(134c) Optimization of a Full-Scale Disinfection Unit Using CFD at Pero WWTP in Milan

In view of coping with more stringent regulatory standards for secondary effluent disinfection, Gruppo CAP is using CFD (Computational Fluid Dynamics) to optimize the 2 disinfection units of one of their biggest wastewater treatment plants (WWTPs). The WWTP of Pero treats 30% of the municipal wastewater of Milan's metropolitan area (Italy), and therefore it is of extreme importance for the surface water quality. The aim of the study was to initially assess and then optimize the disinfectant dosage in the last treatment step of the Pero’s WWTP before discharging into the receiving water body. In this work, CFD was used to identify the most effective location(s) and strategy to cope with the operational variability of the WWTP and the regulatory guidelines. Within the collaboration with Gruppo CAP and AM-Team, CFD modelling was used to perform several virtual experiments at different operational scenarios avoiding onsite tests (i.e., risky, expensive, and time consuming). Some of the scenarios analysed in this project are reported in this abstract. The results highlighted the importance of the location of both the dosage and the mixing device, especially when updating an existing installation to new regulatory requirements (Pero WWTP E. Coli limit value changed form 5000 UFC/100 ml to 1000 UFC/100 ml in 2021 and it will be 500 UFC/100ml in 2022).

Methodological Approach

The geometry of the entire disinfection unit (Figure 1) was derived from the database of Gruppo CAP, and digitally recreated to obtain a 3D domain. There are 2 influent streams to the disinfection unit: the main one comes from an inlet channel (on the top of Figure 1), after the biological treatment. In case of heavy rain events, part of the influent of the WWTP bypasses the biological treatment, and enters the disinfection unit through the bypass channel. The disinfection unit is composed of 2 reaction serpentines; each of them consists of 2 parallel channels that run straight for about 65 m, before taking a 180° U-turn.

The CFD simulations were performed in ANSYS Fluent v21R1 in steady state. The flow and disinfectant transport were simulated by means of a RANS (Reynolds-Averaged Navier-Stokes) approach, specifically with a Realizable k-ε turbulence model, and standard wall functions (Versteeg and Malalasekera, 2007). The mixers were simulated with a Multiple Reference Frame approach (ANSYS, Inc. (2021)). The VOF multiphase model was used to calculate the water level in the plant.

Results and Discussion

In the normal plant operation, the influent coming from the main channel only enters the 1st disinfection serpentine (the gate that connects the 1st and 2nd serpentine is closed), as shown in Figure 2. In the initial configuration the disinfectant dosage point is located on the corner of the basin before entering the serpentine, as shown in the digital recreation of the geometry (Figure 3). For this scenario, an average flowrate of 5000 m3/h (normal conditions) was considered. The results of the simulation of this scenario are shown in Figure 4. The water coming from the inlet channel is accelerated by the gate that leads to the basin with the mixer, before entering the 2 channels of the serpentine. Note that the dosage point is located on a stagnation zone on the side of the basin (the mixer is not able to create full mixing in the basin). This negatively affects the disinfectant distribution. Referring to Figure 4, the variable used to assess the disinfectant distribution relates the local disinfectant fraction to the desired one (if >1, there is excess of disinfectant, while if <1 there is lack of disinfectant). As a consequence of the flow pattern, the disinfectant accumulates on the side of the basin, and mainly enters only one of the two channels (Channel 1). Specifically:

• Channel 1: Local fraction / Desired fraction = 1.8
• Channel 2: Local fraction / Desired fraction = 0.14

To conclude, with this configuration the dosing strategy is not optimal.

Therefore, an improved configuration was suggested and tested, shown in Figure 5. The dosing point was moved in the middle of the inlet channel, with a mixer placed afterwards. This scenario was tested with a high influent flowrate (9000 m3/h). The results of this scenario are shown in Figure 6. The disinfectant plume is broken by the action of the mixer placed after the dosage point. As a result, a good disinfectant distribution is observed even before entering the basin that leads to the serpentine. Therefore, the new configuration proved to bring a substantial improvement in terms of disinfectant usage, as it is well mixed and equally distributed in the channels of the serpentine. However, a potential issue was detected by analysing the hydraulics: the water flowrate coming from the inlet channel is not equally split between the two channels of the serpentine. Despite the action of the impeller in the basin, the path from the turn to Channel 1 is in fact more favourable. As a result, at 9000 m3/h, 70% of the total flowrate enters Channel 1. Such splitting inequality increases as the influent flowrate increases (at 5000 m3/h the splitting was close to 50% - 50%). As a conclusion, at high influent flowrates, the HRT in the serpentine might not be sufficient, because of the high velocity in one of the two channels.

The effect of both disinfection lines (Serpentine 1 and 2) operating together was analysed in the following scenario. In such plant configuration, both the influent coming from the main channel and the bypass stream enter the disinfection unit. The geometry used for this simulation is shown in Figure 7. All gates connecting the serpentines are open, so there are no constraints for the flow. The weirs in the end of both serpentines are included in the model, as they are crucial to determine the water level. For practical reasons, the whole serpentine lengths were not modelled; the weirs were instead placed in the 1st leg of the serpentines. This scenario was modelled under the following conditions: 7000 m3/h as the influent flowrate from the main channel, and 1000 m3/h from the bypass channel. The results are shown in Figure 8. Thanks to on-site measurements, a 28 cm difference in the weirs height between the two serpentines was detected (the weir at the end of Serpentine 2 is 28 cm lower than the one of Serpentine 1). As a consequence, the path through Serpentine 2 is favoured. As shown in Figure 8, the difference in the weir height is such that the total flowrate (8000 m3/h) only enters Serpentine 2. In other words, the overflow height over the weir of Serpentine 2 is lower than the height of the weir of Serpentine 1. As a consequence, stagnation is observed in Serpentine 1. To conclude, if both disinfection lines are operating together, a whole reaction serpentine would be unused.

Conclusions

A full-scale disinfection unit was analysed by CFD modelling. The study allowed to analyse the current state of the plant and, based on that, to spot opportunities to improve it. Specifically, with the original dosage point in the first disinfection line, the disinfectant distribution in the reaction serpentine is not homogeneous, and so the disinfectant is not being optimally used. Thanks to this study, a new dosage location was identified, allowing to save a significant amount of disinfectant (≈7 times less). It was also discovered that in the first disinfection line, at high flowrates, the splitting of the total flowrate into the serpentine channels is not equal; this might lead to HRT issues. Also, when both disinfection lines are operating together, the whole influent flow only enters Serpentine 2. This work helped Gruppo CAP to optimally use the disinfectant and to reduce operational costs.

References

ANSYS, Inc. (2021) ANSYS Fluent Theory Guide

Versteeg, H. K. and Malalasekera, W. (2007) An Introduction to Computational Fluid Dynamics: The Finite Volume Method. Harlow, England: Pearson Education.