Application of two-and three-phase flow in submerged flat-sheet anaerobic membrane bioreactors for dairy wastewater treatment

Sriprasert, Pakpong (2019) Application of two-and three-phase flow in submerged flat-sheet anaerobic membrane bioreactors for dairy wastewater treatment. University of Southampton, Doctoral Thesis, 251pp.

Record type: Thesis (Doctoral)

Abstract

The association of anaerobic process with membrane filtration in anaerobic membrane bioreactors (AnMBR) is a recent development in high rate anaerobic systems. This technology offers complete biomass retention regardless of settling properties, and enables the decoupling of hydraulic retention time (HRT) and solids retention time (SRT), with superior effluent qualities and small footprint areas. However, membrane fouling, an unavoidable circumstance, is still a crucial obstacle for AnMBR implementation.

Gas-liquid two-phase air/gas sparging has been applied extensively for both aerobic membrane bioreactors (AeMBR) and AnMBR systems, but sparging for fouling control purposes is always the largest energy demand for MBRs operation. Hence, sophisticated hydrodynamics design is of prime importance to achieve effective exploitation of the energy input.

The first stage of this study aimed to optimise two-phase flow hydrodynamic parameters such as gas flow rate, nozzle size and tank geometry for a gaslift-loop submerged flat-sheet MBR. The results showed that a Ø 3 mm nozzle sparger gave the largest bubble-membrane contact area. Threshold superficial air velocity was observed at 0.033 m s-1, while the increase in liquid upward flow velocity plateaued at a sparging rate of 6 Lmin-1.

Adding suspended media is regarded as a promising strategy for membrane cleaning, with active adsorbents utilised most widely in either gas-sparged or fluidized bed MBRs. However, once the sorption capacity is diminished, then membrane scouring dominates. The introduction of non-adsorbent suspended particle has also been proposed. It has been reported that external irreversible fouling, which most of the time represents the major contributor to the total fouling, has been diminished effectively by the particles abrasion.

This approach has successfully been operated with various AeMBR, while information on applying non-adsorbent suspended media in AnMBR, in which fouling phenomena are considered to be more intense when compared to AeMBR, is still very limited.

Therefore, the overall aim of this research was to study the use of non-adsorbent particles coupled with conventional gas sparged AnMBR (GSAnMBR) functioning as a three-phase (solid/gas/liquid) flow for fouling mitigation and flux enhancement purposes, in an attempt to operate as a high-rate moving bed AnMBR (MBAnMBR) for dairy wastewater treatment.

Trials were carried out to evaluate the feasibility and the related operational parameters of using non-adsorbent suspended particles associated with gas sparging in AnMBR. The experiments were carried out under simulated MBAnMBR conditions with low density polyethylene beads (SG 0.86 - 0.96, 2 - 3 mm apparent size) at different filling densities. Critical flux (Jcrit), supra-critical flux filtrations and resistance-in-series model experiments were carried out. The results demonstrated that Jcrit for 0, 11.1 and 22.2 % (v/v filling ratio) were 6, 7.5 and 9 L m-2 h-1 (LMH). Total filtration resistances (Rt) were 14.1 x 1012, 8.38 x1012 and 1.61 x 1012 m-1, of which cake resistance (Rc) represented the major contributor to Rt of 95.0, 92.3 and 62.1 %, respectively. This indicated that Rt has been decreased effectively by 40.6 - 88.6 % with the assistance of scouring particles.

Long-term continuous operation for more than 300 days of 6.6 L lab-scale GSAnMBR and MBAnMBR was carried out under mesophilic conditions (37 °C) for synthetic dairy wastewater treatment. Systems were installed with a single A4 size flat sheet membrane (Kubota) made of chlorinated polyethylene with a nominal pore size of 0.4 m and total filtration area of 0.1 m2. Biogas from the reactor headspace was circulated at a rate of 5 Lmin-1 to create gas sparging for both reactors. Polyethylene glycol (PEG) granules, cylindrical in shape with the size and SG of 4 x 4 mm and 1.01 - 1.05 were applied as scouring media in MBAnMBR (10 % by volume fraction). Initial MLVSS was set at 4,500 mgL-1 and controlled to below 14,000 mg L-1. Net fluxes were initially set constant at 3.5 - 3.8 LMH resulting in HRT of 17 - 19 h and OLR were increased stepwise from 0.7 to 5.0 g CODremoved L-1 d-1 by influent concentration adjustment for the first 100 days of operation. Thereafter, influent concentration was fixed in the range of 3,600 - 3,900 mgL-1 until the end of experiments in which attainable OLRs were determined by achievable fluxes and SRTs were varied. The results illustrated that high total COD removal of 98 - 99 % could be achieved for both reactors under stable conditions corresponding to OLRs in range of 2.2 - 7.0 g CODremoved L-1 d-1 with F/M ratios of 0.25 - 0.65 g CODremoved g VSS d-1. Attainable net fluxes for long-term operation without rapid increases in TMP were at 2.8 and 3.7 LMH for GSAnMBR and MBAnMBR. This indicated that under the same given energy input for gas sparging, net flux could be enhanced by at least 24.3 % (even up to 35 %) with the assistance of scouring particles at a lower suction pressure. Additionally, MBAnMBR could be operated continuously for more than 300 days without backflushing or chemical cleaning. The higher VMP range of 0.63 - 2.30 L CH4 L-1 reactor day-1 for the MBAnMBR when compared to 0.68 - 1.86 L CH4 L-1 reactor day-1 for the GSAnMBR was due to higher OLRs causes by greater producible flux. A methane yield of 0.31- 0.32 L CH4
g-1 CODremoved which represents 90 - 92% of theoretical methane conversion was seen from both reactors. Similar observed biomass yields (Yobs) of 0.057 - 0.059 and 0.059 - 0.072 g VSS g-1 CODremoved were found for the GSAnMBR and MBAnMBR. Rc represents the major contributor to the Rt for both reactors. However, Rc in MBAnMBR is about 12-fold lower than in GSAnMBR with values of 4.8 x 1011 m-1 and 57.6 x 1011 m-1, respectively. Cake/gel forming could be alleviated effectively. Hence, particle addition could minimise the frequency of chemical cleaning, possibly prolonging membrane lifespan as well as reducing operational and maintenance expenses. No significant damage was observed on the membrane surface from using PEG granules as scouring agents under 5 L min-1 sparging after 308 days of operation.