Shall we upgrade one-dimensional secondary settler models used in WWTP simulators? – An assessment of model structure uncertainty and its propagation

Abstract

In WWTP models, the accurate assessment of solids inventory in bioreactors equipped with solid-liquid separators, mostly described using one-dimensional (1-D) secondary settling tank (SST) models, is the most fundamental requirement of any calibration procedure. Scientific knowledge on characterising particulate organics in wastewater and on bacteria growth is well-established, whereas 1-D SST models and their impact on biomass concentration predictions are still poorly understood. A rigorous assessment of two 1-DSST models is thus presented: one based on hyperbolic (the widely used Takács-model) and one based on parabolic (the more recently presented Plósz-model) partial differential equations. The former model, using numerical approximation to yield realistic behaviour, is currently the most widely used by wastewater treatment process modellers. The latter is a convection-dispersion model that is solved in a numerically sound way. First, the explicit dispersion in the convection-dispersion model and the numerical dispersion for both SST models are calculated. Second, simulation results of effluent suspended solids concentration (XTSS,Eff), sludge recirculation stream (XTSS,RAS) and sludge blanket height (SBH) are used to demonstrate the distinct behaviour of the models. A thorough scenario analysis is carried out using SST feed flow rate, solids concentration, and overflow rate as degrees of freedom, spanning a broad loading spectrum. A comparison between the measurements and the simulation results demonstrates a considerably improved 1-D model realism using the convection-dispersion model in terms of SBH, XTSS,RAS and XTSS,Eff. Third, to assess the propagation of uncertainty derived from settler model structure to the biokinetic model, the impact of the SST model as sub-model in a plant-wide model on the general model performance is evaluated. A long-term simulation of a bulking event is conducted that spans temperature evolution throughout a summer/winter sequence. The model prediction in terms of nitrogen removal, solids inventory in the bioreactors and solids retention time as a function of the solids settling behaviour is investigated. It is found that the settler behaviour, simulated by the hyperbolic model, can introduce significant errors into the approximation of the solids retention time and thus solids inventory of the system. We demonstrate that these impacts can potentially cause deterioration of the predictive power of the biokinetic model, evidenced by an evaluation of the system's nitrogen removal efficiency. The convection-dispersion model exhibits superior behaviour, and the use of this type of model thus is highly recommended, especially bearing in mind future challenges, e.g., the explicit representation of uncertainty in WWTP models.

Practical identifiability and uncertainty analysis of the one-dimensional hindered-compression continuous settling model

“…The proper assessment of prior parameter uncertainties is significant for the subsequent analysis steps, but usually difficult and laborious. The hindered settling parameters (v 0 , r h ) are well reported in previous studies (Pl osz et al, 2011;Ramin et al, 2014a), while the compression parameters (C g , a, b) remain poorly understood. Table 2 gives the parameter uncertainties used in this study, which are reasonably estimated based on literature reviews and modeling experience.…”

A new settling velocity model to describe secondary sedimentation

“…To calibrate the selected settling velocity models (presented in section 3.1) to the settling measurements, they were implemented in a dynamic 1-D model of the settling column (Pl osz et al, , 2011. For calibration, the adaptive Markov Chain Monte Carlo (MCMC) Bayesian global optimization method DREAM (ZS) (Laloy and Vrugt, 2012) was used.…”

“…The thickening and clarification efficiency of SSTs is significantly influenced by hydraulic disturbances during wet-weather flow conditions. This is of particular concern due to the increasing frequency of hydraulic shock-loads from urban run-offs to WWTPs associated with the effect of climate change on peak rainfall intensities (Larsen et al, 2009;Pl osz et al, 2009). When examining SSTs performance, it is important to consider the non-Newtonian (rheological) and settling behaviour of activated sludge because they have significant impacts on the overall transport and removal of solids (Ekama et al, 1997;Lakehal et al, 1999;De Clercq, 2003).…”

“…Regarding the rheological models for activated sludge, Ratkovich et al (2013) were critical about the procedures currently used to select and calibrate them. They demonstrated the need for high quality measurements and stressed the importance of good modelling practice in rheological modelling.…”

Climate change and urban wastewater infrastructure: There is more to explore

“…Secondary clarifiers are mostly modeled using a 0-D approach (clarifier acts as ideal splitter of flow and solids) or a 1-D flux based approach, such as the 10 layer model proposed by Takács et al (1991), describing the hydrodynamic behavior in 1 dimension and its interaction with the flocs that are settling (Plósz et al, 2012). However, the Takacs model is also not capable of dealing with extreme events (Plósz et al, 2011). An alternative could be the consistent modeling methodology (CMM) introduced by Bürger et al (2011), although a combined model of conversion and clarification-thickening processes using the CMM in a secondary settling tank still needs to be developed.…”