Energy consumption of agitators in activated sludge tanks – actual state and optimization potential

P_D

power density of agitator(s) (W m⁻³)

t_d

daily operating time of agitator(s)/mixing system (h d⁻¹)

W_mix,PE

inhabitant-specific energy consumption of agitator(s) (kWh PE₁₂₀⁻¹ a⁻¹)

W_mix,V

volume-specific energy consumption of agitator(s) (Wh m⁻³d⁻¹)

COD

chemical oxygen demand

MBWWT

mechanical–biological wastewater treatment

PE₁₂₀

population equivalent; assuming 120g COD PE⁻¹ d⁻¹

SVI

sludge volume index (mL g⁻¹)

TSS

total suspended solids (g L⁻¹)

WWTP

wastewater treatment plant

Energy consumption of wastewater treatment plants (WWTPs) amounts to a small part of the overall national energy requirements. In Germany and the USA the share is estimated to about 1% (Goldstein & Smith 2002; Haberkern et al. 2008). However, WWTPs are the largest energy consumer at the level of municipalities (Haberkern et al. 2008). For this reason and due to increasing energy prices the importance of energy optimization in municipal wastewater treatment has been constantly rising (EPA 2010; Krampe 2013; Marner et al. 2016). In this context, it has also become important to optimize agitator operation in activated sludge tanks (Sharma et al. 2011).

Mixing and aerating in activated sludge tanks are energy-intensive sub-processes within wastewater treatment. When aerating demands about 50–60% of the total energy consumption of a WWTP (Krampe 2011b), in this paper it is shown that mixing consumes about 5–20%. However, in contrast to aeration systems, agitators have not been standing in the central focus of energy optimization yet. Therefore, the aim of this paper is to fill this gap giving a sound overview regarding the actual state and the optimization potential of energy consumption of agitators (generally: mixing systems) in activated sludge tanks.

Data analysis in this paper is based on the evaluation of three different energy indicators: (A) inhabitant-specific energy consumption (kWh PE₁₂₀⁻¹ a⁻¹, where PE₁₂₀ is the population equivalent, assuming 120 g chemical oxygen demand (COD) per PE per day), (B) power density (W m⁻³) and (C) volume-specific energy consumption (Wh m⁻³ d⁻¹). As only little reference data was available in literature, the first step was to build up a broad and consistent agitator database of operating WWTPs from Austria.

The database for the study was collected by an Excel-sheet based survey throughout all 942 WWTPs >500 PE₁₂₀ in Austria in the year 2014. Three hundred and ninety-three questionnaires were sent back. Ten per cent of the Excel-sheets had to be excluded because of missing or obviously wrong data. Nine different indicators were built up validating the plausibility of design capacity, average incoming COD load, tank volumes, nominal power and energy consumption of agitators, etc. Only questionnaires that passed the outlier tests of all nine criteria were approved for the study. Finally, the data of 220 treatment plants, ranging from 520 to 950,000 PE₁₂₀ design capacity, could be used for evaluation. The study covers 30% of all Austrian WWTPs with a design capacity ≥ 2,000 PE₁₂₀ and 9% of all plants between 500 and 1,999 PE₁₂₀. However, the 220 evaluated WWTPs are corresponding to a design capacity of about 6 million PE₁₂₀, which is 28% of the total design capacity of Austria (Table 1). The scope of the database is comparable to Krampe (2011a).

The Excel-questionnaire was split into two parts. The first consisted of 13 questions (A–M) regarding the treatment plant as a whole, asking for design capacity, average incoming COD load, total volume of all activated sludge tanks, etc. (see appendix, available with the online version of this paper). The second part consisted of 23 questions (Bx-1 to Bx-23) regarding activated sludge tanks (see appendix). Tanks with the same volume, geometry, aeration type and mixing power were grouped and queried as one tank. So, many small treatment plants filled out the questionnaire for just one tank, whereas bigger plants in most cases filled out for two to four different tanks. Altogether 286 different activated sludge tanks were evaluated, consisting of 228 aerated and 58 unaerated tanks between 52 m³ and 15,000m³ (Table 2).

Apart from a few impulse mixing systems, the evaluation concentrated on the four following agitator types as categorized in DWA (2013):

• horizontal drive shaft: fast-running propeller (100–600 rpm, diameter (D) = 0.5–1 m)

• horizontal drive shaft: slow-running propeller (<100 rpm, D = 1–3 m)

• vertical drive shaft: hyperboloid-shaped mixer (<100 rpm, D = 1–2 m)

• vertical drive shaft: slow-running propeller (<100 rpm, D = 1–3m).

Further, detailed investigations of two WWTPs with low mixing energy consumption (WWTP Wartberg/Krems, WWTP Bad Aussee) were carried out. This included the verification of tank volume, agitator power consumption and energy indicators, measurements of near-ground velocity and tank deposits, recording of exact tank geometry and agitator positions, measurements of sludge volume index (SVI) and total suspended solids (TSS), etc. In addition, the low mixing energy data of WWTP Amperverband (Kopmann 2015: impulse aeration/impulse mixing) was evaluated for the study.

Data evaluation in this paper is based on three different energy indicators.

• W_mix,PE = kWh PE₁₂₀⁻¹ a⁻¹ inhabitant-specific energy consumption of agitators

• W_mix = kWh a⁻¹ annual energy consumption for mixing in activated sludge tanks of a WWTP

• F_COD = PE₁₂₀ yearly average of incoming COD load of the WWTP assuming 120 g COD PE⁻¹ d⁻¹

• P_D = W m⁻³ power density

• P = W power consumption of agitator(s)

• V = m⁻³ tank volume

• W_mix,V = Wh m⁻³ d⁻¹ volume-specific energy consumption

• P_D = W m⁻³ power density

• t_d = h d⁻¹ daily operating time of the mixing system

Due to the large number of specific local process configurations and conditions, it is not reasonable to establish energy benchmarks for mixing only. Nevertheless, as an approximation to benchmarks, so called ‘target values’ and ‘tolerance values’ can be derived from the frequency distributions of the energy indicators. In reference to Haberkern et al. (2008), in this paper they are defined as follows:

• target value = 25%-quantile

• tolerance value = median

If P_D of agitators strongly differs from target values, this may only be due to the agitator (efficiency, dimensioning). Yet in many cases, it is necessary to look more closely at the overall operating and constructional concept of the tank. In this regard, the interaction between tank geometry and P_D, mixing and aeration, as well as the optimal positioning of agitators, plays a key role (Frey 2007; Frey 2011a, 2011b; DWA 2013). For example, in oxidation ditches, the geometric shape of the deflection is a crucial factor regarding the necessary energy input (Figure 1). However, depending on the type of mixer, different reactor shapes can be optimal.

Depending on SVI and TSS, near-ground velocity in tanks should vary between 0.10 and 0.25m s⁻¹ (DWA 2013). The prevention of sludge deposits though does not depend on the average value of near-ground velocity but on ‘peak values and their frequency’ (Frey 2009). Therefore, agitators can be switched off partially or completely during aeration (Haberkern et al. 2008). Also in unaerated phases and tanks intermittent operation of agitators is an energy saving option (Baumann et al. 2014). However, there are some measures to be taken. Most important is the control of nitrate and sludge volume in the outlet of the tank. Further, it has to be ensured that the intermittent operation of agitators is shut down during N-peaks in the feed (Baumann et al. 2014). Other control parameters for the intermittent operation of agitators are the vertical layering of TSS and oxygen concentration (Sharma et al. 2011; Kopmann 2015).

Indicator (A), inhabitant-specific energy consumption W_mix,PE, was calculated for all 220 WWTPs. It showed that it is indirectly proportional to the average incoming COD load (expressed as PE₁₂₀ in Figure 2). So the larger the COD load, the smaller W_mix,PE. In accordance with Lindtner (2008) the target value ranges from 6.8 kWh PE₁₂₀⁻¹ a⁻¹ (<5,000 PE₁₂₀) to 1.3 kWh PE₁₂₀⁻¹ a⁻¹ (>30,000 PE₁₂₀) (Figure 2).

Further, the share of mixing energy within the total energy consumption was of interest. Therefore, the medians of W_mix,PE are compared to the medians of inhabitant-specific energy consumption of mechanical–biological wastewater treatment (MBWWT). MBWWT, a benchmarking term introduced by Lindtner et al. (2004), includes mixing, aeration, return sludge pumping, primary sedimentation tanks and final clarifiers (ÖWAV 2013). It is by far the most energy-intensive sub-process within wastewater treatment (Lindtner 2008; Haslinger et al. 2016). Comparing the medians, it can be shown that mixing demands a share of about 10 to 25% of MBWWT (Figure 2). Now taking into account that aeration consumes about 50–60% of the total energy consumption of a WWTP (Krampe 2011b), but about 70% of MBWWT (Lindtner 2008), it can be calculated that mixing demands a share of about 5 to 20% of the total energy consumption of a WWTP.

Finally, it can be shown that the share of mixing energy increases with decreasing COD loads. So, bigger treatment plants – with influent loads ≥35,000 PE₁₂₀ – need a share of about 10% of MBWWT, whereas smaller treatment plants – with influent loads up to 5,000 PE₁₂₀ – need about 25% of MBWWT (Figure 2); again corresponding to about 5–20% of the total energy consumption of a WWTP.

Indicator (B), power density P_D, was calculated for all 286 tanks. Both P_D and its target values and medians prove to be indirectly proportional to tank volume. Target values range from 5.7 W/m³ (≤200 m³) to 1.1 W/m³ (>2,000 m³), medians from 6.8 W/m³ (≤200 m³) to 1.6 W/m³ (>2,000 m³). Target values are within, and medians tend to be at, the upper limit of the target areas indicated by Baumann et al. (2014) (Figure 3).

The influence of agitator type (fast-running propeller; slow-running propeller; hyperboloid-shaped mixer) on P_D was examined. Fast-running propellers are only used for tank volumes <1,000 m³ and are operated with higher P_D than other agitator types. The target value for fast-running propellers was assessed to 5.2 W m⁻³. Between slow-running propellers (horizontal or vertical drive shaft) and hyperboloid-shaped mixers, no difference regarding P_D could be found. P_D target values for tanks <1,000 m³ range around 2.4 W m⁻³ each.

Tank deposits are an important control parameter regarding low mixing power densities. Data evaluation showed that P_D below the target values does not lead to more deposits than P_D above the target values (Table 3). This proves that the P_D target values elaborated in this paper can be put into praxis without harming sludge suspension.

Indicator (C), volume-specific energy consumption W_mix,V, has more than one main influencing parameter. Therefore, it cannot be evaluated like P_D and W_mix,PE. W_mix,V is mainly dependent on P_D and t_d. Due to the influence of P_D, there is also a slight dependency on tank volume (Figure 4). However, of bigger importance is the aeration. On average, W_mix,V in aerated tanks is twice as low as in unaerated tanks. In aerated tanks the average W_mix,V is about 50 Wh m⁻³ d⁻¹ (Figure 4(a)), in unaerated tanks about 100 Wh m⁻³ d⁻¹ (Figure 4(b)). This is mainly due to the mixing energy contribution of aeration. Consequently, t_d in most aerated tanks can be lower than in unaerated tanks, as there is energy input from aeration (see Table 5). In aerated tanks the spectrum of W_mix,V ranges from 14 to 119 Wh m⁻³ d⁻¹ (Figure 4(a)), in unaerated tanks from 43 to 242 Wh m⁻³ d⁻¹ (Figure 4(b)).

Activated sludge tanks >2,000 m³ have a P_D target value of about 1 W m⁻³ (Figure 3). At continuously mixed tanks (→t_d = 24 h d⁻¹) this corresponds to a low W_mix,V of 24 Wh m⁻³ d⁻¹. For the continuously mixed oxidation ditches at the WWTP Wartberg/Krems (Figure 4(a)) this could be confirmed by on-site measurements. Although a low W_mix,V of 22 Wh m⁻³ d⁻¹, near-ground velocities proved to be sufficiently high. The average values of nine different measurement points inside the oxidation ditch came to be 0.13–0.23 m s⁻¹; the maximum values came to be 0.16–0.30 m s⁻¹. In accordance with that, no significant deposits could be detected on the base of the tanks. An important reason for the good mixing performance of the oxidation ditches at WWTP Wartberg/Krems is to be found in the favourable agitator position and deflection geometry that equals the optimal geometry in Figure 1.

The oxidation ditches of WWTP Wartberg/Krems show that for aerated tanks a low W_mix,V of about 24 Wh m⁻³ d⁻¹ is feasible in practice under favourable conditions with respect to tank volume, tank geometry and agitator position. However, especially examples of impulse aeration show that W_mix,V in aerated tanks can be reduced to even below 24 Wh m⁻³ d⁻¹. This could be confirmed by on-site measurements at the rectangular 1,000 m³ impulse aeration tanks at the WWTP Bad Aussee (W_mix,V = 6 Wh m⁻³ d⁻¹) (Figure 4(a)). Again, no significant tank deposits could be detected.

As W_mix,V of unaerated tanks on average is twice as high as for aerated tanks, for unaerated tanks, an optimization goal of about 48 Wh m³ d⁻¹ can be approximated. This is also indicated by similar frequency distributions: 18% of aerated tanks fall below 24 Wh m⁻³ d⁻¹, 17% of unaerated tanks fall below 48 Wh m⁻³ d⁻¹. So both values are in the range of the 25%-quantile (defined as target value).

The optimization of energy consumption of agitators in activated sludge tanks can be divided into two subareas that can be analyzed separately over a wide range: (1) the reduction of P_D and (2) the reduction of t_d. In the best case, both P_D and t_d are optimized. However, the reduction of P_D, t_d or both P_D and t_d in all cases leads to a reduction of the W_mix,V and hence to mixing energy optimization.

Depending on whether the tolerance value (median) or the target value (25%-quantile) is recognized as an optimization goal, it can be assumed that approximately 50 to 75% of the tanks have an energetic optimization potential regarding P_D. Based on this consideration an estimation of the energy saving potential in regard to P_D was carried out for Austrian WWTPs. The 220 WWTPs of the study were taken into account. If P_D in all 286 tanks is reduced to target value about 40% of mixing energy in activated sludge tanks could be saved. With reduction of P_D to tolerance value, about 25% of mixing energy could be saved (Table 4).

The reduction of t_d (→ intermittent operation of agitators) is an energy saving option already being implemented in many pressure-aerated tanks. However, there are differences between different tank types. In oxidation ditches, the share of agitators being operated intermittently is 26%, in circular tanks it is 37%, in rectangular tanks 67%. This can be explained with different operating requirements; e.g. in oxidation ditches, propellers are needed to induce the direction of the water flow during aeration phases, in rectangular tanks not. Anyway, 33% continuously operated agitators also show an optimization potential at rectangular tanks (Table 5).

For unaerated tanks, there is no significant difference between tank types. In the case of oxidation ditches, circular tanks and rectangular tanks, approximately 85% of the agitators are being operated continuously. Nevertheless, about 15% of the tanks show that the continuous operation of agitators is no operating necessity (Table 5). This is consistent with references in DWA (2013) and Baumann et al. (2014).

It turns out that the range of energy saving options for tank mixing is located between two opposite poles (Figure 5). At one end of the spectrum there are tanks continuously mixed with low P_D (WWTP Wartberg/Krems: P_D = 0.92 W m⁻³). At the other end of the spectrum there are tanks mixed with short high pulses of energy. These energy pulses may either be introduced by air (WWTP Bad Aussee: P_D = 25 W m⁻³; WWTP Amperverband: P_D = 15 W m⁻³) or by mixing (WWTP Amperverband: P_D = 4.5 W m⁻³). The choice of the energy saving option largely depends on the specific operating conditions of the tank. As a consequence, intermittent operation of agitators with P_D < 1 W m⁻³ is not likely to be reasonable. On the other hand, impulse aerating and impulse mixing can be regarded as ‘extreme variants’ of intermittent mixing.

Agitators consume 5 to 20% of the total energy consumption of a WWTP. Up to 75% of the evaluated tanks have an optimization potential regarding energy consumption. This shows that mixing is an important issue in the energy optimization of WWTPs. The energy consumption indicators applied in this paper (W_mix,PE, P_D, W_mix,V) proved to be well applicable to both the calculation of the actual state and of the optimization potential of mixing energy. The first step for operators regarding optimization is thus to compare these indicators with the target values compiled in this paper.

The optimization itself can be split into two mostly independent strategies: the reduction of power density P_D and the reduction of daily operating time t_d. Ideally, both are optimized. However, this approach is limited by the volume-specific energy consumption W_mix,V. The following principle applies: the smaller t_d, the higher P_D on the respective volume-specific energy consumption isoline. Along this isoline, the range of energy saving options varies from continuous mixing with power densities as small as 1 W m⁻³ to mixing by short, intense energy impulses (impulse aeration, impulse mixing). When tank geometry, tank volume, agitator position and aeration are favourable, volume-specific mixing energy can be reduced to 24 Wh m⁻³ d⁻¹ based on available full-scale experiences, and even below.

Aeration shows a major influence on mixing energy consumption W_mix,V. Concerning the complex interdependencies between the energy consumption of mixing and aeration, further research, based on large-scale experiments and/or CFD simulations, is needed.