A control strategy for reducing aeration costs during low loading periods

C. Sahlmann, J.A. Libra, A. Schuchardt, U. Wiesmann and R. Gnirss*

Institute of Chemical Engineering, Technical University Berlin, Sekr. MA 5-7, Str. 17. Juni 135, 10623 Berlin, Germany (E-mail: ch.sahlmann@gmx.de; judy.libra@tu-berlin.de) * Berliner Wasserbetriebe, Postfach 02 10 98, 10122 Berlin, Germany Abstract The efciency of the aeration system in a full-scale activated sludge basin with 3 separately controlled aeration zones was improved for the low loading period in summer. The air ow rate to each aeration zone is currently regulated to hold a preset dissolved oxygen concentration (DO). Four different DO setpoint combinations were tested, each one for a one week period, using dynamic off-gas testing to measure the standardised oxygen transfer efciency (SOTE). As the DO setpoints were lowered, the total air ow rate to the basin decreased initially. A low DO in the rst zones slowed biomass activity and pushed the load towards the end of the aeration basin. The relationship between SOTE and the specic diffuser ow rate qD is different for each zone. In Zone 1 there was a strong decrease in SOTE as qD increased, while Zones 2 and 3 were fairly independent of qD, Zone 2 at a higher level than Zone 3. Aeration costs were reduced by 15% for the most efcient combination. To achieve even more savings, a control strategy adjusting oxygen transfer rates over the aeration basin to the necessary oxygen transfer rates is suggested. It is based on changing the DO setpoints to reach the lowest total air ow rate while meeting the efuent requirements. Keywords Activated sludge; aeration efciency; control strategy; off-gas testing; oxygen transfer efciency

Water Science and Technology Vol 50 No 7 pp 6168 IWA Publishing 2004

Introduction

Wastewater treatment plants (WWTP) are designed for a worst-case-scenario, which for temperate Germany means a low temperature (12C) with a high load, in order to comply with the demands placed on effluent quality under all conditions. This leads unavoidably to situations in which the treatment plant is not fully loaded. Current trends towards higher demands on the discharge quality, especially nutrient removal, and rising energy costs increase the incentive to develop energy conservation strategies in WWTP. Considerable savings can be achieved through the optimisation of the aeration system, one of the largest energy consumers in an activated sludge WWTP. The process control scheme used in most Berlin activated sludge basins is to adjust the air flow rate to hold a constant dissolved oxygen concentration (DO) setpoint of 2 mg/L, so that the air flow rate is reduced automatically at low influent conditions. Additional savings may be achieved by adjusting the DO setpoint over the day to optimise the aeration efficiency, especially at low loading conditions when the removal processes are essentially completed in the first part of the basin. The benefit of decreasing the DO setpoint during these periods is twofold. Firstly, the driving force for mass transfer at lower DO is better. For example, lowering the DO from 2 to 1 mg/L reduces the amount of air needed to transfer the same amount of oxygen into the wastewater by 12%. Secondly, the biological processes are slowed down, especially nitrification, so that the entire aeration basin can be used for removal processes instead of for costly aerobic sludge stabilisation. However, an understanding of the oxygen transfer efficiency in the basin as a function of operating parameters is needed to develop a successful control strategy. This paper presents an overview of the oxygen transfer and energy savings that were achieved in one of

61

Berlins largest WWTP by adjusting the DO setpoint to process conditions in summer and discusses control strategies using flexible setpoint adjustments to increase savings.
Materials and methods

The investigations were carried out in the WWTP Berlin-Wamannsdorf, treating 150,000 m3/d wastewater (dry weather) with biological nutrient removal. Each basin, designed for 30,000 m3/d, has an anaerobic stage for biological phosphorus removal, an anoxic stage for denitrification and an aerated stage with three aeration zones ~112 m long (Figure 1). The number of diffusers per zone decreases with the degree of wastewater treatment, however, the diffusers are distributed equally over the length in each zone. The ceramic tube diffusers (L = 0.75 m, D = 0.07 m) have a minimum specific diffuser flow rate (qD) of 2 Nm3/h m to avoid fouling. Details of the aeration basin, configuration and size as well as typical wastewater concentrations are listed in Table 1 and Figure 1. The air flow rate to each zone is metered and controlled over the DO measured in the middle of each zone. Four different combinations of the DO setpoint in each aeration zone were investigated: (Z1/Z2/Z3) 2.0/2.0/2.0 mg/L, 1.5/1.5/1.0 mg/L; 1.2/1.2/1.5 mg/L and 0.8/1.0/1.2 mg/L. To evaluate the effect of these four different DO setpoint combinations on the aeration efficiency and treatment efficiency, the oxygen transfer efficiency (OTE) and air flow rate (QG) were measured for each setpoint combination over one week of operation. Three comparable dry weather days were chosen and a 24 average was built. The effluent quality met the discharge requirements. The changes in oxygen transfer as a function of operating conditions were measured with the dynamic off-gas method developed at the Technical University Berlin (Libra et al.,

62

C. Sahlmann et al.

Table 1 Technical and operating data for the activated sludge basins
Aeration stage Zone 1 Zone 2 Zone 3 Average

Length Depth (water) Diffuser submergence Distribution of diffusers between Zones Diffuser density (diff. length/ basin area)
Inuent and efuent concentrations (averages for 2000) (BWB, 2002)

112 m 5.1 m 4.8 m 51 % 1.28 m/m2

112 m 5.1 m 4.8 m 29 % 0.74 m/m2

Inuent

102 m 5.1 m 4.8 m 20 % 0.54 m/m2

Efuent

0.88 m/m2
Discharge standards

COD (unltered) BOD5 (unltered) NH4-N N inorg. N total PT (unltered)

mg/ L mg/ L mg/ L mg/ L mg/ L mg/ L

903 375 49.3 69.2 10.1

45 4 0.4 9.8 12.1 0.36

55 15 5 15 0.5

influent

online: NH4-N, NO3-N

Z 1 51 % diffusers Z 2 29 % diffusers Z 3 20 % diffusers

DO

online: NH4-N, NO3-N, PO 4-P

anaerobic stage

anoxic stage

aerobic stage

Figure 1 Activated sludge basin overview with sample points and stages

2002). This means there were no restrictions placed on the operating conditions of the WWTP. The off-gas was collected in each zone with a floating aluminium hood and connected to online gas analysers (O2, CO2) through tubing. The hoods were positioned midzone near the online DO probes; testing over 2% of the basin area showed these positions to be representative. The set-up and calculations followed the ASCE Standard (1997) and ATV M209 (1997). With the online analyser, it was possible to calculate the oxygen uptake rate in every zone at least once an hour.
Results and discussion

C. Sahlmann et al.

In Figure 2 the typical behaviour of air flow rates in Zone 1 for a dry weather day can be seen for the four DO setpoint combinations. Compared to the other zones variation during the day is low. A higher DO setpoint causes a higher air flow rate at all times over the day. Lowering the DO setpoint improved the oxygen transfer efficiency (OTE) in the zone because the driving force for mass transfer increases as DO decreases. Furthermore as the air flow rate per m diffuser qD is lowered aOTE increases (Figure 5). However, the total amount of oxygen transferred in Zone 1 decreased due to lower biomass activity caused by the oxygen limitation (see Table 2). Figure 3 shows the different behaviour of the air flow rate during night and day in Zone 2. Between noon and midnight, QG in Zone 2 is fairly constant, similar to Zone 1, and a
(2.0/2.0/2.0) (1.5/1.5/1.0) (1.2/1.2/1.5) (0.8/1.0/1.2)

7000
air flow rate Q G [Nm3/h]

6000 5000 4000 3000 2000 1000 0 0:00

6:00

12:00

18:00

0:00

Figure 2 Air ow rates (2 h-average for 3 dry weather days) over the day in Zone 1

(2.0/2.0/2.0)

(1.5/1.5/1.0)

(1.2/1.2/1.5)

(0.8/1.0/1.2)

7000

air flow rate Q G [Nm3/h]

6000 5000 4000 3000 2000 1000 0 0:00

6:00

12:00

18:00

0:00
63

Figure 3 Air ow rates (2 h-average for 3 dry weather days) over the day in Zone 2

higher DO setpoint causes higher air flow rates because of the same reasons as above. But during the low loading period at night, the minimum air flow rate (set to prevent flooding of the ceramic diffusers) is reached. This occurs faster when the DO in Zone 1 is high. Only for the combination with the lowest DO setpoints (0.8/1.0/1.2 mg/L) is the load pushed into Zone 2 at night. Figure 4 shows the air flow rates during the day in Zone 3 and demonstrates the relationship between the zones. The DO setpoint in Zone 3 no longer has a notable impact on the air flow rate, it is more what has happened in Zone 1 and 2 that determines the load to be removed in Zone 3 and therefore the air flow rate needed there. For the DO setpoint combination 2.0/2.0/2.0, the air flow rate was at the minimum most of the time. Only during the high loading period at midday did the air flow rate increase slightly. Limiting the DO in the first two zones to 1.5 mg/L did not push load into Zone 3. However, when the DO setpoints in Zone 1 and 2 were reduced further, more air for a longer period was needed in Zone 3, reaching the point that for the DO setpoint combination 0.8/1.0/1.2 more air was needed in Zone 3 than in Zone 2 at midday. Because the diffuser densities are different in each zone, i.e. 1.28 m/m2 in Zone 1 vs. 0.74 m/m2 in Zone 2 or 0.54 m/m2 in Zone 3, the relationship between SOTE and the air flow rate per m diffuser or specific diffuser flow rate qD is different in each zone (Figure 5). Although the range of qD was the smallest in Zone 1, SOTE changes the most there. SOTE decreased strongly as qD increased, while in Zones 2 and 3, SOTE was fairly independent of qD, Zone 2 at a higher level than Zone 3. The lower the setpoint in the first zone was, the more air was saved here because of a higher SOTE at lower qD. In Zones 2 and 3 the wide variation in qD reflects the rate of the removal processes taking place, i.e. low air flow rates with aerobic sludge stabilisation and high air flow rates with nitrification. Since Zone 2 shows the best performance, increasing the amount of air distributed here is the key for reducing the total air flow rate. The present control strategy with a DO-setpoint combination held constant over the day does not consider the different relationship between SOTE and air flow rate in each zone. Air is wasted because the strategy is not able to react adequately to low influent conditions. This can be illustrated using Figure 6, which shows the oxygen transfer rate for each zone as a function of the total air flow rate for the DO setpoint combination 1.2/1.2/1.5 mg/L. With the current strategy, the oxygen transfer rate (OTR) in only one zone is affected by changes in the oxygen demand in the aeration basin (due to increased wastewater flow rates or concentrations). For low total air flow rates (QG) OTR in Zone 1 changes, Zones 2 and 3 are constant at a low level because of the required minimum qD. After reaching a constant

air flow rate Q G [Nm3/h]

64

C. Sahlmann et al.

(2.0/2.0/2.0) 7000 6000 5000 4000 3000 2000 1000 0

(1.5/1.5/1.0)

(1.2/1.2/1.5)

(0.8/1.0/1.2)

0:00

6:00

12:00

18:00

0:00

Figure 4 Air ow rates (2 h-average for 3 dry weather days) over the day in Zone 3

standard oxygen transfer efficiency SOTE [%]

30

25
Zone 2

20

C. Sahlmann et al.

15
Zone 1
Zone 1

Zone 3

10

Zone 2 Zone 3

0 0 2 4 6 8 specific diffuser flow rate q D [Nm3/h m] 10 12 14

Figure 5 SOTE in each aeration zone as a function of the specic diffuser ow rate qD for a one week period (DO setpoints Z1/Z2/Z3 1.2/1.2/1.5 mg/L)
300

oxygen transfer rate [kgO2 /h]

Zone 1

Zone 2

Zone 3

200

150

100

50

0 0 2000 4000 6000 8000

3

minmum flowrate (4600 Nm3/h)

250

10000

12000

14000

total air flow rate [Nm /h]

Figure 6 OTR in each aeration zone as a function of the total air ow rate for dry weather conditions (DO setpoints Z1/Z2/Z3 1.2/1.2/1.5 mg/L)

value in Zone 1, OTR in Zone 2 increases with increasing QG. As oxygen demand increases, only OTR in Zone 3 is modified by an increase in air flow rate to Zone 3. No changes in air flow rate to Zones 1 and 2 are made, so that the higher OTE found in Zone 2 is not exploited. Much better would be a strategy with variable DO-setpoints to distribute the oxygen transfer more efficiently over the entire basin based on the relationship between OTE and the air flow rate in each zone. The performance is in general the same for the other DO setpoint combinations except that different values of OTR are reached at different air flow rates. Table 2 lists the maximum OTR and QG in each zone and in total for dry weather conditions. In Zone 3 the maximum is not reached for dry weather conditions, so the maximum measured is listed. During high loading conditions for dry weather the maximum OTR in the whole basin could be lowered from 680 to 530 kgO2/h without affecting the effluent quality. This showed that approximately 20% of the oxygen transferred at the higher DO setpoint was used for unnecessary aerobic sludge stabilisation. Surprisingly, the combination with the lowest DO setpoints did not achieve the lowest air flow rates. Too much load was pushed in Zone 3, which has a lower SOTE than Zone 2 (see Figure 5), so that at some point the total air flow rate

65

Table 2 Maximum OTR and QG in each zone for the different DO setpoint combinations under dry weather conditions
DO setpoints (Z1/Z2/Z3) [mg/L] OTR [kgO2/h] Zone 1 QG [Nm3/h] OTR [kgO2/h] Zone 2 QG [Nm3/h] OTR [kgO2/h] Zone 3 QG [Nm3/h] OTR [kgO2/h] Total QG [Nm3/h]

2.0/2.0/2.0 1.5/1.5/1.0 1.2/1.2/1.5 0.8/1.0/1.2

280 250 200 180

6,500 6,000 4,500 4,100

280 250 200 180

5,500 5,000 3,500 3,500

120 100 140 170

2,500 2,000 3,000 3,800

680 600 540 530

14,500 13,000 11,000 11,400

air flow rate Q G [Nm3/d]

66

C. Sahlmann et al.

increased. During low loading conditions at night, however, the combination with the lowest DO setpoints was the most efficient. The evaluation of four different DO setpoint combinations at the WWTP Berlin Wamannsdorf shows that the removal processes can be distributed over the aeration zones by decreasing the DO setpoints. The lower DO in the first zone slowed the oxygen transfer rate and pushed the load towards the end of the aeration basin. This translated into lower total air flow rates of up to 15% as the air flow rate was redistributed between the zones (a decrease in Zones 1 and 2 and an increase in Zone 3) and oxygen was transferred more efficiently. The most efficient combination based on the daily air flow rate was 1.2/1.2/1.5 mg/L (Figure 7). This result reflects the overall performance over the day. As already shown before, the DO setpoint combination 1.2/1.2/1.5 is not the most effective at all times. Especially for low loading situations during the night, the control strategy is not well adapted. This can be seen by comparing the calculated oxygen demand for nitrification OD(N) and the amount of oxygen transferred to the whole basin OTR (Figure 8). The portion of oxygen transferred used for nitrification %OT(N) shows large variation over the day, from 30% at night to 90% at high loading. At night the oxygen needed for nitrification is transferred in Zone 1. Because of the minimum qD Zones 2 and 3 are over aerated. DO increases up to 4 mg/L and oxygen is unnecessarily used for aerobic sludge stabilisation. To achieve even more savings, an automated strategy that can react to the varying loads over the day and the transfer efficiency in each zone is necessary. The goal is to change air flow rates simultaneously in all three zones instead of one after another as shown in Figure 6. The analysis of the dynamics during the day leads to a control strategy based on changing the DO setpoints to reach the lowest total air flow rate while meeting the effluent

300000 100 % 250000

29500

-6 % zone 3
28000

-15 %

-13 %

200000
43000 80000 55000

150000 100000
132500

zone 2

77000 64000 69000

50000 0 2.0/2.0/2.0

zone 1

122000 99000 87000

1.5/1.5/1.0

1.2/1.2/1.5

0.8/1.0/1.2

DO-setpoints (Z1/Z2/Z3) [mg/L]

Figure 7 Inuence of the four different combinations of DO setpoints on the total daily air ow rate into the three aeration zones (setpoints constant over a week, dry weather days)

1000

100

oxygen transfer rate [kgO2/h]

600

60

portion for nitrification [%]

800

80

C. Sahlmann et al.

400

40

200 OTR 29.7.01 30.7.01 31.7.01 OD(N) 1.8.01 2.8.01 % OT(N) 3.8.01 4.8.01

20

0 28.7.01

0 5.8.01

Figure 8 Oxygen demand for nitrication OD(N), total oxygen transferred OTR and the ratio between them % OT(N) over a one week period for the DO setpoint combination 1.2/1.2/1.5 mg/L

requirements. For WWTP Berlin Wamannsdorf a combined feed forward/feed back strategy working in three steps is proposed: 1. Measure SOTE for each zone as a function of qD. The current transfer efficiency has to be measured in each zone, and the distribution of the SOTE and air flow rate QG over the basin as well as over the day investigated. After calculating SOTE as a function of qD, the necessary air flow rate in each zone to obtain a certain oxygen transfer rate can be predicted. 2. Estimate the total oxygen amount necessary for the influent load. To get a quick and reliable prediction of the oxygen needed in the aeration basin, NH4-load and wastewater flow rate to the aerated stage, calculated using the online instruments already installed, as well as estimates based on off-gas measurements could be used (as seen in Figure 8). 3. Optimise the air flow rate distribution over the zones based on the function measured in step 1 to achieve the necessary oxygen input. The lowest possible total airflow to achieve the necessary oxygen transfer is calculated using the present relationship between air flow rate and oxygen transfer rate in each zone. This distribution of air into the zones can achieve nitrification in all zones and avoid aerobic sludge stabilisation and over-aeration in parts of the basin.
Conclusions

Aeration costs were reduced by 15% for the low loading conditions of summer by adjusting oxygen transfer rates over the aeration basin to control nitrification rates through manual changes in the DO-setpoint. Potential for further savings exists, especially during the low loading period at night, when the nitrification rate was still so high that the process finished midway in the aeration basin, wasting air for aerobic sludge stabilisation in the second half. A new strategy, which can adjust oxygen transfer rates over the aeration basin to the necessary nitrification rates, is suggested. The strategy is based on changing the DO setpoints to reach the lowest total air flow rate while meeting the effluent requirements. This operation strategy will be tested soon, with the goal of adapting it to other waste water treatment plants. Besides reducing energy consumption it will improve the reliability of the process by allowing a clearer assessment of the oxidation processes over the basin.
References
Libra, J.A., Schuchardt, A., Sahlmann, C., Handschag, J., Wiesmann, U. and Gnirss, R. (2002). Comparision
67

of the efficiency of large scale ceramic and membrane aeration systems with the dynamic off-gas method. Wat. Sci. Tech., 46(45), 317324. Messung der Sauerstoffzufuhr von Belftungseinrichtungen in Belebungsanlagen in Reinwasser und in belebtem Schlamm (1997). ATV Arbeitsblatt M 209, ATV, Hennef. Standard guidelines for in-process oxygen transfer testing (1997). ASCE 1896, ASCE American Society of Civil Engineers, New York.
C. Sahlmann et al.
68