Diagnosis of membrane bioreactor performance ... - Science Direct

1 downloads 0 Views 5MB Size Report
Abstract. It is recognised that the performance of a membrane bioreactor (MBR) is dependent upon the flow profile and hydraulic conditions in the reactor, the ...
Desalination 236 (2009) 120–126

Diagnosis of membrane bioreactor performance through residence time distribution measurements — a preliminary study Y. Wang, Sanly, M. Brannock, G. Leslie* UNESCO Centre for Membrane Science and Technology, University of New South Wales, Sydney, Australia email: [email protected] Received 30 June 2007; revised accepted 7 October 2007

Abstract It is recognised that the performance of a membrane bioreactor (MBR) is dependent upon the flow profile and hydraulic conditions in the reactor, the literature is silent on this aspect of MBR design. Consequently, there is poor understanding of the hydraulic conditions that exist within MBR’s. One method to characterise the hydraulic conditions is based on the concept of residence time distribution (RTD). Careful measurement of the RTD profile reveals information on the degree of mixing in the reactor and any effects that the configuration of the membrane may have on the hydraulic conditions. This work focuses on developing a RTD technique for the study of the hydrodynamics and relating it to MBR process performance. The technique would be used as a tool to determine the impact of membrane geometry, orientation and mixing efficiency on MBR performance. RTD profiles were generated using the conservative tracer, lithium chloride, for a bench scale MBR’s operating at the UNESCO centre in Sydney Australia and a pilot plant MBR located at South Windsor Australia. Analysis of the RTD profiles from the lead and tail membrane positions in the reactor in the absence of aeration indicated that different performance could occur in the same reactor. Increasing the aeration rate from 3.8 to 8.1 L/min did not affect the RTD data significantly in a bench scale MBR. However, it would be an important factor associated with reactor mixing and energy in larger municipal installations. The ‘‘Tanks-in-Series’’ model was used to simulate the performance of a pilot plant MBR. However, it was found that experimental RTD profiles could not be reproduced with this simple model. Consequently, we conclude that a more fundamental approach, based on computational fluid dynamics is needed to model the hydraulic behaviour in MBR’s. Keywords: MBR; Residence time distribution; Mixing; Energy

*Corresponding author. Presented at the International Membrane Science and Technology Conference, IMSTEC 07, 5–9 November 2007, Sydney, Australia 0011-9164/09/$– See front matter # 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.desal.2007.10.058

Y. Wang et al. / Desalination 236 (2009) 120–126

121

1. Introduction

1.1. Theoretical considerations

The membrane bioreactor (MBR) has become a popular process for the treatment and reuse of municipal wastewater. Optimisation of MBRs, however, requires detailed understanding of the kinetics of biological nutrient removal (BNR), the performance of the microporous membranes and the hydraulic conditions in the bioreactor. Hydrodynamics, plays an important role since it determines the reactor’s residence time and liquid distribution in the entire reactor [1–2]. Mixing characteristics are very important for the MBR systems because they can affect both the efficiency of organic removal and the settling of the sludge [3]. Good mixing promotes the transfer of substrates and heat to the microorganisms and ensures the effective use of the entire reactor volume. Although the hydrodynamics of MBR system is of critical importance to the performance of the system, MBRs are usually assumed as completely mixed flow reactors (CSTR) and designed based mainly on the biokinetics and fouling potential of the treatment system. Ideal flow is hardly attainable in practice or requires perfect mixing. Consequently, the hydraulic profile of MBR has been an insufficiently understood aspect of MBR design. One method to characterize the hydraulic profile is based on the concept of residence time distribution (RTD). The degree of mixing affects the output response describing the system’s flow regimes and can be expressed by the RTD profiles. Unlike conventional process where treated effluent is recovered at a single position determined by the overflow weir of the gravity settler, MBRs may have multiple extraction points (depending on membrane position in the reactor). This may lead to multiple RTDs for an MBR system. The following paper considers the theoretical basis for mixing and an experimental approach for measuring the degree of mixing in MBR systems.

To characterize the non-ideal flow within vessels, many types of models have been developed. The two most frequently used hydraulic models are the dispersion model and Tanks-inSeries model. Based on the superposition of the longitudinal dispersion with an ideal plug flow, dispersion model is described with a single parameter, D/uL or the dispersion number, which is the ratio of radial to axial mixing from the inlet to the outlet of the reactor. Completely mixed conditions occur when the dispersion number approaches infinity, while negligible dispersion or plug flow conditions exist as the dispersion number approaches zero. The Tanks-in-Series model considers the real reactor as a series of equal sized ideal stirred tanks [4]. The mixing regime, from completely mixed flow (N ¼ 1) to plug flow (N ¼ 1) could be covered by this model. The theoretical C(t) curve of the Tanks-in-Series model can be predicted using Martin Method which uses a Gamma distribution to allow non-integer numbers of tanks in series to be modelled (Eq. (1)) [5].  N 1 NN t   eðN t=tm Þ CðtÞ ¼ C0  tm ðN Þ ð1Þ where C0 is the tracer concentration at time zero, tm is the mean residence time. In this work, tracer studies using lithium chloride were carried out to acquire RTD profiles of a bench scale MBR and an intermediate scale MBR (South Windsor MBR pilot plant, Australia). The effects of membrane positions and aeration rate on RTDs were evaluated. These are the preliminary studies on the hydrodynamics of MBRs to show the importance of mixing on MBR systems. In addition, Tanks-in-series model for the pilot MBR was developed for comparison against the tracer study results.

122

Y. Wang et al. / Desalination 236 (2009) 120–126

2. Methods 2.1. Tracer studies for bench scale membrane bioreactor Bench scale experiments were performed in a plastic rectangular tank with dimension (L  W  H) of 30  23.5  12.5 cm3 (Fig. 1). Microporous polypropylene hollow fibres with a median pore size of 0.2 microns, an outside diameter of 0.65 mm, a wall thickness of 0.13 mm (US Filter Memcor, Windsor Australia) were mounted in two modules (total surface area of 0.0375 m2 per module). The modules were orientated in parallel and in series relative to the direction of flow through the tank (Fig. 1). The feed was regulated by the feeding pump, whilst the effluent was drawn directly from the MBR through the membrane by a variable speed suction pump (Cole-Parmer Masterflex pump model 7518-00). The permeate was sucked through the fibres and the filtration thus operating from the external skin of the fibres inwards. Two air diffusers of approximately 23 cm in length were used aerate the tank. A T-junction was built to evenly split the air flow into two channels connected to the two aerators. The air flowrate could be varied and was measured using a flowmeter. The flowrate through the each membrane was maintained at 66.7 mL/ min. Each membrane module was connected Membranes

Inlet jet

Pump

Sampling point

straight to the pump so that samples could be taken from both membrane outlets. Lithium chloride was dissolved in small amount of water and then injected into the feed line of the system with a syringe over as short time as possible. The experiment was carried out in single pass mode, where pure water was pumped from a reservoir into the reactor and samples were taken at both membrane outlets at predetermined interval and the rest are discharged. The inlet jet was positioned at the middle of the tank and the flow was directed at the tank wall to ensure that the inlet jet would not cause additional mixing effect. Sampling was undertaken for three hydraulic residence time ensuring 100% recovery of tracer.

2.2. Tracer studies for intermediate scale membrane bioreactor (South Windsor pilot plant) A MBR pilot plant with a treatment capacity of 100 kL/day was operated on municipal effluent wastewater at a sewage treatment plant located in Windsor New South Wales (Fig. 2). The primary treatment processes included a 3 mm screen, a grit chamber, and a 1 mm screen. A membrane module containing 0.1 micron PVDF (polyvinylidene fluoride) hollow fibre membranes with total surface area of 10 m2 was immersed in the reactor tank. The membranes were operated in a cross flow with a pre-aerated feed pumped past the membrane to create conditions of true two phase flow (air and liquid). The modules were periodically backwashed using a combination of compressed air and screened Dose point

Pump

Grit chamber

Q2

LS

Flowmeter

Aerators 3 mm screen

Pump

1 mm screen

T1 Anoxic

T2 Aerobic

Q1

Sample point T3 Q3 Membrane

To compressed air supply

Fig. 1. Diagram of bench scale MBR.

Fig. 2. Schematic diagram for MBR pilot plant in Windsor.

Y. Wang et al. / Desalination 236 (2009) 120–126

Table 1 Operating parameters for the MBR pilot plant during tracer study Volume of T1, L Volume of T2, L Volume of T3, L Intermittent inlet Q to anoxic tank (L/h) Flowrate from anoxic to aerobic tank (L/h) Q1 (L/h) Q2 (L/h) Q3 (L/h) Aerobic tank air flow (m3/h) Membrane tank air flow (m3/h) Total volume of sludge wastage (L/day) Frequency of sludge wastage

1072 1191 92 3600 L/h for 80 s every 20 min 400 2000 1600 400 5.5–6.5 3 155

Table 2 Wastewater characteristics of pilot plant MBR in Windsor Influent BOD (mg/L)

200

Effluent BOD (mg/L) Influent COD (mg/L) Effluent COD (mg/L) MLSS membrane tank