Modelling of secondary sedimentation under wet-weather and filamentous bulking conditions

Secondary settling tanks (SSTs) are the most hydraulically sensitive unit operations in wastewater treatment plants (WWTPs). Performance of SSTs influences the solids inventory in the activated sludge unit and consequently impacts the biological treatment efficiency. Moreover, SSTs limit the maximum permissible flow rate entering the WWTPs during wet-weather conditions. Therefore, modelling the dynamics in the SSTs is an essential part of integrated sewer- WWTP modelling for the purpose of optimization and control, specifically under wet-weather conditions. One-dimensional (1-D) SST models with first-order type equations are widely used among researches and practitioners for dynamic WWTP simulations. Several drawbacks of the first-order models have however been reported in the literature, which have led to the development of more advanced second-order 1-D SST models. Unfortunately, the second-order models have not yet found their way into practice. This thesis aims at encouraging a broader application of secondorder 1-D SST models by assessing their significance for WWTP modelling by means of global sensitivity analysis (GSA). Moreover, laboratory and numerical (computational fluid dynamics, CFD) tools were developed for the identification and calibration of the settling sub-model in the SST models. The developed CFD tool is a potential tool for the development of a more mechanistic based flow (and design) dependent hydraulic sub-model in the second-order 1-D SST. In this thesis, a rigorous comparative evaluation of the first- and second-order SST models in WWTP modelling was performed by means of GSA. In the first GSA study using the Benchmark Simulation Model No. 2 with first- and second-order SST models, the settling parameters were included in the sensitivity analysis. Interestingly, the settling parameters were found to be among the most influential parameters for predicting the WWTP performance in terms of biogas production and quality of treated water. Importantly, it was observed that the choice of 1-D SST type model influences the sensitivity
measures of the parameters and consequently result in different parameter sub-sets for the calibration of WWTP models. Furthermore, the limitations of first-order SST models with relevance to the calibration of WWTP models were discussed. In the second GSA study of this thesis, the aim was to supplement the protocol recently published by the International Water Association on good modelling practice for activated sludge systems with practical findings on the calibration of 1-D SST models for dynamic WWTP simulations under ideal and non-ideal flow (dry- or wet-weather) and settling (good settling and/or bulking) boundary conditions. The Benchmark simulation model No. 1 in combination with first- and second-order 1-D SST models was used. An assessment was performed on the sensitivity of WWTP model outputs to uncertainties intrinsic to 1-D SST model structures and parameters under different boundary conditions imposed to WWTP simulation models. Further, the relative sensitivity to these uncertainties indicated potential parameter subsets for WWTP model calibration and the optimal choice of 1-D settling model structure under the different boundary conditions. Importantly, the hydraulic parameters in the second-order 1-D SST model were found significant under dynamic wet-weather flow conditions. The results of this study illustrated the advantages of second-order 1-D SST models for dynamics WWTP simulations under wider flow and bulking conditions, and furthermore, highlighted the necessity of developing a more mechanistic based flow-dependent hydraulic sub-model in second-order 1-D SST models in the future. A significant part of the thesis was dedicated to the development of a CFD model of a circular conical SST with the open source OpenFOAM CFD toolbox. The focus was mainly on identifying the settling and rheology submodels using data obtained from laboratory batch experiments. A simple, novel settling column experimental set-up was developed to evaluate the accuracy of the state-of-the-art settling velocity models. For calibration the Bayesian optimization method DREAM(ZS) was used. Consequently, a new settling velocity model, including hindered, transient and compression settling, was developed. In addition, a rheology model of activated sludge was selected and calibrated to high quality rheological measurements from the optimized batch experiments. New correlations between rheology model parameters and sludge concentration were identified. A 2-D axisymmetric CFD model of a circular SST containing the new settling velocity and rheology sub-models were validated with full-scale measurements. Finally, it was shown that the representation of compression settling in the CFD model greatly influences the prediction of sludge distribution in the SSTs. The validated CFD model was further used in the last study of this thesis to model the impact of filamentous bulking on the sludge distribution and transport in SSTs by calibrating the rheology and settling sub-models to measurements of
sludge with high and low filamentous bacteria content.