- Email: [email protected]
New modules for membrane bioreactors: Improving fouling control
Accepted Manuscript Title: New modules for membrane bioreactors: improving fouling control Authors: Robson Rodrigues Moror´o, Cristiano Piacsek Borges, Frederico de Araujo Kronemberger PII: DOI: Reference:
S0263-8762(18)30274-0 https://doi.org/10.1016/j.cherd.2018.05.035 CHERD 3199
To appear in: Received date: Revised date: Accepted date:
5-3-2018 27-4-2018 25-5-2018
Please cite this article as: Moror´o, Robson Rodrigues, Borges, Cristiano Piacsek, Kronemberger, Frederico de Araujo, New modules for membrane bioreactors: improving fouling control.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2018.05.035 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
New modules for membrane bioreactors: improving fouling control
1
IP T
Robson Rodrigues Mororó1, Cristiano Piacsek Borges1, Frederico de Araujo Kronemberger1,*
Chemical Engineering Program – COPPE/Federal University of Rio de Janeiro, Brazil. P.O.
*
N
U
Corresponding author email: [email protected]
SC R
Box: 68502, ZIP CODE: 21941-972, Rio de Janeiro, RJ.
M
New module for submerged membrane bioreactors with integrated aeration was evaluated;
ED
A
Highlights:
The lower the packing density, the smaller the transport resistance;
A high number of air injectors in the module base resulted in elevated flux values;
Reduction in transport resistance was correlated to the aeration energy expenditure.
CC E
PT
A
Abstract: New hollow fiber membrane modules with air injectors coupled to their base are presented in this paper. The transport resistances were analyzed through permeation tests with yeast cells suspensions varying parameters such as surface air velocity and filtration pressure. Five different modules were investigated, and the lowest resistances were observed in the module with 64 holes in the air injector and packing density equal to 650 m².m-3. The
1
modules with 32 holes in the air injector and packing density of 650 m².m -3 showed the best relation between the efficiency in reducing transport resistance and the aeration energy expenditure.
SC R
IP T
Keywords: Fouling; transport resistance; hydrodynamics; aeration; MBR.
1 Introduction
U
The use of membranes to assist biological treatment of wastewaters, whether industrial
N
or domestic, has enormous potential in the solid-liquid separation, replacing traditional side
A
decanters [1-5]. Membrane Bioreactors (MBR) have many advantages over conventional
M
biological treatment of wastewater, such as the production of better quality effluent and up to 99% removal of pollutant loading, since all suspended biomass is retained by membranes [6].
ED
It is also possible to operate with higher sludge concentration, as the decanting step is
PT
replaced by this physical barrier. Regarding the equipment, MBR are significantly more compact, which is becoming a key issue, especially taking into account that the physical space
CC E
for installation of treatment units is limited in most of the metropolitan areas. The introduction of MBR with submerged membranes (sMBR) significantly reduced energy consumption associated with the operation, compared to the MBR in which membranes are
A
allocated outside the bioreactor (eMBR) [3]. As for all membrane separation processes, fouling problems are crucial in achieving
an efficient sMBR process, taking into account the energy expenditure used to control it [2,3,5-8]. During sMBR operation, flocs containing colloidal particles, macromolecules and microorganisms are driven to the membranes and accumulate in their pores and surface, 2
reducing the permeability, i.e., increasing the transport resistance [3]. To overcome membrane fouling problems, new membranes can be developed or modified using different techniques [9], but several scientific papers in the literature indicate that hydrodynamics is the key parameter to control fouling [5,10-14]. There are numerous recent developments in fouling control in sMBR, including specific turbulence generators [15,16]. But among the factors
IP T
affecting hydrodynamics in an sMBR and, consequently, fouling control, the most studied are air sparging and module design. Air injection is used to produce shear forces in the layer
SC R
deposited on the membrane. Module design is related to the type and position of the air
injector, and the geometry and packing density of membranes. These design variables present a strong influence on aeration patterns [7,14,17]. Hollow fiber membranes are the most
U
commonly used geometry, offering higher filtration area per volume of module, even though
N
helical membrane and folded plate/tilted membrane modules are being experimented in
A
laboratory scale [18-20]. Concerning aeration, most of the published works are related to the
M
influence of air flow rate and the distribution and size of the bubbles in the filtration
ED
resistance [5,14]. Regarding the design of membrane modules, most of the publications are focused on modules with air injectors attached to their bases in a centralized manner, which
PT
can lead to stagnant regions near the membrane, causing intense accumulation of suspended material. Vibrating or rotating hollow fiber membrane modules are being studied recently
CC E
[21,22], but the design of new stationary modules is still required. In the present work, new hollow fiber membrane modules were developed and
A
evaluated. These modules, with several air injectors coupled to their base in a uniform distributed manner, were evaluated aiming at reducing the overall transport resistance. Five modules were tested, with different numbers of air injectors and different packing densities, in order to determine the one with better efficiency, taking into account fouling control and air energy expenditure. For this purpose, transport resistances caused by fouling (Rf) were
3
obtained from the constant pressure permeation tests of fresh yeast suspensions under various conditions, such as different surface air velocities and applied pressures. The suspensions are used as a model to screen and compare different module configurations in order to define which of the five modules could be tested in a MBR.
IP T
2 Methods
SC R
2.1 Hollow fiber membrane modules
The sMBR modules were developed by PAM-Membranas Seletivas Ltda. (Rio de
U
Janeiro, Brazil) with microfiltration poly(ether imide) hollow fiber membranes (pore diameter
A
N
< 0.1 m). The average outer diameter of the fibers was 1 mm and they were fixed to the
M
bottom of the module coupled to the air injectors in a uniform distributed manner. The membranes were allocated within a perforated poly(vinyl chloride) (PVC) pipe with 44 mm
ED
of inner diameter and 200 mm height, together with a PVC screen for protection, preventing the intense movement of the membranes with the rising bubbles, which could lead to
PT
disruption. The modules comprised an inlet for the air at their base and a permeate outlet at
CC E
the top. The basic design of one module and the scheme of the base with the air injectors are presented in Figure 1. Since air is injected in a dispersed and uniform way among the fibers, its distribution is better, enhancing fouling control. Technical specifications of the modules
A
evaluated are shown in Table 1.
4
2.2 Permeation system
The modules were characterized in submerged operation the module was immersed in the feed tank with the suspension to be filtered. The driving force for permeation was obtained using a suction pump, connected to the lumen side of the fibers, maintaining the
IP T
pressure in the feed tank equal to the atmospheric one. Figure 2 shows the flowchart of the
SC R
system used in this study.
N
U
2.3 Permeation test conditions
A
The permeation tests were carried out with suspensions of Saccharomyces cerevisiae
M
cells (Itaiquara®-Tapiratiba/SP) in pure water (distilled, demineralized and microfiltered). The first objective was the individual evaluation of the modules, with respect to the influence
ED
of operational parameters, pressure difference and air superficial velocity (Ug), in the fouling formation. Then, the modules were compared to each other for the evaluation of their design
PT
parameters the number of air injectors and the packing densities. Initially, pure water permeabilities were determined at 0.7 bar of pressure difference – the highest pressure applied
CC E
– to exclude the effect of irreversible pore deformation. Finally, the pure water hydraulic permeability was evaluated. Then, critical conditions were evaluated through filtration
A
experiments at constant pressure values. The critical operational conditions determination – critical flux or critical pressure – is important in order to minimize fouling. Critical pressure can be defined as the pressure beyond which concentration polarization will occur, increasing fouling due to the higher suspended matter concentration in the membrane surface.
5
Permeation tests were performed using suspensions with 8,000 mg.L-1 of the fresh yeast with superficial air velocities (Ug) of 0.0, 1.3, 2.7 and 5.0 m.s-1, at filtration pressures of 0.3 and 0.7 bar, using all modules. The Ug is the ratio between the air flow and the total area of the air injectors. In these tests, the same Ug values were used to ensure that the hydrodynamic conditions near the surface of the membranes were similar, rather than using
IP T
the same air flow rate. The values of the parameters evaluated in this study were based on
SC R
typical literature values for sMBR.
2.4 Transport resistances caused by fouling (Rf)
U
The analysis of the transport resistances caused by fouling (Rf) is a powerful tool for
N
the evaluation of different modules, disregarding membrane morphological characteristics.
A
The model of resistances in series (Equation 1) was considered. Consequently, the total
M
transport resistance (RT) was the sum of membrane intrinsic resistance (Rm), the adsorption
ED
resistance (Ra), the pore blocking resistance (Rpb) and the fouling resistance (Rf), considering that the concentration polarization resistance value was comprised into the fouling resistance,
PT
since it is difficult to dissociate them. The adsorption and pore blocking resistances were
CC E
measured and considered negligible data not shown – while the membrane intrinsic resistance value was obtained from the pure water permeability. Thus, the fouling resistance value (Rf) was obtained by subtracting the membrane resistance from the total transport
A
resistance value.
𝑅𝑇 = 𝑅𝑚 + 𝑅𝑎 + 𝑅𝑝𝑏 + 𝑅𝑓
Equation 1
6
RT was calculated at the end of each permeation test (180 min), using the permeability final value (Lpfinal), according to Equation 2. In this model, the permeate flux in a membrane is described by Darcy's law:
𝐽𝑝𝑠𝑡𝑎𝑏𝑖𝑙𝑖𝑧𝑒𝑑 = 𝐿𝑝𝑓𝑖𝑛𝑎𝑙 . ∆𝑃 =
∆𝑃
Equation 2
𝜇.𝑅𝑇
IP T
where Jp is permeate flow (m3.m-2.h-1), ∆P is the pressure difference across the
SC R
membrane (Pa), μ is the fluid viscosity (Pa.s) and RT is the total resistance (m-1).
U
2.5 Analysis of the overall efficiency of the modules
N
Aeration is responsible for approximately 90% of total energy expenditure of a sMBR,
A
hence special attention should be given to this parameter [5]. The relationship between
M
transport resistance caused by fouling and the energy expenditure in the aeration, named from now on as the overall efficiency of the module, must be optimized. Aeration should not be
ED
used without restriction in fouling control, as the energy demand should be reduced and an excessive aeration could lead to the disruption of some flocs, liberating polymeric material
PT
which could enhance fouling [23]. Therefore, in order to evaluate the overall efficiency of the
CC E
modules, the aeration power in yeast permeation tests, obtained according to Equation 3, was
A
analyzed together with the transport resistance caused by fouling (Rf).
𝑃𝑜𝑤𝑒𝑟𝑎𝑖𝑟 = 𝑃𝑎𝑖𝑟 . 𝑄𝑎𝑖𝑟
Equation 3
where Pair is the air pressure measured at the injector inlet and Qair is the air flow rate. The values of Powerair and Rf were normalized between 0 and 1 and plotted against each other. 7
3 Results and discussion
3.1 Pressure and air velocity influence on filtration resistance
IP T
Prior to the filtration experiments, critical conditions were determined for all five
modules. As an example, the result obtained for module 32-650 at the highest air flow rate is
SC R
presented in Figure 3. Critical conditions were evaluated using yeast suspensions with the same concentration employed in the subsequent experiments and applying different air flow rates. Using the highest air flow rate, two modules, 32-650 and 64-650, presented critical
U
pressures between 0.10 and 0.15 bar. Considering the other modules and lower air velocities,
N
the critical pressure was below 0.10 bar. Since it would not be possible to operate all modules
A
at subcritical conditions using the chosen concentration, all tests were performed above
ED
M
critical conditions.
PT
The average value for the pure water permeability of the fibers, used in the resistances calculations, was 571.2 L.h-1.m-2.bar-1. Concerning the experiments with the yeast
CC E
suspensions, all tests were carried out for 180 min, so that the permeate flux would be constant, indicating the stability of the fouling layer formed. Figure 4 shows the time course
A
profiles of the hydraulic permeability normalized by its initial value (Lp.Lpo-1) for the superficial air velocities (Ug) tested for the module 32-650.
8
The initial permeability values were lower than the pure water hydraulic permeabilities (value not shown) due to the concentration polarization phenomenon, which is established in first moments of test, added to the intense initial fouling. The first observation from the results presented in Figure 4 is that all profiles present the same pattern. The permeability drops promptly in the first 10 minutes, stabilizing after 90 minutes of
IP T
permeation. For all air velocities investigated, it was observed that the flux decline was higher
for the pressure difference of 0.7 bar. This observation could be attributed to the thickness and
SC R
the compactness of the fouling layer. Particle accumulation in the fouling layer was enhanced
as transmembrane pressure increased because more suspended solids were dragged to the membrane surface by the increased permeate flux. Furthermore, the fouling layer could be
U
denser at high transmembrane pressures. Hong et al. [1] studied the impact of transmembrane
N
pressure on fouling during microfiltration tests in an activated sludge submerged membrane
A
bioreactor (sMBR). The authors tested 0.1, 0.2 and 0.3 bar and observed the largest decline in
M
hydraulic permeability (Lp) for 0.3 bar. The permeability reduction for this pressure was 66%
ED
in the first 100 minutes of operation, against 45% and 40% for 0.2 and 0.1 bar, respectively. While in the remainder of the experiments, until 300 minutes, almost no variation in Lp was
0.2 and 0.3 bar.
PT
observed when applying 0.1 bar, a further decrease of about 27% was observed when using
CC E
It is also noticed that increasing air velocity enhances the hydrodynamic conditions
inside the sMBR system, since the permeability values are also raised. Considering the
A
highest pressure difference applied, the permeability values for 150 min of experiment are equal to 14%, 30%, 36% and 41%, in relation to the initial ones, for the air velocities of 0.0, 1.3, 2.7 and 5.0 m.s-1, respectively.
9
Typical results of the influence of aeration on the membrane permeability of the modules during the yeast suspensions filtration are shown in Figure 5, also for the module with 32 air injectors and packing density of 650 m2.m-3.
IP T
It is clearly observed from Figure 5 that the injection of air during the yeast suspension permeation tests in the module 32-650 increased the productivity. The membrane
SC R
permeabilities undergo a greater decline with respect to the accumulated permeate volume obtained during the tests without aeration (0.0 m.s-1) than in the test using air under a superficial velocity of 5.0 m.s-1 (Figure 5, a and c), showing that the latter operation is more
U
stable over time, which would result in lower cleaning frequency. It is interesting to observe
N
that, without air injection, the flux reduces to merely 17% of the initial value with 30 L.m -2 of
A
permeate volume, considering the pressure of 0.7 bar (Figure 5, c). Even with a lower
M
pressure, 0.3 bar (Figure 5, a), the flux is reduced to 28% of the initial one, for the same
ED
volume of permeate. When sparging air, these values increase to 45% and 52%, for 0.7 and 0.3 bar, respectively. This indicates clearly that the aeration, or the hydrodynamics conditions
PT
near the fibers, is a key factor to control fouling in sMBR systems, above all, considering that the fouling could be much more severe when considering the real biological media. Another
CC E
confirmed pattern is the permeate flux dependence with the pressure difference. The greater the pressure difference is, the higher the permeate flux declines. Even considering that the
A
cumulative permeated volume is higher for the pressure difference of 0.7 bar during the whole experiments (Figure 5, b and d), the energy expended per permeated volume, including the membrane cleaning, will not make this high pressure operation economically viable. It is also noticed that the cumulative permeated volume is lower when no aeration was used, mainly for
10
higher filtration times, when the accumulation of yeast particles on the surface of hollow fibers is more severe, increasing the transport resistance.
3.2 Comparison among different module designs
IP T
In order to compare the different module designs, the variation in the transport resistance values caused by fouling (Rf) and the final average permeability of the membranes
SC R
(Lpfinal) are presented in Figure 6 as a function of the superficial velocity of the air bubbles
U
(Ug) in the permeation tests with the yeast suspensions.
N
It can be confirmed that the use of air was effective in reducing fouling on the
A
membranes, as noticed in the transport resistance variation from Ug = 0.0 m.s-1 to Ug = 1.3
M
m.s-1, for all models of the permeator. Nevertheless, subsequent increases in the superficial
ED
velocity of air (Ug) will not lead to a linear augmentation in the average permeability final values (Lpfinal) nor in a linear reduction in the transport resistance caused by fouling (Rf). For
PT
example, for the module 32-650, the variation of Ug from 0.0 to 1.3 m.s-1 caused a decrease of approximately 50% of the Rf value, while the variation from 2.7 to 5.0 m.s-1 produced a
CC E
further decrease of only 14% in the transport resistance caused by fouling. These results are attributed to the turbulence promoted by air bubbles, which creates a field of shear forces on
A
the cake adhered on the surface of the hollow fibers, causing the particles to return to the yeast suspension. The higher Ug, the higher is the velocity of the rising air bubbles and greater the turbulence and the intensity of shear force vectors. But there is a critical value for which the increase of Ug is not accompanied by an increase in the bubbles velocity and the effect on the reduction of fouling is not changed [2,3,5,24-29]. The most probable reasons for the
11
existence of this Ug limit value, or for the plateau in the curves of Rf in function of Ug, may be linked to the fact that the increase of Ug is not completely accompanied by a proportional increase in the rise rate of the bubbles, apart from that the shape of the bubbles may be changing, as well as their size and number, which influences the coalescence. Comparing the filtration pressures tested, it is observed, for 0.7 bar, that the use of air
IP T
becomes less effective in controlling fouling formation due to an imbalance between the shear
stress and the permeation driving force, the latter one being dominant, which causes a higher
SC R
amount of suspended particles to move to the membrane surface and remain adhered to it. Furthermore, the fouling layer porosity can be reduced by compression.
Li et al. [30] treated sewage in a MBR with submerged hollow fibers, at a constant
U
permeate flux equal to 12.5 L.h-1.m-2, and found Rf values above 175 x 1011 m-1 within 20
N
days of filtration, using an air flow rate of 36.0 m³.h-1 per membrane square meter. This
.m-2. Chu and Li [31] found Rf values of 400 x 1011 m-1 after 70 days of operation in a
M
1
A
resistance was reduced to 100 x 1011 m-1 when the aeration rate was increased to 64.8 m³.h-
ED
submerged hollow fiber MBR treating domestic sewage. Lesage et al. [32] also treated sewage in a hollow fiber sMBR using an aeration rate of 1.2 m³.h-1.m-2 and observed Rf values
PT
exceeding 2,500 x 1011 m-1. Han et al. [33] tested submerged flat sheet membranes in a MBR treating synthetic sewage, using air flow of 1.0 m³.h-1.m-2 and a fouling resistance of 1,070 x
CC E
1011 m-1 was observed.
For the new modules studied in the present work, different air flow rates per
A
membrane area were used, depending on the module design. The maximum rates, which correspond to Ug of 5.0 m.s-1 were 1.15, 2.30, 4.60, 1.87 and 1.50 m³.h-1.m-2 for the modules 16-650, 32-650, 64-650, 32-800 and 32-1000, respectively. The resistance values for these modules varied from 21.2 x 1011 to 45.3 x 1011 m-1, but these values cannot be directly
12
compared to the ones presented in the scientific literature since they were obtained using a synthetic suspension, as the main goal of this work was to compare different module designs. In order to assess the modules' performance in relation to the number of air injectors and the packing density of the fibers, the fouling resistance data were plotted again, segregating the data for each of these parameters. The comparison of the values of Rf resulting
IP T
from permeation tests of fresh yeast suspensions using permeators with the same packing density (650 m2.m-3) and the different number of air injectors (16, 32 and 64) is presented in
SC R
Figure 7, for both 0.3 and 0.7 bar.
U
The number of air injectors in the base of the modules has a major influence in fouling
N
control. It is observed that a higher number of air injectors reduces Rf for both filtration
A
pressures investigated. These results are related to the fact that a better distribution of bubbles
M
among the hollow fibers is obtained with a higher number of injectors distributed in the base
ED
of the modules. The homogeneity of the shear forces due to the rise of bubbles among the membranes constitutes a good hydrodynamic pattern and may depend on the proper design of
PT
the aeration system [34]. Mayer et al. [35] recommended the use of complex systems, provided with numerous holes, which would make the distribution of air more homogeneous
CC E
and thus more effective, which is consistent with the results presented in Figure 7. Yet, the modules with more air injectors may generate a greater amount of bubbles in a confined
A
space, increasing coalescence. According to Braak et al. [5], bubbles with large diameters formed by the coalescence of smaller ones promote increased turbulence regions. And the results are notoriously related to the fact that the air flow rate is higher in the modules with more air injectors, for the same air velocity. The energy expenditure is thus higher in these cases, which will be discussed later in this paper.
13
Figure 8 compares the result of Rf obtained in the permeation tests using pressure differences of 0.3 and 0.7 bar, but now analyzing the modules 32-650, 32-800 and 32-1000, which have the same number of air injectors attached to their base (32) and fiber packing
IP T
densities of 650, 800 and 1,000 m2.m-3, respectively.
The increase of fibers packing density severely increases fouling on the surface of the
SC R
membranes, particularly for the highest filtration pressure, which offers the greatest driving
force to the flow, causing a great amount of particles to adhere to the membrane surface and making the fouling layer more compact (less porous). These results are probably due to the
U
fact that increased packing density of the hollow fibers might hinder the contact of bubbles
N
with the inner fibers due to physical blocking, even with the uniform distribution of the
A
injectors, limiting the effects of aeration on the control of fouling. It has to be noted that the
M
increased packing density of fibers reduces the percentage ratio of the air injectors per number
ED
of hollow fibers. Models 32-650, 32-800 and 32-1000 have 10.1%, 8.3% and 6.6% of injectors in relation to the number of fibers, respectively. This also contributes to a worse air
PT
distribution for the modules with increased packing density. For example, at Ug of 5.0 m.s-1, Rf is 29.5 x 1011, 36.4 x 1011 and 45.3 x 1011 m-1 for 10.1%, 8.3% and 6.6% of the number of
CC E
holes in relation to the number of hollow fibers (650, 800 and 1,000 m².m-3), respectively. Among the modules tested, the one that presented the lowest value of Rf in all
A
superficial air velocities (Ug), i.e., the best efficiency in fouling control, was the module 64650, which has the largest number of air injectors (64) and lower packing density, equal to 650 m2.m-3. In this module, it is believed that the hydrodynamic conditions are more appropriate for the fouling control, due to better distribution of bubbles, which excludes the presence of stagnant zones, which have great potential for fouling. The lowest fouling
14
resistance (21.2 x 1011 m-1) was observed for this module with the highest air velocity tested, 5.0 m.s-1.
3.3 Power consumption analysis
IP T
In sMBR, aeration is responsible for approximately 90% of total energy expenditure,
so fouling control should not be investigated regardless of this subject [5]. The control of
SC R
fouling is important to ensure process stability, resulting in good permeate productivity,
decreased number of cleanings and increased lifetime of membranes, reducing the operational costs. In contrast, the operating cost related to the energy spent by the aeration used to control
U
fouling cannot be so high that it would economically derail the process. Therefore, a balance
N
between the benefits of fouling control and the energy expenditure in aeration is required.
A
Since both energy consumption and fouling resistance should be minimized, one possible way
M
to evaluate these variables together is to normalize them from 0 to 1, in order to avoid weight
ED
influence, and then to plot them in one graph. Figure 9 shows the aeration power influence on the transport resistance caused by fouling on the surface of the hollow fibers. The resistance
PT
values were obtained during the fresh yeast suspensions permeation tests in the different
CC E
modules using 0.3 and 0.7 bar.
A
The objective in any permeation analysis in sMBR is to obtain the lowest fouling
resistance spending as little energy as possible. Therefore, analyzing the graphics presented in Figure 9, the objective is to get as close as possible to the dashed line, representing the reduction of both energy consumption and transport resistance. It can be seen that, for both pressures evaluated in the permeation tests, 0.3 and 0.7 bar, the module 32-650 presented the
15
best commitment between the two variables in evidence, Powerair and Rf. Considering the lowest pressure, the modules 32-800 and 16-650 also seem to be considered good alternatives. But for 0.7 bar, only the module 16-650 could be an alternative for the 32-650. These results reflects that, although module 64-650 presents the best hydrodynamic conditions among all modules tested, since the lowest transport resistances (Rf) were observed in the permeation
IP T
tests performed with this module, it does not present the best commitment between the
efficiency in controlling fouling and the energy expenditure, i.e., the overall efficiency. The
SC R
lowest resistance values observed for this module was accompanied by a humongous energy expenditure, which would not desired when applying it in a sMBR.
N
U
4 Conclusions
A
New modules for submerged membrane bioreactors (sMBR) with air injectors attached to
M
their base were compared in terms of fouling resistance and aeration energy expenditure.
ED
Increasing pressure in all permeation tests with model suspensions of fresh yeast cells caused a greater decrease in the permeability throughout the experiments, but the air injection could
PT
reduce this effect in both pressures tested. However, the relation between the average permeability of the membranes (Lpfinal) and the superficial velocity of the air bubbles (Ug) is
CC E
nonlinear: there is a limiting value for which the increase of Ug does not represent a further influence in the Lpfinal value. Concerning the comparison among different modules, model 64-
A
650 (with 64 air injectors and 650 m2.m-3 of packing density) presented the lowest transport resistances caused by fouling (Rf) among all the models tested, for all superficial air velocities (Ug). The lowest value (21.2 x 1011 m-1) was observed for this module with the highest air velocity, 5.0 m.s-1. This indicated that the best results in terms of fouling resistance would be achieved with lower packing density and higher air injectors number. But since the energy
16
spent in air sparging corresponds of the majority of the operational costs of a sMBR, this simplistic evaluation is not completely accurate. Through an aeration power/resistance analysis, the module 32-650 presented the best ratio between the efficiency in controlling fouling resistance (Rf) and energy expenditure (Powerair). The decision on whether to use one
operational parameters, related to the real biological medium operation.
SC R
5 Acknowledgements
IP T
or another of these modules and about the operating conditions will rely on specific
R. R. Mororó would like to thank CAPES - Coordenação de Aperfeiçoamento de Pessoal de
U
Nível Superior (Brazil) for his scholarship. The authors would like to thank PAM-Membranas
A
CC E
PT
ED
M
A
N
Seletivas (Rio de Janeiro, Brazil) for gently providing the modules used in this study.
17
References [1] Hong SP, Bae TH, Tak TM, et al. Fouling control in activated sludge submerged hollow fiber membrane bioreactors. Desalination. 2002;143:219-228. [2] Kim J, Chang I, Shin D, Park H. Membrane fouling control through the change of the depth of a membrane module in a submerged membrane bioreactor for advanced wastewater
IP T
treatment. Desalination. 2008;231:35-43.
[3] Park H, Lee YH, Kim H, et al. Reduction of membrane fouling by simultaneous upward
SC R
and downward air sparging in a pilot-scale submerged membrane bioreactor treating municipal wastewater. Desalination. 2010;251:75-82.
[4] Kraume M, Drews A. Membrane bioreactors in waste water treatment – status and trends.
U
Chem Eng Technol. 2010;33(8):1251-1259.
N
[5] Braak E, Alliet M, Schetrite S, Albasi C. Aeration and hydrodynamics in submerged
A
membrane bioreactors. J Membrane Sci. 2011;379:1-18.
M
[6] Judd S. Principles and Applications of Membrane Bioreactors in Water and Wastewater
ED
Treatment. San Diego (CA): Elsevier; 2006.
[7] Viana PZ, Nobrega R, Jordão EP, Azevedo JPS. Optimizing the operational conditions of
2005;48:119.
PT
a membrane bioreactor used for domestic wastewater treatment. Braz Arch Biol Techn.
CC E
[8] Viero AF, Demelo T, Torres A, et al. The effects of long-term feeding of high organic loading in a submerged membrane bioreactor treating oil refinery wastewater. J Membrane
A
Sci. 2008;319:223-230. [9] Deowan SA, Galiano F, Hoinkis J, et al. Novel low-fouling membrane bioreactor (MBR) for industrial wastewater treatment. J Membrane Sci. 2016;510:524-532. [10] Evenblij H, Geilvoet S, van der Graaf JH, van der Roest HF. Filtration characterization for assessing MBR performance: three cases compared. Desalination. 2005;178:115-124.
18
[11] Brannock MW, De Wever H, Wang Y, Leslie G. Computational fluid dynamics simulations of MBRs: Inside submerged versus outside submerged membranes. Desalination. 2009;236:244-251. [12] Marselina Y, Lifia, Le-Clech P, et al. Characterization of membrane fouling deposition and removal by direct observation technique. J Membrane Sci. 2009;341:163-171.
IP T
[13] Khalili-Garakani A, Mehrnia MR, Mostoufi N, Sarrafzadeh MH. Analyze and control
fouling in an airlift membrane bioreactor: CFD simulation and experimental studies. Process
SC R
Biochem. 2011;46:1138-1145.
[14] Ratkovich N, Naessens W, Maere T, et al. Critical review of membrane bioreactor models – Part 2: Hydrodynamic and integrated models. Bioresource Technol. 2012;122:107-
U
118.
N
[15] Pourbozorg M, Li T, Law AW. Effect of turbulence on fouling control of submerged
A
hollow fibre membrane filtration. Water Res. 2016;99:101-111.
M
[16] Krzeminski P, Leverette L, Malamis S, Katsou E. Membrane bioreactors – A review on
ED
recent developments in energy reduction, fouling control, novel configurations, LCA and market prospects. J Membrane Sci. 2017;527:207-227.
PT
[17] Liu X, Wang Y, Waite TD, Leslie G. Numerical simulations of impact of membrane module design variables on aeration patterns in membrane bioreactors. J Membrane Sci.
CC E
2016;520:201-213.
[18] Jie L, Liu L, Yang F, et al. The configuration and application of helical membrane
A
modules in MBR. J Membrane Sci. 2012;392-393:112-121. [19] Zhang H, Zhang J, Jiang W, et al. A new folded plate membrane module for hydrodynamic characteristics improvement and flux enhancement. Sep Purif Technol. 2015;154:36-43.
19
[20] Eliseus A, Bilad MR, Nordin NAHM, et al. Tilted membrane panel: A new module concept to maximize the impact of air bubbles for membrane fouling control in microalgae harvesting. Bioresource Technol. 2017;241:661-668. [21] Ruigómez I, Vera L, González E, Rodríguez-Sevilla J. Pilot plant study of a new rotating
J Membrane Sci. 2016;514:105-113.
IP T
hollow fibre membrane module for improved performance of an anaerobic submerged MBR.
[22] Zsirai T, Qiblawey H, A-Marri MJ, Judd S. The impact of mechanical shear on
SC R
membrane flux and energy demand. J Membrane Sci. 2016;516:56-63.
[23] Wisniewski C, Grasmick A, Leon Cruz A. Critical particle size in membrane bioreactors: Case of a denitrifying bacterial suspension. J Membrane Sci. 2000;178(1-2):141-150.
U
[24] Ueda T, Hata K, Kikuoka Y, Seino O. Effects of aeration on suction pressure in a
N
submerged membrane bioreactor. Water Res. 1997;31(3):489-494.
A
[25] Cabassud C, Laborie S, Lainé JM. How Slug Flow can Improve Ultrafiltration Flux in
M
Organic Hollow Fibres? J Membrane Sci. 1997;128:93-101.
ED
[26] Chang S, Fane AG. The effect of fibre diameter on filtration and flux distribution– relevance to submerged hollow fibre modules. J Membrane Sci. 2001;184:221-231.
PT
[27] Le-Clech P, Chen V, Fane AG. Fouling in membrane bioreactors used in wastewater treatment. J Membrane Sci. 2006;284:17-53.
CC E
[28] Cui ZF, Chang S, Fane AG. The use of gas bubbling to enhance membrane processes. J Membrane Sci. 2003;221:1-35.
A
[29] Sofia A, Ng WJ, Ong SL. Engineering design approaches for minimum fouling in submerged MBR. Desalination. 2004;160:67-74. [30] Li Y, He Y, Liu Y, et al. Comparison of the filtration characteristics between biological powdered activated carbon sludge and activated sludge in submerged membrane bioreactors. Desalination. 2005;174:305-314.
20
[31] Chu HP, Li XY. Membrane fouling in a membrane bioreactor (MBR): Sludge cake formation and fouling characteristics. Biotechnol Bioeng. 2005;90:323-331. [32] Lesage N, Spérandio M, Cabassud C. Performances of a hybrid adsorption/submerged membrane biological process for toxic waste removal. Water Sci Technol. 2005;51:173-180. [33] Han SS, Bae TH, Jang GG. Tak, T.M., Influence of sludge retention time on membrane
IP T
fouling and bioactivities in membrane bioreactor system. Process Biochem. 2005;40:23932400.
SC R
[34] Fane AG, Yeo A, Law A, et al. Low pressure membrane processes – doing more with less energy. Desalination. 2005;185:159-165.
[35] Mayer M, Braun R, Fuchs W. Comparison of various aeration devices for air sparging in
A
CC E
PT
ED
M
A
N
U
crossflow membrane filtration. J Membrane Sci. 2006;277:258-269.
21
IP T SC R
U
Figure 1. Hollow fiber membrane module: (a), simplified scheme of the tested modules, showing the air
N
injectors coupled to the base of the module, among the hollow fibers; (b), (c) and (d), schematic drawing
A
showing the injectors coupled to the base of the module black circles represent the air injectors and the fibers
A
CC E
PT
ED
M
are evenly potted in the light gray region of the base – with 16, 32 and 64 air injectors, respectively.
22
Figure 2. Flowchart of the permeation system. Blue lines and grey lines stand for permeation streams and backflush streams separately, while dashed grey line represents the air stream. T1 and T2 are the feed and
A
N
U
SC R
IP T
permeate tanks, respectively, V stands for the control and on-off valves and M1 indicates the membrane module.
A
CC E
PT
ED
M
Figure 3. Critical conditions evaluation at constant pressure values. Module 32-650; 5.0 m.s-1.
23
IP T SC R U N A
M
Figure 4. Normalized permeability values for the yeast suspension permeation tests. Module 32-650; 0.3 and 0.7 bar; 0.0, 1.3, 2.7 and 5.0 m.s-1. The symbols indicate the average value of 3 independent experiments and the
A
CC E
PT
ED
error bars indicate the standard deviation of the data.
24
IP T SC R U N A M
Figure 5. Lp.Lpo-1 as function of the cumulative permeated volume (a, c) and cumulative permeated volume
A
CC E
PT
ED
profiles (b, d) for yeasts suspension filtration. Module 32-650; 0.3 (a, b) and 0.7 bar (c, d); 0.0 and 5.0 m.s-1.
Figure 6. Profiles of Rf and Lpfinal as a function of Ug for two of the tested modules.
25
IP T
SC R
Figure 7. Comparison between the Rf obtained at the end of the 8,000 mg.L-1 yeast permeation tests in the
PT
ED
M
A
N
U
modules 16-650, 32-650 and 64-650 at filtration pressure of 0.3 and 0.7 bar.
Figure 8. Comparison between the Rf obtained at the end of the 8,000 mg.L-1 yeast permeation tests using
A
CC E
modules 32-650, 32-800 and 32-1000 at filtration pressures of 0.3 and 0.7 bar.
26
IP T
SC R
Figure 9. Fouling permeation resistance as a function of the aeration power consumption in the permeation tests.
A
CC E
PT
ED
M
A
N
U
The variables Powerair and Rf were normalized from 0 to 1 considering their extreme values.
27
Table1: Technical specifications of the modules Fibers
packing
Number
of Number of air Injectors/Fibers
Membrane area (m2) density (m2.m-3)
fibers
injectors
(%)
16-650
650
0.20
318
16
5.0
32-650
650
0.20
318
32
10.1
64-650
650
0.20
318
64
20.1
32-800
800
0.24
387
32
8.3
32-1000
1,000
0.30
484
32
6.6
IP T
Module code*
A
CC E
PT
ED
M
A
N
U
SC R
* The first two numbers represent the number of air injectors and the remaining numbers represent the packing density of hollow fibers.
28
S0263-8762(18)30274-0 https://doi.org/10.1016/j.cherd.2018.05.035 CHERD 3199
To appear in: Received date: Revised date: Accepted date:
5-3-2018 27-4-2018 25-5-2018
Please cite this article as: Moror´o, Robson Rodrigues, Borges, Cristiano Piacsek, Kronemberger, Frederico de Araujo, New modules for membrane bioreactors: improving fouling control.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2018.05.035 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
New modules for membrane bioreactors: improving fouling control
1
IP T
Robson Rodrigues Mororó1, Cristiano Piacsek Borges1, Frederico de Araujo Kronemberger1,*
Chemical Engineering Program – COPPE/Federal University of Rio de Janeiro, Brazil. P.O.
*
N
U
Corresponding author email: [email protected]
SC R
Box: 68502, ZIP CODE: 21941-972, Rio de Janeiro, RJ.
M
New module for submerged membrane bioreactors with integrated aeration was evaluated;
ED
A
Highlights:
The lower the packing density, the smaller the transport resistance;
A high number of air injectors in the module base resulted in elevated flux values;
Reduction in transport resistance was correlated to the aeration energy expenditure.
CC E
PT
A
Abstract: New hollow fiber membrane modules with air injectors coupled to their base are presented in this paper. The transport resistances were analyzed through permeation tests with yeast cells suspensions varying parameters such as surface air velocity and filtration pressure. Five different modules were investigated, and the lowest resistances were observed in the module with 64 holes in the air injector and packing density equal to 650 m².m-3. The
1
modules with 32 holes in the air injector and packing density of 650 m².m -3 showed the best relation between the efficiency in reducing transport resistance and the aeration energy expenditure.
SC R
IP T
Keywords: Fouling; transport resistance; hydrodynamics; aeration; MBR.
1 Introduction
U
The use of membranes to assist biological treatment of wastewaters, whether industrial
N
or domestic, has enormous potential in the solid-liquid separation, replacing traditional side
A
decanters [1-5]. Membrane Bioreactors (MBR) have many advantages over conventional
M
biological treatment of wastewater, such as the production of better quality effluent and up to 99% removal of pollutant loading, since all suspended biomass is retained by membranes [6].
ED
It is also possible to operate with higher sludge concentration, as the decanting step is
PT
replaced by this physical barrier. Regarding the equipment, MBR are significantly more compact, which is becoming a key issue, especially taking into account that the physical space
CC E
for installation of treatment units is limited in most of the metropolitan areas. The introduction of MBR with submerged membranes (sMBR) significantly reduced energy consumption associated with the operation, compared to the MBR in which membranes are
A
allocated outside the bioreactor (eMBR) [3]. As for all membrane separation processes, fouling problems are crucial in achieving
an efficient sMBR process, taking into account the energy expenditure used to control it [2,3,5-8]. During sMBR operation, flocs containing colloidal particles, macromolecules and microorganisms are driven to the membranes and accumulate in their pores and surface, 2
reducing the permeability, i.e., increasing the transport resistance [3]. To overcome membrane fouling problems, new membranes can be developed or modified using different techniques [9], but several scientific papers in the literature indicate that hydrodynamics is the key parameter to control fouling [5,10-14]. There are numerous recent developments in fouling control in sMBR, including specific turbulence generators [15,16]. But among the factors
IP T
affecting hydrodynamics in an sMBR and, consequently, fouling control, the most studied are air sparging and module design. Air injection is used to produce shear forces in the layer
SC R
deposited on the membrane. Module design is related to the type and position of the air
injector, and the geometry and packing density of membranes. These design variables present a strong influence on aeration patterns [7,14,17]. Hollow fiber membranes are the most
U
commonly used geometry, offering higher filtration area per volume of module, even though
N
helical membrane and folded plate/tilted membrane modules are being experimented in
A
laboratory scale [18-20]. Concerning aeration, most of the published works are related to the
M
influence of air flow rate and the distribution and size of the bubbles in the filtration
ED
resistance [5,14]. Regarding the design of membrane modules, most of the publications are focused on modules with air injectors attached to their bases in a centralized manner, which
PT
can lead to stagnant regions near the membrane, causing intense accumulation of suspended material. Vibrating or rotating hollow fiber membrane modules are being studied recently
CC E
[21,22], but the design of new stationary modules is still required. In the present work, new hollow fiber membrane modules were developed and
A
evaluated. These modules, with several air injectors coupled to their base in a uniform distributed manner, were evaluated aiming at reducing the overall transport resistance. Five modules were tested, with different numbers of air injectors and different packing densities, in order to determine the one with better efficiency, taking into account fouling control and air energy expenditure. For this purpose, transport resistances caused by fouling (Rf) were
3
obtained from the constant pressure permeation tests of fresh yeast suspensions under various conditions, such as different surface air velocities and applied pressures. The suspensions are used as a model to screen and compare different module configurations in order to define which of the five modules could be tested in a MBR.
IP T
2 Methods
SC R
2.1 Hollow fiber membrane modules
The sMBR modules were developed by PAM-Membranas Seletivas Ltda. (Rio de
U
Janeiro, Brazil) with microfiltration poly(ether imide) hollow fiber membranes (pore diameter
A
N
< 0.1 m). The average outer diameter of the fibers was 1 mm and they were fixed to the
M
bottom of the module coupled to the air injectors in a uniform distributed manner. The membranes were allocated within a perforated poly(vinyl chloride) (PVC) pipe with 44 mm
ED
of inner diameter and 200 mm height, together with a PVC screen for protection, preventing the intense movement of the membranes with the rising bubbles, which could lead to
PT
disruption. The modules comprised an inlet for the air at their base and a permeate outlet at
CC E
the top. The basic design of one module and the scheme of the base with the air injectors are presented in Figure 1. Since air is injected in a dispersed and uniform way among the fibers, its distribution is better, enhancing fouling control. Technical specifications of the modules
A
evaluated are shown in Table 1.
4
2.2 Permeation system
The modules were characterized in submerged operation the module was immersed in the feed tank with the suspension to be filtered. The driving force for permeation was obtained using a suction pump, connected to the lumen side of the fibers, maintaining the
IP T
pressure in the feed tank equal to the atmospheric one. Figure 2 shows the flowchart of the
SC R
system used in this study.
N
U
2.3 Permeation test conditions
A
The permeation tests were carried out with suspensions of Saccharomyces cerevisiae
M
cells (Itaiquara®-Tapiratiba/SP) in pure water (distilled, demineralized and microfiltered). The first objective was the individual evaluation of the modules, with respect to the influence
ED
of operational parameters, pressure difference and air superficial velocity (Ug), in the fouling formation. Then, the modules were compared to each other for the evaluation of their design
PT
parameters the number of air injectors and the packing densities. Initially, pure water permeabilities were determined at 0.7 bar of pressure difference – the highest pressure applied
CC E
– to exclude the effect of irreversible pore deformation. Finally, the pure water hydraulic permeability was evaluated. Then, critical conditions were evaluated through filtration
A
experiments at constant pressure values. The critical operational conditions determination – critical flux or critical pressure – is important in order to minimize fouling. Critical pressure can be defined as the pressure beyond which concentration polarization will occur, increasing fouling due to the higher suspended matter concentration in the membrane surface.
5
Permeation tests were performed using suspensions with 8,000 mg.L-1 of the fresh yeast with superficial air velocities (Ug) of 0.0, 1.3, 2.7 and 5.0 m.s-1, at filtration pressures of 0.3 and 0.7 bar, using all modules. The Ug is the ratio between the air flow and the total area of the air injectors. In these tests, the same Ug values were used to ensure that the hydrodynamic conditions near the surface of the membranes were similar, rather than using
IP T
the same air flow rate. The values of the parameters evaluated in this study were based on
SC R
typical literature values for sMBR.
2.4 Transport resistances caused by fouling (Rf)
U
The analysis of the transport resistances caused by fouling (Rf) is a powerful tool for
N
the evaluation of different modules, disregarding membrane morphological characteristics.
A
The model of resistances in series (Equation 1) was considered. Consequently, the total
M
transport resistance (RT) was the sum of membrane intrinsic resistance (Rm), the adsorption
ED
resistance (Ra), the pore blocking resistance (Rpb) and the fouling resistance (Rf), considering that the concentration polarization resistance value was comprised into the fouling resistance,
PT
since it is difficult to dissociate them. The adsorption and pore blocking resistances were
CC E
measured and considered negligible data not shown – while the membrane intrinsic resistance value was obtained from the pure water permeability. Thus, the fouling resistance value (Rf) was obtained by subtracting the membrane resistance from the total transport
A
resistance value.
𝑅𝑇 = 𝑅𝑚 + 𝑅𝑎 + 𝑅𝑝𝑏 + 𝑅𝑓
Equation 1
6
RT was calculated at the end of each permeation test (180 min), using the permeability final value (Lpfinal), according to Equation 2. In this model, the permeate flux in a membrane is described by Darcy's law:
𝐽𝑝𝑠𝑡𝑎𝑏𝑖𝑙𝑖𝑧𝑒𝑑 = 𝐿𝑝𝑓𝑖𝑛𝑎𝑙 . ∆𝑃 =
∆𝑃
Equation 2
𝜇.𝑅𝑇
IP T
where Jp is permeate flow (m3.m-2.h-1), ∆P is the pressure difference across the
SC R
membrane (Pa), μ is the fluid viscosity (Pa.s) and RT is the total resistance (m-1).
U
2.5 Analysis of the overall efficiency of the modules
N
Aeration is responsible for approximately 90% of total energy expenditure of a sMBR,
A
hence special attention should be given to this parameter [5]. The relationship between
M
transport resistance caused by fouling and the energy expenditure in the aeration, named from now on as the overall efficiency of the module, must be optimized. Aeration should not be
ED
used without restriction in fouling control, as the energy demand should be reduced and an excessive aeration could lead to the disruption of some flocs, liberating polymeric material
PT
which could enhance fouling [23]. Therefore, in order to evaluate the overall efficiency of the
CC E
modules, the aeration power in yeast permeation tests, obtained according to Equation 3, was
A
analyzed together with the transport resistance caused by fouling (Rf).
𝑃𝑜𝑤𝑒𝑟𝑎𝑖𝑟 = 𝑃𝑎𝑖𝑟 . 𝑄𝑎𝑖𝑟
Equation 3
where Pair is the air pressure measured at the injector inlet and Qair is the air flow rate. The values of Powerair and Rf were normalized between 0 and 1 and plotted against each other. 7
3 Results and discussion
3.1 Pressure and air velocity influence on filtration resistance
IP T
Prior to the filtration experiments, critical conditions were determined for all five
modules. As an example, the result obtained for module 32-650 at the highest air flow rate is
SC R
presented in Figure 3. Critical conditions were evaluated using yeast suspensions with the same concentration employed in the subsequent experiments and applying different air flow rates. Using the highest air flow rate, two modules, 32-650 and 64-650, presented critical
U
pressures between 0.10 and 0.15 bar. Considering the other modules and lower air velocities,
N
the critical pressure was below 0.10 bar. Since it would not be possible to operate all modules
A
at subcritical conditions using the chosen concentration, all tests were performed above
ED
M
critical conditions.
PT
The average value for the pure water permeability of the fibers, used in the resistances calculations, was 571.2 L.h-1.m-2.bar-1. Concerning the experiments with the yeast
CC E
suspensions, all tests were carried out for 180 min, so that the permeate flux would be constant, indicating the stability of the fouling layer formed. Figure 4 shows the time course
A
profiles of the hydraulic permeability normalized by its initial value (Lp.Lpo-1) for the superficial air velocities (Ug) tested for the module 32-650.
8
The initial permeability values were lower than the pure water hydraulic permeabilities (value not shown) due to the concentration polarization phenomenon, which is established in first moments of test, added to the intense initial fouling. The first observation from the results presented in Figure 4 is that all profiles present the same pattern. The permeability drops promptly in the first 10 minutes, stabilizing after 90 minutes of
IP T
permeation. For all air velocities investigated, it was observed that the flux decline was higher
for the pressure difference of 0.7 bar. This observation could be attributed to the thickness and
SC R
the compactness of the fouling layer. Particle accumulation in the fouling layer was enhanced
as transmembrane pressure increased because more suspended solids were dragged to the membrane surface by the increased permeate flux. Furthermore, the fouling layer could be
U
denser at high transmembrane pressures. Hong et al. [1] studied the impact of transmembrane
N
pressure on fouling during microfiltration tests in an activated sludge submerged membrane
A
bioreactor (sMBR). The authors tested 0.1, 0.2 and 0.3 bar and observed the largest decline in
M
hydraulic permeability (Lp) for 0.3 bar. The permeability reduction for this pressure was 66%
ED
in the first 100 minutes of operation, against 45% and 40% for 0.2 and 0.1 bar, respectively. While in the remainder of the experiments, until 300 minutes, almost no variation in Lp was
0.2 and 0.3 bar.
PT
observed when applying 0.1 bar, a further decrease of about 27% was observed when using
CC E
It is also noticed that increasing air velocity enhances the hydrodynamic conditions
inside the sMBR system, since the permeability values are also raised. Considering the
A
highest pressure difference applied, the permeability values for 150 min of experiment are equal to 14%, 30%, 36% and 41%, in relation to the initial ones, for the air velocities of 0.0, 1.3, 2.7 and 5.0 m.s-1, respectively.
9
Typical results of the influence of aeration on the membrane permeability of the modules during the yeast suspensions filtration are shown in Figure 5, also for the module with 32 air injectors and packing density of 650 m2.m-3.
IP T
It is clearly observed from Figure 5 that the injection of air during the yeast suspension permeation tests in the module 32-650 increased the productivity. The membrane
SC R
permeabilities undergo a greater decline with respect to the accumulated permeate volume obtained during the tests without aeration (0.0 m.s-1) than in the test using air under a superficial velocity of 5.0 m.s-1 (Figure 5, a and c), showing that the latter operation is more
U
stable over time, which would result in lower cleaning frequency. It is interesting to observe
N
that, without air injection, the flux reduces to merely 17% of the initial value with 30 L.m -2 of
A
permeate volume, considering the pressure of 0.7 bar (Figure 5, c). Even with a lower
M
pressure, 0.3 bar (Figure 5, a), the flux is reduced to 28% of the initial one, for the same
ED
volume of permeate. When sparging air, these values increase to 45% and 52%, for 0.7 and 0.3 bar, respectively. This indicates clearly that the aeration, or the hydrodynamics conditions
PT
near the fibers, is a key factor to control fouling in sMBR systems, above all, considering that the fouling could be much more severe when considering the real biological media. Another
CC E
confirmed pattern is the permeate flux dependence with the pressure difference. The greater the pressure difference is, the higher the permeate flux declines. Even considering that the
A
cumulative permeated volume is higher for the pressure difference of 0.7 bar during the whole experiments (Figure 5, b and d), the energy expended per permeated volume, including the membrane cleaning, will not make this high pressure operation economically viable. It is also noticed that the cumulative permeated volume is lower when no aeration was used, mainly for
10
higher filtration times, when the accumulation of yeast particles on the surface of hollow fibers is more severe, increasing the transport resistance.
3.2 Comparison among different module designs
IP T
In order to compare the different module designs, the variation in the transport resistance values caused by fouling (Rf) and the final average permeability of the membranes
SC R
(Lpfinal) are presented in Figure 6 as a function of the superficial velocity of the air bubbles
U
(Ug) in the permeation tests with the yeast suspensions.
N
It can be confirmed that the use of air was effective in reducing fouling on the
A
membranes, as noticed in the transport resistance variation from Ug = 0.0 m.s-1 to Ug = 1.3
M
m.s-1, for all models of the permeator. Nevertheless, subsequent increases in the superficial
ED
velocity of air (Ug) will not lead to a linear augmentation in the average permeability final values (Lpfinal) nor in a linear reduction in the transport resistance caused by fouling (Rf). For
PT
example, for the module 32-650, the variation of Ug from 0.0 to 1.3 m.s-1 caused a decrease of approximately 50% of the Rf value, while the variation from 2.7 to 5.0 m.s-1 produced a
CC E
further decrease of only 14% in the transport resistance caused by fouling. These results are attributed to the turbulence promoted by air bubbles, which creates a field of shear forces on
A
the cake adhered on the surface of the hollow fibers, causing the particles to return to the yeast suspension. The higher Ug, the higher is the velocity of the rising air bubbles and greater the turbulence and the intensity of shear force vectors. But there is a critical value for which the increase of Ug is not accompanied by an increase in the bubbles velocity and the effect on the reduction of fouling is not changed [2,3,5,24-29]. The most probable reasons for the
11
existence of this Ug limit value, or for the plateau in the curves of Rf in function of Ug, may be linked to the fact that the increase of Ug is not completely accompanied by a proportional increase in the rise rate of the bubbles, apart from that the shape of the bubbles may be changing, as well as their size and number, which influences the coalescence. Comparing the filtration pressures tested, it is observed, for 0.7 bar, that the use of air
IP T
becomes less effective in controlling fouling formation due to an imbalance between the shear
stress and the permeation driving force, the latter one being dominant, which causes a higher
SC R
amount of suspended particles to move to the membrane surface and remain adhered to it. Furthermore, the fouling layer porosity can be reduced by compression.
Li et al. [30] treated sewage in a MBR with submerged hollow fibers, at a constant
U
permeate flux equal to 12.5 L.h-1.m-2, and found Rf values above 175 x 1011 m-1 within 20
N
days of filtration, using an air flow rate of 36.0 m³.h-1 per membrane square meter. This
.m-2. Chu and Li [31] found Rf values of 400 x 1011 m-1 after 70 days of operation in a
M
1
A
resistance was reduced to 100 x 1011 m-1 when the aeration rate was increased to 64.8 m³.h-
ED
submerged hollow fiber MBR treating domestic sewage. Lesage et al. [32] also treated sewage in a hollow fiber sMBR using an aeration rate of 1.2 m³.h-1.m-2 and observed Rf values
PT
exceeding 2,500 x 1011 m-1. Han et al. [33] tested submerged flat sheet membranes in a MBR treating synthetic sewage, using air flow of 1.0 m³.h-1.m-2 and a fouling resistance of 1,070 x
CC E
1011 m-1 was observed.
For the new modules studied in the present work, different air flow rates per
A
membrane area were used, depending on the module design. The maximum rates, which correspond to Ug of 5.0 m.s-1 were 1.15, 2.30, 4.60, 1.87 and 1.50 m³.h-1.m-2 for the modules 16-650, 32-650, 64-650, 32-800 and 32-1000, respectively. The resistance values for these modules varied from 21.2 x 1011 to 45.3 x 1011 m-1, but these values cannot be directly
12
compared to the ones presented in the scientific literature since they were obtained using a synthetic suspension, as the main goal of this work was to compare different module designs. In order to assess the modules' performance in relation to the number of air injectors and the packing density of the fibers, the fouling resistance data were plotted again, segregating the data for each of these parameters. The comparison of the values of Rf resulting
IP T
from permeation tests of fresh yeast suspensions using permeators with the same packing density (650 m2.m-3) and the different number of air injectors (16, 32 and 64) is presented in
SC R
Figure 7, for both 0.3 and 0.7 bar.
U
The number of air injectors in the base of the modules has a major influence in fouling
N
control. It is observed that a higher number of air injectors reduces Rf for both filtration
A
pressures investigated. These results are related to the fact that a better distribution of bubbles
M
among the hollow fibers is obtained with a higher number of injectors distributed in the base
ED
of the modules. The homogeneity of the shear forces due to the rise of bubbles among the membranes constitutes a good hydrodynamic pattern and may depend on the proper design of
PT
the aeration system [34]. Mayer et al. [35] recommended the use of complex systems, provided with numerous holes, which would make the distribution of air more homogeneous
CC E
and thus more effective, which is consistent with the results presented in Figure 7. Yet, the modules with more air injectors may generate a greater amount of bubbles in a confined
A
space, increasing coalescence. According to Braak et al. [5], bubbles with large diameters formed by the coalescence of smaller ones promote increased turbulence regions. And the results are notoriously related to the fact that the air flow rate is higher in the modules with more air injectors, for the same air velocity. The energy expenditure is thus higher in these cases, which will be discussed later in this paper.
13
Figure 8 compares the result of Rf obtained in the permeation tests using pressure differences of 0.3 and 0.7 bar, but now analyzing the modules 32-650, 32-800 and 32-1000, which have the same number of air injectors attached to their base (32) and fiber packing
IP T
densities of 650, 800 and 1,000 m2.m-3, respectively.
The increase of fibers packing density severely increases fouling on the surface of the
SC R
membranes, particularly for the highest filtration pressure, which offers the greatest driving
force to the flow, causing a great amount of particles to adhere to the membrane surface and making the fouling layer more compact (less porous). These results are probably due to the
U
fact that increased packing density of the hollow fibers might hinder the contact of bubbles
N
with the inner fibers due to physical blocking, even with the uniform distribution of the
A
injectors, limiting the effects of aeration on the control of fouling. It has to be noted that the
M
increased packing density of fibers reduces the percentage ratio of the air injectors per number
ED
of hollow fibers. Models 32-650, 32-800 and 32-1000 have 10.1%, 8.3% and 6.6% of injectors in relation to the number of fibers, respectively. This also contributes to a worse air
PT
distribution for the modules with increased packing density. For example, at Ug of 5.0 m.s-1, Rf is 29.5 x 1011, 36.4 x 1011 and 45.3 x 1011 m-1 for 10.1%, 8.3% and 6.6% of the number of
CC E
holes in relation to the number of hollow fibers (650, 800 and 1,000 m².m-3), respectively. Among the modules tested, the one that presented the lowest value of Rf in all
A
superficial air velocities (Ug), i.e., the best efficiency in fouling control, was the module 64650, which has the largest number of air injectors (64) and lower packing density, equal to 650 m2.m-3. In this module, it is believed that the hydrodynamic conditions are more appropriate for the fouling control, due to better distribution of bubbles, which excludes the presence of stagnant zones, which have great potential for fouling. The lowest fouling
14
resistance (21.2 x 1011 m-1) was observed for this module with the highest air velocity tested, 5.0 m.s-1.
3.3 Power consumption analysis
IP T
In sMBR, aeration is responsible for approximately 90% of total energy expenditure,
so fouling control should not be investigated regardless of this subject [5]. The control of
SC R
fouling is important to ensure process stability, resulting in good permeate productivity,
decreased number of cleanings and increased lifetime of membranes, reducing the operational costs. In contrast, the operating cost related to the energy spent by the aeration used to control
U
fouling cannot be so high that it would economically derail the process. Therefore, a balance
N
between the benefits of fouling control and the energy expenditure in aeration is required.
A
Since both energy consumption and fouling resistance should be minimized, one possible way
M
to evaluate these variables together is to normalize them from 0 to 1, in order to avoid weight
ED
influence, and then to plot them in one graph. Figure 9 shows the aeration power influence on the transport resistance caused by fouling on the surface of the hollow fibers. The resistance
PT
values were obtained during the fresh yeast suspensions permeation tests in the different
CC E
modules using 0.3 and 0.7 bar.
A
The objective in any permeation analysis in sMBR is to obtain the lowest fouling
resistance spending as little energy as possible. Therefore, analyzing the graphics presented in Figure 9, the objective is to get as close as possible to the dashed line, representing the reduction of both energy consumption and transport resistance. It can be seen that, for both pressures evaluated in the permeation tests, 0.3 and 0.7 bar, the module 32-650 presented the
15
best commitment between the two variables in evidence, Powerair and Rf. Considering the lowest pressure, the modules 32-800 and 16-650 also seem to be considered good alternatives. But for 0.7 bar, only the module 16-650 could be an alternative for the 32-650. These results reflects that, although module 64-650 presents the best hydrodynamic conditions among all modules tested, since the lowest transport resistances (Rf) were observed in the permeation
IP T
tests performed with this module, it does not present the best commitment between the
efficiency in controlling fouling and the energy expenditure, i.e., the overall efficiency. The
SC R
lowest resistance values observed for this module was accompanied by a humongous energy expenditure, which would not desired when applying it in a sMBR.
N
U
4 Conclusions
A
New modules for submerged membrane bioreactors (sMBR) with air injectors attached to
M
their base were compared in terms of fouling resistance and aeration energy expenditure.
ED
Increasing pressure in all permeation tests with model suspensions of fresh yeast cells caused a greater decrease in the permeability throughout the experiments, but the air injection could
PT
reduce this effect in both pressures tested. However, the relation between the average permeability of the membranes (Lpfinal) and the superficial velocity of the air bubbles (Ug) is
CC E
nonlinear: there is a limiting value for which the increase of Ug does not represent a further influence in the Lpfinal value. Concerning the comparison among different modules, model 64-
A
650 (with 64 air injectors and 650 m2.m-3 of packing density) presented the lowest transport resistances caused by fouling (Rf) among all the models tested, for all superficial air velocities (Ug). The lowest value (21.2 x 1011 m-1) was observed for this module with the highest air velocity, 5.0 m.s-1. This indicated that the best results in terms of fouling resistance would be achieved with lower packing density and higher air injectors number. But since the energy
16
spent in air sparging corresponds of the majority of the operational costs of a sMBR, this simplistic evaluation is not completely accurate. Through an aeration power/resistance analysis, the module 32-650 presented the best ratio between the efficiency in controlling fouling resistance (Rf) and energy expenditure (Powerair). The decision on whether to use one
operational parameters, related to the real biological medium operation.
SC R
5 Acknowledgements
IP T
or another of these modules and about the operating conditions will rely on specific
R. R. Mororó would like to thank CAPES - Coordenação de Aperfeiçoamento de Pessoal de
U
Nível Superior (Brazil) for his scholarship. The authors would like to thank PAM-Membranas
A
CC E
PT
ED
M
A
N
Seletivas (Rio de Janeiro, Brazil) for gently providing the modules used in this study.
17
References [1] Hong SP, Bae TH, Tak TM, et al. Fouling control in activated sludge submerged hollow fiber membrane bioreactors. Desalination. 2002;143:219-228. [2] Kim J, Chang I, Shin D, Park H. Membrane fouling control through the change of the depth of a membrane module in a submerged membrane bioreactor for advanced wastewater
IP T
treatment. Desalination. 2008;231:35-43.
[3] Park H, Lee YH, Kim H, et al. Reduction of membrane fouling by simultaneous upward
SC R
and downward air sparging in a pilot-scale submerged membrane bioreactor treating municipal wastewater. Desalination. 2010;251:75-82.
[4] Kraume M, Drews A. Membrane bioreactors in waste water treatment – status and trends.
U
Chem Eng Technol. 2010;33(8):1251-1259.
N
[5] Braak E, Alliet M, Schetrite S, Albasi C. Aeration and hydrodynamics in submerged
A
membrane bioreactors. J Membrane Sci. 2011;379:1-18.
M
[6] Judd S. Principles and Applications of Membrane Bioreactors in Water and Wastewater
ED
Treatment. San Diego (CA): Elsevier; 2006.
[7] Viana PZ, Nobrega R, Jordão EP, Azevedo JPS. Optimizing the operational conditions of
2005;48:119.
PT
a membrane bioreactor used for domestic wastewater treatment. Braz Arch Biol Techn.
CC E
[8] Viero AF, Demelo T, Torres A, et al. The effects of long-term feeding of high organic loading in a submerged membrane bioreactor treating oil refinery wastewater. J Membrane
A
Sci. 2008;319:223-230. [9] Deowan SA, Galiano F, Hoinkis J, et al. Novel low-fouling membrane bioreactor (MBR) for industrial wastewater treatment. J Membrane Sci. 2016;510:524-532. [10] Evenblij H, Geilvoet S, van der Graaf JH, van der Roest HF. Filtration characterization for assessing MBR performance: three cases compared. Desalination. 2005;178:115-124.
18
[11] Brannock MW, De Wever H, Wang Y, Leslie G. Computational fluid dynamics simulations of MBRs: Inside submerged versus outside submerged membranes. Desalination. 2009;236:244-251. [12] Marselina Y, Lifia, Le-Clech P, et al. Characterization of membrane fouling deposition and removal by direct observation technique. J Membrane Sci. 2009;341:163-171.
IP T
[13] Khalili-Garakani A, Mehrnia MR, Mostoufi N, Sarrafzadeh MH. Analyze and control
fouling in an airlift membrane bioreactor: CFD simulation and experimental studies. Process
SC R
Biochem. 2011;46:1138-1145.
[14] Ratkovich N, Naessens W, Maere T, et al. Critical review of membrane bioreactor models – Part 2: Hydrodynamic and integrated models. Bioresource Technol. 2012;122:107-
U
118.
N
[15] Pourbozorg M, Li T, Law AW. Effect of turbulence on fouling control of submerged
A
hollow fibre membrane filtration. Water Res. 2016;99:101-111.
M
[16] Krzeminski P, Leverette L, Malamis S, Katsou E. Membrane bioreactors – A review on
ED
recent developments in energy reduction, fouling control, novel configurations, LCA and market prospects. J Membrane Sci. 2017;527:207-227.
PT
[17] Liu X, Wang Y, Waite TD, Leslie G. Numerical simulations of impact of membrane module design variables on aeration patterns in membrane bioreactors. J Membrane Sci.
CC E
2016;520:201-213.
[18] Jie L, Liu L, Yang F, et al. The configuration and application of helical membrane
A
modules in MBR. J Membrane Sci. 2012;392-393:112-121. [19] Zhang H, Zhang J, Jiang W, et al. A new folded plate membrane module for hydrodynamic characteristics improvement and flux enhancement. Sep Purif Technol. 2015;154:36-43.
19
[20] Eliseus A, Bilad MR, Nordin NAHM, et al. Tilted membrane panel: A new module concept to maximize the impact of air bubbles for membrane fouling control in microalgae harvesting. Bioresource Technol. 2017;241:661-668. [21] Ruigómez I, Vera L, González E, Rodríguez-Sevilla J. Pilot plant study of a new rotating
J Membrane Sci. 2016;514:105-113.
IP T
hollow fibre membrane module for improved performance of an anaerobic submerged MBR.
[22] Zsirai T, Qiblawey H, A-Marri MJ, Judd S. The impact of mechanical shear on
SC R
membrane flux and energy demand. J Membrane Sci. 2016;516:56-63.
[23] Wisniewski C, Grasmick A, Leon Cruz A. Critical particle size in membrane bioreactors: Case of a denitrifying bacterial suspension. J Membrane Sci. 2000;178(1-2):141-150.
U
[24] Ueda T, Hata K, Kikuoka Y, Seino O. Effects of aeration on suction pressure in a
N
submerged membrane bioreactor. Water Res. 1997;31(3):489-494.
A
[25] Cabassud C, Laborie S, Lainé JM. How Slug Flow can Improve Ultrafiltration Flux in
M
Organic Hollow Fibres? J Membrane Sci. 1997;128:93-101.
ED
[26] Chang S, Fane AG. The effect of fibre diameter on filtration and flux distribution– relevance to submerged hollow fibre modules. J Membrane Sci. 2001;184:221-231.
PT
[27] Le-Clech P, Chen V, Fane AG. Fouling in membrane bioreactors used in wastewater treatment. J Membrane Sci. 2006;284:17-53.
CC E
[28] Cui ZF, Chang S, Fane AG. The use of gas bubbling to enhance membrane processes. J Membrane Sci. 2003;221:1-35.
A
[29] Sofia A, Ng WJ, Ong SL. Engineering design approaches for minimum fouling in submerged MBR. Desalination. 2004;160:67-74. [30] Li Y, He Y, Liu Y, et al. Comparison of the filtration characteristics between biological powdered activated carbon sludge and activated sludge in submerged membrane bioreactors. Desalination. 2005;174:305-314.
20
[31] Chu HP, Li XY. Membrane fouling in a membrane bioreactor (MBR): Sludge cake formation and fouling characteristics. Biotechnol Bioeng. 2005;90:323-331. [32] Lesage N, Spérandio M, Cabassud C. Performances of a hybrid adsorption/submerged membrane biological process for toxic waste removal. Water Sci Technol. 2005;51:173-180. [33] Han SS, Bae TH, Jang GG. Tak, T.M., Influence of sludge retention time on membrane
IP T
fouling and bioactivities in membrane bioreactor system. Process Biochem. 2005;40:23932400.
SC R
[34] Fane AG, Yeo A, Law A, et al. Low pressure membrane processes – doing more with less energy. Desalination. 2005;185:159-165.
[35] Mayer M, Braun R, Fuchs W. Comparison of various aeration devices for air sparging in
A
CC E
PT
ED
M
A
N
U
crossflow membrane filtration. J Membrane Sci. 2006;277:258-269.
21
IP T SC R
U
Figure 1. Hollow fiber membrane module: (a), simplified scheme of the tested modules, showing the air
N
injectors coupled to the base of the module, among the hollow fibers; (b), (c) and (d), schematic drawing
A
showing the injectors coupled to the base of the module black circles represent the air injectors and the fibers
A
CC E
PT
ED
M
are evenly potted in the light gray region of the base – with 16, 32 and 64 air injectors, respectively.
22
Figure 2. Flowchart of the permeation system. Blue lines and grey lines stand for permeation streams and backflush streams separately, while dashed grey line represents the air stream. T1 and T2 are the feed and
A
N
U
SC R
IP T
permeate tanks, respectively, V stands for the control and on-off valves and M1 indicates the membrane module.
A
CC E
PT
ED
M
Figure 3. Critical conditions evaluation at constant pressure values. Module 32-650; 5.0 m.s-1.
23
IP T SC R U N A
M
Figure 4. Normalized permeability values for the yeast suspension permeation tests. Module 32-650; 0.3 and 0.7 bar; 0.0, 1.3, 2.7 and 5.0 m.s-1. The symbols indicate the average value of 3 independent experiments and the
A
CC E
PT
ED
error bars indicate the standard deviation of the data.
24
IP T SC R U N A M
Figure 5. Lp.Lpo-1 as function of the cumulative permeated volume (a, c) and cumulative permeated volume
A
CC E
PT
ED
profiles (b, d) for yeasts suspension filtration. Module 32-650; 0.3 (a, b) and 0.7 bar (c, d); 0.0 and 5.0 m.s-1.
Figure 6. Profiles of Rf and Lpfinal as a function of Ug for two of the tested modules.
25
IP T
SC R
Figure 7. Comparison between the Rf obtained at the end of the 8,000 mg.L-1 yeast permeation tests in the
PT
ED
M
A
N
U
modules 16-650, 32-650 and 64-650 at filtration pressure of 0.3 and 0.7 bar.
Figure 8. Comparison between the Rf obtained at the end of the 8,000 mg.L-1 yeast permeation tests using
A
CC E
modules 32-650, 32-800 and 32-1000 at filtration pressures of 0.3 and 0.7 bar.
26
IP T
SC R
Figure 9. Fouling permeation resistance as a function of the aeration power consumption in the permeation tests.
A
CC E
PT
ED
M
A
N
U
The variables Powerair and Rf were normalized from 0 to 1 considering their extreme values.
27
Table1: Technical specifications of the modules Fibers
packing
Number
of Number of air Injectors/Fibers
Membrane area (m2) density (m2.m-3)
fibers
injectors
(%)
16-650
650
0.20
318
16
5.0
32-650
650
0.20
318
32
10.1
64-650
650
0.20
318
64
20.1
32-800
800
0.24
387
32
8.3
32-1000
1,000
0.30
484
32
6.6
IP T
Module code*
A
CC E
PT
ED
M
A
N
U
SC R
* The first two numbers represent the number of air injectors and the remaining numbers represent the packing density of hollow fibers.
28