conversion rates of ammonium in the aerobic zones of the DN process (left) and step-feed process (right). T ¼ 8 C. Figure 4 Net conversion rates of ammonium in the aerobic zones of the DN process (left) and step-feed process (right). T ¼ 14 C. From Figure 3 we see that ammonium conversion rates in the studied processes are quite close to each other in the temperature of þ8 C. The average conversion rates, weighted with zone volumes, are 5.5 gn/m 3 /h for the DN process and 6.0 gn/m 3 /h for step-feed process. The net conversion of nitrogen per basin volume is, thus, about 10% faster in the step-feed process. When temperature is þ14 C, the conversion rates are practically equal: the average rates, weighted with zone volumes, are now 8.2 gn/m 3 /h for the DN process and 8.0 gn/m 3 /h for the step-feed process. In the DN process, where all wastewater is fed to the beginning of aeration line, the first 1 2 aerobic zones are always working at full speed, which is very efficient. The variation of influent
load is reflected in the conversion rates of the latter zones. Step-feeding of wastewater distributes the load variations directly along the length of the process. As a result, maximum conversion rates are higher in the step-feed process and the conversion rates do not reach zero even at the end of the basin. It seems that despite the higher MLSS concentration in the beginning of the basin, biological conversion of substrates does not proceed considerably faster in the step-feed process than it does in the DN-process. This observation may be explained by the fact that the concentrations of dissolved substrates (ammonium nitrogen, biodegradable COD) at the beginning of the step-feed process are lower than they are in the DN-process. Simulations, Phase 2 To improve the performance of the step-feed process, different feeding ratios of wastewater and methanol were tested. In addition, a four-step process was included in the study, and gas removal zones were implemented in both step-feed configurations. The results are presented in Table 3. Optimization of process design and operation brought only marginal (1 2%) savings in the required volume of the step-feed process. In the dimensioning temperature of þ8 C, methanol dose could be reduced by approximately 10% (from 65 to 57 mg/l), whereas in the four-step process no improvement was achieved. Although the total anoxic retention time is equal in both configurations, the immediate retention time in each anoxic compartment is shorter in the four-step process, whereby more methanol is left unused. This is reflected clearly in Figure 5, where the residual RBCOD (S S ) at the end of anoxic zones of the step-feed processes is plotted. In the dimensioning temperature, methanol consumption of the step-feed process was reduced by 20 30%, in practice to a level equal with the DN process. Consumption of aeration air in the dimensioning temperature (þ8 C) is now practically equal for all studied process options. The optimized step-feed configurations consume slightly less air than DN process, mostly due to their smaller aerated volume. In the average temperature (þ14 C), DN process still consumes least air, but a reduction of 10 14% has been achieved in step-feed process by optimization of the process design and operation. As a whole, both step-feed configurations are more economical in the dimensioning temperature than the DN process. In average conditions, DN is the most economical one, but the four-step process reaches almost the same air and methanol consumptions. Keeping in mind that due to a longer SRT, the step feed produces approximately 10% less excess sludge than the DN process, the studied processes can be deemed equally economical within the accuracy range of simulation results. Discussion Most of the literature published on the step-feed process concentrates on optimizing the process by means of influent division ratio and process dimensioning (e.g. Larrea et al. 2001, 2002; debarbadillo et al. 2002; Johnson et al. 2005; Tang et al. 2007; Zhu et al. 2007). Most sources state that basin volume is used more effectively with the step-feed process than with traditional activated sludge configurations. This opinion seems to be based primarily on theoretical analysis and pilot-scale tests. Obviously, it is difficult to arrange a full-scale comparison between different configurations. However, Johnson et al. (2005) present comparison data on parallel activated sludge lines, one of which was operated as a DN process and the other one as a step-feed process at the Rock Creek WWTP (Hillsboro, Oregon, USA). They operated step-feed with almost 50% higher load than DN process, with equal treatment results. However, treatment results were presented only by effluent ammonium nitrogen concentration, which was so low (,0.1 mgn/l) in both lines, that neither line cannot have been operating close to its maximum capacity. The possible over-capacity of the basins and other important parameters, which may have influence on the results, cannot be estimated based on the data
Table 3 Results of the simulations of Phase 2 (optimization of the step-feed configurations) Dimensioning 8 C Comparison 14 C Parameter Unit DN 3-step 3-step, optimized 4-step, optimized DN 3-step 3-step, optimized 4-step, optimized Volume of basin m 3 24 200 23 000 22 700 22 500 24 200 23 000 22 700 22 500 Division of wastewater 42:33:25 42:33:25 35:30:25:10 42:33:25 42:33:25 35:30:25:10 Division of methanol 33:33:33 70:15:15 70:15:15:0 33:33:33 45:30:25 45:30:25:0 Aerated of total % 71 67 65 66 48 50 48 47 SRT, total d 13.5 15 14.9 14.3 12.1 12.7 12.0 11.9 SRT, aerobic d 9.6 10.0 9.7 9.4 5.8 6.4 5.8 5.6 Methanol dosage mg/l 80 65 57 65 15 22 17 15 Air flow, average Nm 3 /h 11 940 13 020 11 600 11 500 10 360 12 380 11 100 10 700 Water Practice & Technology Vol 7 No 3
Figure 5 Residual S S at the end of anoxic zones in the three-step (right) and four-step (left) processes, T ¼þ8 C. presented. Johnson et al. (2005) give several examples of implementation of step-feed at different WWTPs, but only the one described above is a true comparison of parallel configurations. It also seems that comparison of DN and step-feed processes has not been conducted by simulation. debarbadillo et al. (2002) and Larrea et al (2001, 2002) present simulation results of the step-feed process but do not compare it to the DN process in terms of capacity or dimensioning. The total nitrogen reduction rates (75 85%) and effluent total nitrogen concentrations (5 10 mg/l) presented for the step-feed process in literature imply good nitrogen removal performance. However, these figures can be achieved also with the DN process. It must be noted that in debarbadillo et al. (2002), effluent ammonium nitrogen concentration was quite high (4 7 mg/l), i.e. nitrification was not complete, and in Tang et al. (2007) the BOD 5 :N TOT ratio was on the higher side (4 7), which facilitates high denitrification rates. If denitrification is limited by the availability of organic carbon, the step-feed process requires a gas removal compartment at the end of each step. This increases the total volume required by 3 5%. It must be noted that the alpha values applied in the aerobic zones have a large effect on calculated airflows. The alpha values varied from 0.4 to 0.6, increasing towards the end of the aeration basin. They were chosen so that their effect on each of the studied processes was the same, to the extent that this was possible. A constant specific oxygen transfer efficiency of 7.0%/m of basin depth was assumed for all processes. CONCLUSIONS DN process and step-feed configurations of three and four steps were compared with help of dynamic process simulation. The following conclusions are presented. Step-feed process required 5 10% less volume than DN process for treating an equal load in equal conditions. This is in contrast with literature values, which claim a difference of 20 25% to the favour of the step-feed process. The suggested reason for the small difference is that, contrary to intuitive assumption, the higher MLSS concentration at the first zones of the step-feed process is compensated by low concentrations of soluble substrates. Therefore, biokinetic conversion rates of substances are more or less equal to those of a DN process. The common opinion on reaching significantly more efficient use of basin volume with the step-feed process may be partly erroneous, due to rare and partly incomplete comparisons of parallel processes operated at full capacity and lack of comparative simulation studies. Consumption of aeration air and additional carbon source of denitrification (methanol) was practically equal for all studied processes in average temperature. In cold conditions, step-feed processes presented lower consumption rates than DN process.
Step-feed process could be operated with 10 15% higher SRT than DN process without increasing the solids load to secondary sedimentation. This is in accordance with literature. Economical operation of the step-feed process requires implementation of gas removal zones before each feeding point. In the step-feed process, DO has to be elevated from zero to setpoint level in each step, whereas in the DN process this has to be done only once, and all aerobic zones save the first one receive free oxygen from the preceding zones. This is reflected as higher air consumption in the step-feed process. Increasing the number of steps in the step-feed process from three to four did not have a significant effect on required volume and consumption of air and methanol. The division of methanol dosing in different steps of the step-feed process has a major effect on methanol consumption in cold conditions, when denitrification is limited by immediate retention time in the anoxic zones. In warmer conditions, when more anoxic volume is available, optimization of methanol dosing is less important. REFERENCES debarbadillo, C., Carrio, L., Mahoney, K., Anderson, J., Passarelli, N., Streett, F. & Abraham, K. 2002 Practical considerations for design of a stepfeed biological nutrient removal system. Florida Water Resources Journal January 2002, 18 35. Daigger, G. T. & Parker, D. S. 2000 Enhancing nitrification in North American activated sludge plants. Water Science and Technology 41 (9), 97 105. Johnson, B. R., Goodwin, S., Daigger, G. T. & Crawford, G. V. 2005 A comparison between the theory and reality of fullscale stepfeed nutrient removal systems. Water Science and Technology 52 (1011), 587 596. Larrea, L., Irizar, I. & Hidalgo, M. E. 2002 Improving the predictions of ASM2d through modelling in practice. Water Science and Technology 45 (6), 199 208. Larrea, L., Larrea, A., Ayesa, E., Rodrigo, J. C., Lopez-Carrasco, M. D. & Cortacans, J. A. 2001 Development and verification of design and operation criteria for the step feed process with nitrogen removal. Water Science and Technology 43 (1), 261 268. Phillips, H. M., Sahlstedt, K. E., Frank, K., Bratby, J., Brennan, W., Rogowski, S., Pier, D., Anderson, W., Mulas, M., Copp, J. B. & Shirodkar, N. 2009 Wastewater treatment modelling in practice: a collaborative discussion of the state of the art. Water Science and Technology 59 (4), 695 704. Tang, C. C., Kuo, J. & Weiss, J. S. 2007 Maximum nitrogen removal in the stepfeed activated sludge process. Water Environment Research 79 (4), 367 374. Zhu, G., Peng, Y., Wang, S., Wu, S. & Ma, B. 2007 Effect of influent flow rate distribution on the performance of stepfeed biological nitrogen removal process. Chemical Engineering Journal 131 (13), 319 328.