Validation Of The Novel Settling Prediction Model Framework For Various Full-Scale Systems

INTRODUCTION
A clear trend towards intensified and resource-efficient technologies is observed as wastewater treatment plants are moving towards resource recovery facilities. Given that current secondary clarifiers are often the largest footprint bottleneck to achieve intensification, improving settleability, clarifier capacity, and effluent quality have become an essential feature of new technologies (Ekama et al., 1997). Recently, Ford et al. (2018) and Sturm & De Clippeleir (2020) described different degrees of settling improvements using the inDENSE technology based on a combination of physical (cyclones) and microbial selection (Bio-P and plug flow configurations). However, it was unclear how easily transferrable the latter observations were to other plants as the effects of operational conditions on settleability are challenging to estimate and extrapolate to new systems. Unlike the use of physical and microbial selection for shortcut nitrogen removal, where kinetic measurements or molecular tools help to quantify selection or selective wasting (Van Winckel et al., 2019), no clear quantification for settling selection is available, and thus transferability of results is challenging.
In previous work, the threshold of flocculation (TOF) (Mancell-Egala et al., 2017a), describing the collision efficiency of particles, was successfully introduced as a simple experimental method to calculate the flocculent settling coefficient (rp) in one-dimensional (1D) clarifier models and enhanced the ability to predict clarifier effluent solids (Ngo et al., 2021a; Figure 1). TOF also showed to have a clear relationship with effluent quality making it a useful operational parameter to detect flocculation limitations. Besides that, a strong correlation was found between the composition of extracellular polymeric substances (EPS) expressed as EPS-TKN/COD and TOF. The relationship between EPS composition and TOF was introduced as a means of linking biological process models (Sumo2C model) and 1D clarifier models (Ngo et al., 2021b; Figure 1). The latter model framework (Figure 1) shows the potential to predict settling behavior from operational conditions and would open up opportunities to extrapolate concepts from one facility to another. Preliminary validation using literature data showed settling trends related to solids retention time changes well. However, full validation using full-scale wastewater, process, and settling data was missing. In this study, we are filling this gap by detailed monitoring of three activated sludge systems at Blue Plains (two secondary contact stabilization systems, one biological nutrient removal system). In addition, sampling at five additional full-scale locations is initiated and will be added to the full paper to gain confidence that the proposed model framework is generally applicable. Two major validations are aimed for (1) validation of the two empirical equations (Figure 1) and (2) validation that process models can predict settling changes in full-scale systems based on operational conditions.
MATERIALS AND METHODS
Two high-rate secondary systems and one biological nutrient removal systems were samples for at least six months consisting of wastewater characteristics, flocculation settling, Vesilind parameters, and sludge composition including EPS. Similar measurement at five additional full-scale systems is initiated to generate similar data set for a broader range of operational conditions and systems. At least two of the full-scale systems include hydrocyclone selection. Empirical equations were validated directly from data (EPS~TOF) or using steady-state clarifier modeling (rp~TOF). Data gathered was then used to set up and calibrate the Sumo2C model (Hauduc et al., 2019) and validate predicted TOF and clarifier performance with the observed dataset. The latter action is ongoing and will be fully reported in the full paper.
RESULTS AND DISCUSSIONS
Validation of empirical equations A wide range of full-scale activated sludge systems was selected for validation to test the potential global applicability. Two years of validation data (2020 and 2021) from EAST and WEST secondary systems at Blue Plains were sampled (Figure 2 and Table 1). In addition, each secondary system was operated under different configurations including stepfeed (SF) and contact-stabilization mode (feast-famine) mode. Besides that, another five new full-scale WWTPs are sampled later this year to include larger variability in SRT, effluent quality, organic loading, and reactor configurations. Initial testing results are presented in figure 1B and 1D. Despite the significant differences in operational conditions of the systems, all validation data fell well within 95% prediction interval of the training data that was established based on stepfeed secondary system and BNR systems at Blue Plains systems (Figure 1). This confirmed that the developed equations are valid and transferrable. Process model-based prediction of settling behavior The overall model concept and an example of the model setup for the two secondary systems at Blue Plains are shown in Figure 1. This step included a detailed and longer-term sampling of full-scale systems, including detailed wastewater characterization (Table 1) to allow dynamic process models of the latter systems, in addition to the settling behavior and EPS (Figure 3). First, a steady-state model run will be aimed to get a sense of TOF prediction of the model, and this phase will be followed by dynamic modeling of the more than six-month period to see how well the model can pick up changes in settling over time. Both the stepfeed mode and contact stabilization mode will be simulated, which in practice resulted in improved settling behavior (Figure 2). The latter is a good example of operational condition change within one system. In addition to that, sampling campaigns within five additional full-scale systems are ongoing, which will provide a broader view of the applicability of the modeling framework. We believe that this work can be a gamechanger in approaching settleability and settling predictions.
CONCLUSION
This is the first time that process conditions were directly linked to settleability in a way model prediction of settling improvements based on changes in operational conditions are made possible. The power of the proposed settling prediction model is being validating this study based on eight full-scale systems. It brings us one step closer to a practical approach for enhancing the ability to predict clarifier effluent quality and performance as a result of operational and wastewater changes.