Treatment of the Supernatant of Anaerobically Digested Organic Fraction of Municipal Solid Waste in a Demo-Scale Mesophilic External Anaerobic Membrane Bioreactor

Abstract

Conventional aerobic biological treatments of digested organic fraction of municipal solid waste (OFMSW) slurries–usually conventional activated sludge or aerobic membrane bioreactor (AeMBR)–are inefficient in terms of energy and economically costly because of the high aeration requirements and the high amount of produced sludge. In this study, the supernatant obtained after the anaerobic digestion of OFMSW was treated in a mesophilic demo-scale anaerobic membrane bioreactor (AnMBR) at cross flow velocities (CFVs) between 1 and 3.5 m⋅s–1. The aim was to determine the process performance of the system with an external ultrafiltration unit, in terms of organic matter removal and sludge filterability. In previous anaerobic continuous stirred tank reactor (CSTR) tests, without ultrafiltration, specific gas production between 40 and 83 NL CH4⋅kg–1 chemical oxygen demand (COD) fed and removals in the range of 10–20% total COD (tCOD) or 59–77% soluble COD (sCOD) were obtained, for organic loading rates (OLR) between 1.7 and 4.4 kg COD⋅m–3reactor d–1. Data helped to identify a simplified model with the aim of understanding and expressing the process performance. Methane content in biogas was in the range of 74–77% v:v. In the AnMBR configuration, the COD removal has been in the ranges of 15.6–38.5 and 61.3–70.4% for total and sCOD, respectively, with a positive correlation between solids retention time (SRT, ranging from 7.3 to 24.3 days) and tCOD removal. The constant used in the model expressing inhibition, attributable to the high nitrogen content (3.6 ± 1.0 g N-NH4+⋅L–1), indicated that this inhibition decreased when SRT increased, explaining values measured for volatile fatty acids concentration, which decreased when SRT increased and OLR, measured per unit of volatile suspended solids in the reactor, decreased. The alkalinity was high enough to allow a stable process throughout the experiments. Constant CFV operation resulted in excessive fouling and sudden trans-membrane pressure (TMP) increases. Nevertheless, an ultrafiltration regime based on alternation of CFV (20 min with a certain CFVi and then 5 min at CFVi + 1 m⋅s–1) allowed the membranes to filter at a flux (standardized at 20°C temperature) ranging from 2.8 to 7.3 L⋅m–2⋅h–1, over 331 days of operation, even at very high suspended solids concentrations (>30 g total suspended solids⋅L–1) in the reactor sludge. This flux range confirms that fouling is the main issue that can limit the spread of AnMBR potential for the studied stream. No clear correlation was found between CFV or SRT vs. fouling rate, in terms of either TMP⋅time–1 or permeability⋅time–1. As part of the demo-scale study, other operational limitations were observed: irreversible fouling, scaling (in the form of struvite deposition), ragging, and sludging. Because ragging and sludging were also observed in the existing AeMBR, it can be stated that both are attributable to the stream and to the difficulty of removing existing fibers. All the mentioned phenomena could have contributed to the high data dispersion of experimental results.