A model of reactive settling of activated sludge: comparison with experimental data

TL;DR

This paper develops a nonlinear PDE-based reactive settling model for activated sludge, incorporating hindered settling, compression, dispersion, and mixing, and validates it against pilot plant experimental data.

Contribution

It introduces a comprehensive reactive-settling model with a numerical scheme, fitting it to experimental data and improving predictions with dispersion modeling.

Findings

Model accurately simulates hindered settling and compression.

Fitting to pilot data validates the model's applicability.

Inclusion of dispersion improves steady-state predictions.

Abstract

A non-negligible part of the biological reactions in the activated sludge process for treatment of wastewater takes place in secondary settling tanks that follow biological reactors. It is therefore of interest to develop models of so-called reactive settling that describe the spatial variability of reaction rates caused by the variation of local concentration of biomass due to hindered settling and compression. A reactive-settling model described by a system of nonlinear partial differential equations and a numerical scheme are introduced for the simulation of hindered settling of flocculated particles, compression at high concentrations, dispersion of the flocculated particles in the suspension, dispersion of the dissolved substrates in the fluid, and the mixing that occurs near the feed inlet. The model is fitted to experiments from a pilot plant where the sedimentation tank has a varying cross-sectional area. For the reactions, a modified version of the activated sludge model no. 1 (ASM1) is used with standard coefficients. The constitutive functions for hindered settling and compression are adjusted to a series of conventional batch settling experiments after the initial induction period of turbulence and reflocculation has been transformed away. Further (but not substantial) improvements of prediction of experimental steady-state scenarios can be achieved by also fitting additional terms modelling hydrodynamic dispersion.