You are on page 1of 41

Ministère de l'enseignement supérieur

et de la Recherche Scientifique
*** * ***
Université de Carthage
*** * ***
Institut National des Sciences
Appliquées Sciences et Technologies

End-of-year project
Industrial Chemistry
4th year

Modeling and Optimization of Transmembrane Pressure


in Aerobic Membrane Bioreactors

Submitted by: Supervised by:



Iheb Driouech Ms. Fatma Ellouze


Zeineb Khlifi

Defended on June 5, 2023

Jury Members:

Ms. Fatma Ellouze


Ms. Souhir Abdelmoula

Academic year: 2022 / 2023


Acknowledgements

We would like to express our sincere gratitude to our supervisor, Ms. Fatma Ellouze, for her
invaluable guidance, support, and expertise throughout the course of this work. Her dedication
and insightful feedback have been instrumental in shaping the direction of this report.

We would also like to thank Ms. SouhirAbdelmoula for taking the time to attend our
presentation and for her support and encouragement. We are grateful for her presence and
interest in our findings.

We would also like to extend our sincere gratitude to the professors and faculty of INSAT
(Institut National des Sciences Appliquées et de Technologie) for their invaluable contribution
to our work. Their expertise, extensive knowledge, and theoretical insights have played a
crucial role in enhancing the effectiveness of our research.

Lastly, we would like to acknowledge our friends and family's unwavering support and
encouragement throughout this research endeavor. Their belief in us and their constant
encouragement has been a source of motivation during this journey.

Without the support and guidance of Ms. Fatma Ellouze and the contributions of all those
mentioned above, this research would not have been possible. We are truly grateful for their
assistance.

Thank you.

Iheb Driouech

Zeineb Khlifi

i
Table of Contents

GENERAL INTORDUCTION .................................................................................................. 1

I. LITERATURE REVIEW ................................................................................................... 3

Introduction ............................................................................................................................ 3

I.1 Aerobic Membrane Bioreactors (AMR) Technology .................................................. 3

I.2 Fouling Mechanisms in Aerobic Membrane Bioreactors (AMRs) ............................. 4

I.3 Factors Affecting Membrane Fouling ......................................................................... 6

Conclusion .............................................................................................................................. 8

II. MATHEMATICAL MODELING OF MEMBRANE FILTRATION ........................... 9

Introduction ............................................................................................................................ 9

II.1 Mathematical Modeling of Membrane Filtration ........................................................ 9

II.1.1 Filtration Phase ..................................................................................................... 9

II.1.2 Backwash Phase ................................................................................................... 9

II.1.3 Combined Filtration Dynamics .......................................................................... 10

II.1.4 Transmembrane Pressure in Constant Flux Filtration ........................................ 10

II.2 MATLAB Simulation and Computational Tools ...................................................... 10

Conclusion ............................................................................................................................ 12

III. Experimental References ............................................................................................... 13

Introduction .......................................................................................................................... 13

III.1 Characteristics of the Effluent and Influent in AMRs ............................................... 13

III.2 Experimental Input Data ............................................................................................ 14

Conclusion ............................................................................................................................ 16

IV. Simulation of Transmembrane Pressure ....................................................................... 17

Introduction .......................................................................................................................... 17

IV.1 Parameters to Optimise .............................................................................................. 17

IV.1.1 Specific Cake Resisitance .................................................................................. 17

ii
IV.1.2 Shear Force ......................................................................................................... 17

IV.1.3 Cake BackwashingEfficiency ............................................................................ 18

IV.2 Results and Discussion .............................................................................................. 18

IV.2.1 Filtration Simulation .......................................................................................... 18

IV.2.2 Filtration and Backwash Simulation .................................................................. 21

Conclusion ............................................................................................................................ 23

V. Modeling of Transmembrane Pressure ......................................................................... 24

Introduction .......................................................................................................................... 24

V.1 Error Analysis ............................................................................................................ 24

V.2 Global ParameterOptimization .................................................................................. 25

V.2.1 Coding Approach ............................................................................................... 25

V.2.2 Resultsand Discussion ........................................................................................ 26

V.3 Time-DividedParameterOptimization ....................................................................... 27

V.3.1 Coding Approach .............................................................................................. 27

V.3.2 Results and Discussion ...................................................................................... 28

Conclusion ............................................................................................................................ 31

General Conclusion and Perspectives ...................................................................................... 32

Appendix .................................................................................................................................. 33

Bibliography ............................................................................................................................. 34

iii
List of figures
Figure I-1 Immersed Aerobic Membrane Reactor Configuration [1]. .................................................................... 3

Figure I-2 Fouling Mechanisms in Membrane Filtration [2]. ................................................................................. 5

Figure I-3 Factors Affecting Membrane Fouling .................................................................................................... 7

Figure I-4 Membrane Backwash Mechanism [4]. ................................................................................................... 7

Figure II-1 MATLAB Logo [5]. ........................................................................................................................... 11

Figure III-1 Time-dependent Variation in Processed Transmembrane Pressure ................................................... 15

Figure IV-1 Transmembrane Pressure Evolution as Function of Time for 5 minutes Filtration. .......................... 19

Figure IV-2 Adjusted Transmembrane Pressure Evolution as Function of Time for 5 minutes Filtration. .......... 20

Figure IV-3 Transmembrane Pressure Evolution as Function of Time for 30 minutes Filtration. ........................ 21

Figure IV-4 Evolution of SimulatedTransmembrane Pressure over Time Throughout the Entire Process. .......... 22

Figure V-1 Evolution of ModeledTransmembrane Pressure over Time Throughout the Entire Process. ............. 26

Figure V-2 Time-divided Optimization of Modeled and experimental Transmembrane Pressure over Time. ..... 29

iv
List of Tables
Table III-1 Input Data for The Simulation ............................................................................................................ 15

Table IV-1 Input Parameters for Transmembrane Pressure Simulation. ............................................................... 22

Table V-1Fmincon Function Input Arguments for Global Optimisation. ............................................................. 26

Table V-2 Global Parameters OptimisationResults. ............................................................................................. 27

Table V-3Fmincon Function Input Arguments for Time-divided Optimisation. .................................................. 28

Table V-4 Time-divided Parameters Optimisation Results. .................................................................................. 30

v
List of Abbreviations

MBR Membrane Bioreactor

AMR Aerobic Membrane Bioreactor

RIS Resistance In Series

EPS Extracellular Polymeric Substances

TMP Transmembrane Pressure

COD Chemical Oxygen Demand

m Meter

kg Kilogram

s Second

Pa Pascal

L Liter

mg Milligram

°C Celcius degrees

vi
GENERAL INTORDUCTION

he ever-increasing demands of industrialization, urbanization, and population

T growth have led to an alarming surge in wastewater production in today's society.


This wastewater, laden with a diverse range of chemical pollutants, poses grave
environmental and health risks. Urgent prioritization of wastewater treatment is
paramount to safeguarding the well-being of both human beings and the fragile ecosystem.

Within wastewater, a myriad of pollutants such as heavy metals, pesticides, pharmaceuticals,


and organic compounds persist and wreak havoc on aquatic life, soil quality, and human
health when left untreated in the environment. Their longevity facilitates bioaccumulation and
biomagnification, disrupting the delicate balance of the food chain.

Addressing this critical issue necessitates the adoption of advanced treatment technologies
that enhance the efficiency of wastewater treatment. Two highly effective methods, namely
the aerobic membrane bioreactor (MBR) and biological sludge treatment, have emerged as
crucial approaches to removing chemical pollutants from wastewater.

Efficiency assumes a pivotal role in wastewater treatment, particularly in the context of


pollutant removal. Striving for high removal rates while minimizing energy consumption and
operational costs becomes paramount for the sustainable management of wastewater
treatment facilities. Implementing advanced treatment technologies such as aerobic MBR and
biological sludge treatment can significantly bolster the efficiency of the treatment process,
ensuring the effective removal of chemical pollutants and curbing environmental impact.

In conclusion, prioritizing the treatment of wastewater contaminated with chemical pollutants


is an urgent matter in modern society. By prioritizing the implementation of effective
treatment methods like aerobic MBR and biological sludge treatment, we can mitigate the
risks associated with chemical pollution and safeguard the integrity of our environment and
the health of future generations. This thesis report aims to explore this prioritization,
examining the principles, mechanisms, and applications of these treatment methods. Through
comprehensive analysis and a review of pertinent case studies and scientific literature, this
study aims to provide valuable insights and practical recommendations for policymakers,

1|Page
researchers, and practitioners involved in wastewater treatment and environmental
management. By taking proactive steps to address this pressing issue, we can pave the way for
sustainable development and ensure a cleaner, healthier future for all.

The report will be structured into four chapters to provide a comprehensive analysis of the
topic. The first chapter will consist of a literature review aiming to explore the theoretical
principles underlying aerobic membrane bioreactors and gain in-depth insights into the
challenges intended for optimization.

Moving on to the second chapter, it will introduce the mathematical modeling of the filtration
membrane's functioning in aerobic bioreactors. This chapter will establish the foundational
groundwork for modeling and optimizing the performance of the filtration membrane.

Chapter three will focus on the simulation work conducted for the filtration and backwash
cycles of the membrane. The results obtained from the simulations will be discussed,
providing valuable insights derived from the analysis.

Lastly, the fourth and final chapter will present the culmination of the efforts in modeling the
membrane's functionality and optimizing the intervening parameters. This chapter will
showcase the outcomes of the work on the membrane's operation and the optimization of
associated parameters.

By following this structured approach, the report aims to provide a comprehensive


understanding of the topic, from the theoretical foundations to practical insights gained
through simulation and optimization.

2|Page
I. LITERATURE REVIEW

Introduction

The first chapter of this report establishes the foundation for the study by presenting an
extensive understanding of the theoretical background. The focus is on the technology of
Aerobic Membrane Bioreactors (AMRs), examining the mechanisms behind membrane
fouling and its primary contributing factors. These fouling mechanisms encompass pore
constriction and cake formation, while the biological and physical factors that influence
fouling will also be discussed. This acquired knowledge serves as a robust base for devising
strategies to enhance AMR performance and alleviate fouling, ultimately promoting more
efficient and sustainable wastewater treatment processes.

I.1 Aerobic Membrane Bioreactors (AMR) Technology

Aerobic Membrane Bioreactors (AMRs) are an innovative wastewater treatment technology


that combines the benefits of the activated sludge process and membrane filtration. The
activated sludge process, a widely used biological treatment method, relies on the activity of
microorganisms to break down organic pollutants in wastewater. On the other hand,
membrane filtration serves as a physical barrier, effectively separating suspended solids,
bacteria, and other contaminants from the treated water.

Figure I-1 Immersed Aerobic Membrane Reactor Configuration [1].

3|Page
What sets AMRs apart from conventional treatment systems is the integration of these two
processes within a single tank. In an AMR, the filtration membrane is directly immersed in
the activated sludge tank, facilitating the simultaneous removal of biomass and purification of
water. This configuration eliminates the need for a separate tank for filtration, streamlining
the treatment process and reducing the system's overall footprint.

The term "aerobic" in AMRs refers to the provision of oxygen to the microorganisms
involved in the treatment process. Air is injected into the bottom of the tank, creating a highly
oxygenated environment that enables the microorganisms to carry out their metabolic
activities effectively.

Combining the activated sludge process and membrane filtration in AMRs offers several
advantages. Firstly, the membrane acts as a physical barrier, preventing the discharge of
suspended solids, bacteria, and other contaminants into the treated water, producing a higher-
quality effluent. Secondly, the direct immersion of the membrane in the activated sludge tank
promotes the formation of a highly active microbial community, enhancing the
biodegradation of pollutants and achieving efficient wastewater treatment. Lastly, the
compact design of AMRs reduces the system's footprint, making it suitable for both new
installations and retrofits in existing wastewater treatment plants.

I.2 Fouling Mechanisms in Aerobic Membrane Bioreactors (AMRs)

Membrane fouling is a significant challenge that impacts the performance of Aerobic


Membrane Bioreactors (AMRs) by decreasing permeate flux in constant-pressure filtration or
increasing transmembrane pressure in constant-flux filtration. It occurs when particles from
the treated wastewater accumulate on the membrane surface or get trapped within its pores,
thereby diminishing the permeate flux, which represents the volume of treated water that
passes through the membrane. This accumulation of obstacles on the membrane surface or
within its pores hampers the volume of treated water that can flow through the membrane,
ultimately reducing the overall system efficiency.

There are three primary mechanisms by which membrane fouling can occur in AMRs: pore
constriction, pore blocking, and cake formation. Pore constriction takes place when particles,
significantly smaller than the pore size, enter the membrane and cause the pores to shrink,
impeding water flow. Pore blocking occurs when particles of similar size to the pores
accumulate on the membrane surface, leading to complete obstruction of the pores.

4|Page
Additionally, intermediate blocking can occur when particles partially block the pores,
resulting in reduced water flow through the membrane. Lastly, cake formation happens when
particles larger than the pore size accumulates on the membrane surface, forming a layer that
hampers the effectiveness of membrane filtration.

Figure I-2 Fouling Mechanisms in Membrane Filtration [2].

These fouling mechanisms collectively contribute to the resistance encountered during


filtration. As particles accumulate on the membrane surface or within its pores through pore
constriction, pore blocking, and cake formation, the overall resistance to filtration increases.
The total resistance to filtration can be expressed by Equation (1), derived from Darcy’s law:

𝑇𝑀𝑃
𝑅𝑡𝑜𝑡 =
𝐽. 𝜇

Where :

𝐽 : represents the permeate flux (𝑚3 . 𝑚−2 . 𝑠 −1)

𝑇𝑀𝑃 : represents the transmembrane pressure (𝑃𝑎)

𝜇 : represents the permeate viscosity (𝑃𝑎. 𝑠)

𝑅𝑡𝑜𝑡 : is the total resistance to filtration (𝑚−1)

Darcy’s law enables the calculation of the total resistance as a combination of the original
membrane resistance, Rm, and the resistances deriving from the different fouling
mechanisms.

5|Page
Therefore, by applying the RIS model, the total resistance of the fouled membrane can be
quantified as the sum of the intrinsic resistance of the clean membrane Rm and the fouling
resistance Rf. The latter, as aforementioned, can be split into different contributions due to the
specific fouling mechanisms. The former approach, mostly adopted at the beginning,
expresses the fouling resistance Rf as the sum of the cake deposition Rc, and internal fouling
(or pore blocking) [3].

𝑅𝑡𝑜𝑡 = 𝑅𝑚 + 𝑅𝑓

𝑅𝑡𝑜𝑡 = 𝑅𝑚 + 𝑅𝑐 + 𝑅𝑃𝐵

Understanding the factors that contribute to membrane fouling is indeed crucial for
minimizing its effects and optimizing the operating conditions of Aerobic Membrane
Bioreactors (AMRs). These factors can be broadly categorized into biological and physical
factors, each playing a significant role in the fouling process.

I.3 Factors Affecting Membrane Fouling

To comprehensively address the challenges of membrane fouling in Aerobic Membrane


Bioreactors (AMRs), it is essential to examine the contributing factors from both biological
and physical perspectives.

Biological factors play a significant role in membrane fouling within AMRs. The growth of
microorganisms, the formation of biofilms, and the production of extracellular polymeric
substances (EPS) contribute to fouling. Microbial growth on the membrane surface leads to
the development of biofilms, which decrease permeability EPS, produced by microorganisms,
can accumulate on the membrane and contribute to pore blocking. Furthermore, metabolites
and by-products from microbial activities can absorb onto the membrane surface, further
exacerbating fouling.

Effectively understanding and managing these biological factors are crucial for mitigating
fouling and optimizing the performance of AMRs.

In addition to biological factors, flow dynamics, particularly the flow rate and its impact on
transmembrane pressure (TMP), significantly influence membrane fouling in AMRs. Flow
rate within an AMR plays a critical role in determining the extent of fouling. Higher flow
rates result in increased TMP, intensifying the driving force for particle deposition and cake
formation, thus promoting fouling.

6|Page
Figure I-3 Factors Affecting Membrane Fouling

The choice of backwashing method in AMRs also plays a vital role in membrane fouling.
Backwashing is a widely used technique to mitigate fouling by flushing the membrane surface
and dislodging accumulated particles and biofilm. Different backwashing methods, such as air
scouring, water backwashing, or a combination of both, can be employed. The effectiveness
of backwashing depends on various physical factors, including the duration of filtration
cycles, the frequency and duration of backwashing, and the flow velocity during the process.
Inadequate backwashing, such as shorter durations or lower flow velocities, may lead to
higher fouling rates due to incomplete removal of deposited materials. Conversely, optimizing
backwashing parameters, such as longer durations, higher flow velocities, and adequate
frequency, can enhance fouling mitigation and improve the overall performance of the
membrane system.

Figure I-4 Membrane Backwash Mechanism [4].

7|Page
Effectively understanding and managing these physical factors, including flow dynamics,
transmembrane pressure, and backwashing methods, are crucial for minimizing membrane
fouling in AMRs. By implementing appropriate backwashing strategies and optimizing flow
dynamics, operators can reduce fouling rates, enhance system efficiency, and extend the
lifespan of the membrane, ensuring sustained performance in wastewater treatment processes.

Conclusion

In conclusion, this first chapter has established a solid foundation for the study by providing
an extensive understanding of the theoretical background of Aerobic Membrane Bioreactors
(AMRs). The focus has been on examining the mechanisms behind membrane fouling and
identifying its primary contributing factors. These fouling mechanisms encompass pore
constriction and cake formation, while the biological and physical factors that influence
fouling have also been discussed.It is evident that Aerobic Membrane Bioreactors (AMRs) are
effective technologies for wastewater treatment. However, the challenge of membrane fouling
remains a significant issue. Therefore, it is crucial to simulate and model this fouling
phenomenon to optimize AMR performance in the future. By gaining a comprehensive
understanding of the fouling mechanisms and factors, strategies can be devised to enhance
AMR efficiency and alleviate fouling, ultimately promoting more efficient and sustainable
wastewater treatment processes.

8|Page
II. MATHEMATICAL MODELING OF MEMBRANE FILTRATION

Introduction

This chapter aims to provide a comprehensive understanding of the factors influencing the
performance of membrane filtration in aerobic membrane bioreactors (AMRs) by examining
specific operating conditions and utilizing MATLAB simulations. By representing the
operating conditions in our case study, we can gain insights into the key factors that impact
membrane filtration. Furthermore, we will explore how MATLAB simulations can serve as a
powerful tool to optimize the performance of membrane filtration in AMRs, offering practical
insights and strategies.

II.1 Mathematical Modeling of Membrane Filtration

A simplified mathematical model will be utilized to describe the dynamics of constant flux
filtration. The primary focus of this model is to capture the accumulation and decay of cake
on the membrane surface, aiming to simplify the complexity of the filtration process. Below is
an overview of the model and its approximations.

II.1.1 Filtration Phase

The filtration phase is characterized by the deposition of matter on the membrane surface. We
will begin by examining the evolution of the deposited mass over time, denoted as
𝑚(𝑡) (𝑘𝑔. 𝑚²). This evolution can be described by Equation (2), which relates the deposited
mass to the concentration of suspended matter 𝐶 (𝑘𝑔. 𝑚−3 ) , volumetric
flux 𝐽𝑣 (𝑚3 . 𝑚−2 . 𝑠 −1 ), and shear force β (m2.𝑘𝑔−1 ):

𝑚 = 𝑓𝑓 (𝑚) = −𝐶. 𝐽𝑣. 𝛽 × 𝑚 + 𝐶. 𝐽𝑣

II.1.2 Backwash Phase

Following the filtration phase, the backwash phase occurs, during which the deposited mass
on the membrane surface decreases. We will analyze the dynamics of this decrease using
Equation (3):

𝑚 = 𝑓𝑏𝑤 (𝑚) = −𝑎 × 𝑚

9|Page
In Equation (3), the parameter 𝑎(𝑠 −1 ) reflects the efficiency of cake backwashing, influenced
by the cake resistance and the intensity of backwashing.

II.1.3 Combined Filtration Dynamics

To fully understand the overall dynamics of the membrane filtration process, we consider a
control variable 𝑢(𝑡) that takes the value of 1 during the filtration phase and -1 during the
backwash phase. By incorporating this control variable, we can express the evolution of the
mass as follows in Equation (4):

1+𝑢 1−𝑢
𝑚̇ = 𝑓𝑓 (𝑚) + 𝑓 (𝑚)
2 2 𝑏𝑤

Here, 𝑚̇ represents the derivative of the mass with respect to time. The initial condition
𝑚(0) = 𝑚0 is given, where 𝑚0 is the initial mass.

II.1.4 Transmembrane Pressure in Constant Flux Filtration

In a constant flux filtration system, the accumulation of matter on the membrane surface leads
to an increase in transmembrane pressure (TMP) that can beexpressed by Equation (5).

𝑄
𝑇𝑀𝑃(𝑚) = 𝜇𝑅 (𝑚)
𝐴 𝑡

As previously stated, the total resistance is expressed as the sum of the intrinsic resistance of
the clean membrane Rm, the cake deposition Rc, and the resistance to internal fouling RPB.
However, in this mathematical model, the resistance to internal fouling, RPB, is not accounted
for in such a way that:

𝑅𝑡 = 𝑅𝑚 + 𝑅𝑐

𝑅𝑡 = 𝑅𝑚 + 𝛼𝑚

Where :

𝛼 : Specific cake resistance (𝑚. 𝑘𝑔−1 )

II.2 MATLAB Simulation and Computational Tools

MATLAB, an abbreviation for "MATrixLABoratory," is a widely used software tool in


various scientific and engineering domains. It offers a comprehensive environment for
numerical computation, data visualization, and algorithm development. MATLAB provides

10 | P a g e
researchers with a user-friendly interface to perform complex calculations, analyze data, and
simulate systems efficiently. With its extensive library of built-in functions, MATLAB allows
users to tackle a wide range of tasks, including mathematical modeling, signal processing,
image analysis, and control system design. Its versatile programming language enables the
implementation of custom algorithms and automation of repetitive tasks, enhancing
productivity and reproducibility. The graphical capabilities of MATLAB facilitate the visual
representation of data, enabling intuitive interpretation and presentation of research findings.
MATLAB is a suitable choice for modeling and optimizing transmembrane pressure over
time due to its comprehensive mathematical modeling and optimization functions. It provides
researchers with a versatile platform for integrating numerical simulations, data analysis, and
visualization tasks, streamlining the workflow. MATLAB's user-friendly interface and
programming language allow for rapid prototyping and easy debugging, ensuring efficient
development and refinement of models. Furthermore, MATLAB's widespread adoption and
active community support provide access to extensive resources, including documentation,
tutorials, and user forums, enhancing researchers' productivity, accuracy, and effectiveness in
addressing the complexities of transmembrane pressure modeling and optimization. In the
study, two important MATLAB functions, namely 'fmincon' and 'ODE45,' were utilized to
enhance the modeling and optimization of transmembrane pressure over time. The 'fmincon'
function served as a robust optimization tool to minimize pressure variations while
considering system constraints. By defining the objective function, constraints, and bounds,
the iterative search capability of 'fmincon' was leveraged to find the optimal solution.
Additionally, the 'ODE45' function played a crucial role in solving the differential equations
involved in the transmembrane pressure model. 'ODE45' enabled numerical simulation and
analysis of the system's dynamic behavior, providing valuable insights into pressure evolution
over time. The combined use of MATLAB, 'fmincon,' and 'ODE45' allowed for effective
exploration, modeling, optimization, and analysis of transmembrane pressure, contributing to
a deeper understanding of the research subject [4].

Figure II-1 MATLAB Logo [5].

11 | P a g e
Conclusion

In this chapter, the exploration of operating conditions, mathematical modeling, and


MATLAB simulations for membrane filtration in aerobic membrane bioreactors (AMRs) was
conducted. The mathematical model simplified the filtration process by emphasizing cake
formation while incorporating specific approximations. The different methods of membrane
operation were discussed, and simulations were performed with a filtration duration of 5
minutes and a backwash duration of 30 seconds. MATLAB demonstrated its effectiveness as
a powerful tool for modeling and simulation tasks, with functions such as fmincon and
ODE45 greatly enhancing the research efforts.

12 | P a g e
III. Experimental References

Introduction

In this chapter on Experimental References, we will explore the characteristics of the effluent
and influent, as well as the crucial experimental input data utilized in our study. These
parameters play a key role in the simulation of transmembrane pressure and subsequent
modeling of fouling. Understanding the characteristics of the effluent and influent is essential
for comprehending the composition and properties of the wastewater under investigation. By
examining key parameters such as suspended solids, conductivity, turbidity, COD, total
nitrogen, phosphorus, and pH levels, we can gain valuable insights into the nature of the
wastewater and its potential impact on the overall treatment process.

The experimental data provided for this study were obtained from Mr. Aymen Chaaben's
final-year project. These data serve as the primary source of information for our research,
forming the foundation for our analysis and modeling.

III.1 Characteristics of the Effluent and Influent in AMRs

• pH, conductivity, and turbidity

The pH was measured using a Mettler ToledoTM pH meter.

The electrical conductivity σ was measured using a VWR SB70C Benchtop conductivity
meter.

The Turbidity is the amount of suspended particles in water that cause cloudiness. Turbidity
was measured using a WTWTM TURB550 turbidimeter, which conforms to the NF EN ISO
7027 standard and provides measurements up to 10,000 NTU (nephelometric turbidity units).

Calibration of these instruments was performed periodically, following the manufacturer's


recommended calibration interval [6].

• Suspended solids

Suspended solids (SS) were measured using the filtration method on 0.7-micron and 0.45-
micron glass fiber filters for the influent and effluent, respectively. First, the empty filter (m0)
was weighed, and then a sampled volume of water (Vp) was poured through it. After the
filtered volume was vacuum filtered using a Buchner setup, the filter was recovered and

13 | P a g e
placed in an oven at 105°C for 1.5 hours. Finally, the filter was reweighed (mf) to calculate
the ratio of the difference between the two masses to the sampled volume [6].
𝑚1 − 𝑚𝑖
𝑆𝑆 =
𝑣𝑝𝑡

Where:

SS: Suspended Solids (mg. L−1 )

𝑚1 : Mass of Empty Filter (mg)

𝑚𝑖 : Mass of Filter after Filtration and Drying (mg)

𝑣𝑝𝑡 : Volume of the Sampled Water (L)

• Chemical oxygen demand (COD)

COD measurement was performed immediately after sample collection using WTWTM
252070 "COD kits." This test is compliant with the EN ISO 15705 standard. The "COD kits"
consist of tubes containing sulfuric acid, potassium dichromate, mercuric sulfate, and silver
sulfate. To analyze the sample, 3mL of the sample was poured into a tube, which was then
placed in a thermoreactor at 148°C for 2 hours. After cooling the tube, COD was measured
using a WTWTMPhotoLabTM S6 filter photometer [6].

• Total nitrogen and phosphorus

The determination of total nitrogen and ortho-phosphates (𝑃𝑏𝑂43− ) is done directly using a
photometric method with WTWTM cell test kits. The method involves adding a predetermined
volume of the sample and corresponding reagents, allowing them to rest for a specific time.
Finally, the concentration measurement is determined using a WTWTMPhotoLab photometer
[6].

Please refer to the appendix for experimental values of these characteristics.

III.2 Experimental Input Data

Experimental data on transmembrane pressure over time was obtained, collected during
multiple cycles of 5 minutes of filtration and 30 seconds of backwashing. The raw
experimental data corresponding to transmembrane pressure during backwashing will be
presented in the following data.

14 | P a g e
0.14

Experimental Transmembrane Pressure (bar)


0.12

0.1

0.08

0.06

0.04

0.02

0
-1120 880 2880 4880 6880 8880 10880 12880
Time (s)

Figure III-1 Time-dependent Variation in Processed Transmembrane Pressure

To facilitate the simulations and ensure an overall understanding, all the relevant data and
parameters have been compiled in the table below.

Table III-1 Input Data for The Simulation

Parameter Value Unit

Concentration of suspended matter (C) 2,2 𝑘𝑔. 𝐿−1

Volumetric flux (Jv) 1,9. 10−6 𝑚3 . 𝑚2 . 𝑠

Initial deposit mass (m0) 10−3 𝑘𝑔. 𝑚2

Dynamic viscosity (u) 10−3 𝑃𝑎. 𝑠

Membrane resistance 2,0021. 1012 𝑠 −1

15 | P a g e
Conclusion

In conclusion, this chapter on experimental references has yielded valuable insights into the
characteristics of the effluent and influent, as well as the crucial experimental input data for
our study. The reliable and primary source of information was derived from Mr. Aymen
Chaaben's final-year project, serving as the foundation for our analysis and modeling efforts.

Ultimately, this chapter underscores the significance of leveraging experimental references


and data analysis in the pursuit of advancing membrane filtration processes.

16 | P a g e
IV. Simulation of Transmembrane Pressure

Introduction

This chapter focuses on the simulation of transmembrane pressure during the filtration phase.
The objective is to align the pressure values obtained from the mathematical model with the
experimental data for a fixed filtration time by adjusting the shear force and specific cake
resistance values. This alignment aims to improve the accuracy of the model and enhance the
efficiency of the membrane in practical applications.

IV.1 Parameters to Optimise

IV.1.1 Specific Cake Resisitance

Specific Cake Resistance refers to the resistance encountered by the filtrate as it passes
through the cake layer formed on the membrane surface. It is a measure of how easily the
filtrate can flow through the cake layer.Specific Cake Resistance has a direct influence on
transmembrane pressure. As the cake resistance increases, the flow of the filtrate becomes
more restricted, leading to an increase in transmembrane pressure. Conversely, when the cake
resistance decreases, the filtrate can flow more easily, resulting in a decrease in
transmembrane pressure.Within the context of the case study, the specific cake resistance
interval considered ranges from 5. 1013 to 5. 1015 m.kg-1.

IV.1.2 Shear Force

Shear force refers to the force applied to the cake layer during the backwashing or cleaning
processes. It is the force that acts parallel to the membrane surface, causing the cake layer to
detach and be removed from the membrane.

The Shear Force has a significant influence on the transmembrane pressure. When the Shear
Force is increased, it helps to dislodge and remove the cake layer more effectively, resulting
in a decrease in transmembrane pressure. On the other hand, if the Shear Force is too low, the
cake layer may not be adequately removed, leading to an increase in transmembrane pressure.

The considered range for the Shear Force is between 0 and 100 𝑚2 . 𝑘𝑔−1 . This interval
enables the exploration of various levels of Shear Force and the analysis of their impact on
transmembrane pressure.

17 | P a g e
IV.1.3 Cake BackwashingEfficiency

Cake backwashing efficiency refers to the effectiveness of the backwashing process in


removing the accumulated cake layer from the membrane surface. It measures how efficiently
the backwash operation can dislodge and flush away the cake material, allowing the
membrane to regain its optimal filtration capacity.

The efficiency of cake backwashing has a direct impact on transmembrane pressure. Higher
backwashing efficiency results in better removal of the cake layer, leading to a decrease in
transmembrane pressure. Conversely, if the backwashing efficiency is low, the cake layer may
not be effectively removed, causing an increase in transmembrane pressure as the cake layer
continues to accumulate on the membrane surface.

In the case study, an interval of 0 to 0.2 𝑠 −1 will be considered for the cake backwashing
efficiency.

IV.2 Results and Discussion

To generate the next set of results, a MATLAB code was developed using the predefined
function ODE45. This function facilitated the solution of the differential equation that
describes the deposition mass as a function of time. The calculated deposition mass served as
an intermediate variable for determining the transmembrane pressure. By employing ODE45,
accurate modeling and simulation of the system's behavior were achieved, ultimately leading
to the desired results.

IV.2.1 Filtration Simulation

After conducting experiments, it was observed that the specific cake resistance parameter has
a more significant impact on the variation in transmembrane pressure than the shear force
parameter. Consequently, a fixed value of the shear force parameter was chosen, namely 50,
which lies in the middle of the interval [0, 100]. Initially the focus was on a filtration duration
of 5 minutes, specifically the first filtration cycle. The obtained results can be observed in the
provided figures.

18 | P a g e
Figure IV-1 Transmembrane Pressure Evolution as Function of Time for 5 minutes Filtration.

Since specific cake resistance values greater than 5.1015 m.kg-1 resulted in higher
transmembrane pressure values compared to experimental data, it was decided to use
intermediate values between 5.1013 and 5.1014 m.kg-1. By selecting these intermediate
values, it is expected to achieve a better balance between cake resistance and transmembrane
pressure, bringing the simulated results closer to the experimental data.

19 | P a g e
Figure IV-2 Adjusted Transmembrane Pressure Evolution as Function of Time for 5 minutes Filtration.

To obtain a more accurate estimation of the interval for the specific cake resistance parameter,
it is necessary to consider a longer period to better represent the process. However, it is
important to keep in mind that the model in this particular section does not take into account
the backwashing part. Therefore, we have chosen transmembrane pressure values slightly
higher than the experimental values to account for the effect of backwashing. The results are
shown in these figures.

20 | P a g e
Figure IV-3 Transmembrane Pressure Evolution as Function of Time for 30 minutes Filtration.

Based on the two previous curves, it can be observed that values of the specific cake
resistance parameter greater than 5. 1014 result in significantly high transmembrane pressures.
Therefore, it is necessary to narrow down the upper limit of the specific cake resistance
parameter interval to a value equal to 5. 1014 .It is therefore recommended that the specific
cake resistance parameter range be reduced to [5. 1013 , 5. 1014 ] to achieve better alignment
between the modeled transmembrane pressures and the experimental values.

IV.2.2 Filtration and Backwash Simulation

In order to achieve a comprehensive understanding of the entire process, the incorporation of


backwashing cycles was accomplished by modifying the previous code to include the
equation with the control variable u(t). Additionally, the consideration of the backwashing
coefficient efficiency was taken into account.

During the simulations, a series of tests were conducted where the values of three parameters
were randomly and simultaneously altered. Among these simulations, a combination of
parameter values was identified in the provided table, which yielded the best results. The
DTM_model values closely matched the experimental data for this specific combination.

21 | P a g e
Table IV-1 Input Parameters for Transmembrane Pressure Simulation.

Parameter Value Unit

Specific Cake Resistance 5. 1014 𝑚. 𝑘𝑔−1

Shear Force 50 𝑚2 . 𝑘𝑔−1

Backwashing Coefficient
0,0045 𝑠 −1
Efficiency

The corresponding results are shown in this figure:

Figure IV-4 Evolution of SimulatedTransmembrane Pressure over Time Throughout the Entire Process.

However, it is important to note that this simulation only compared the initial and final values
of the experimental transmembrane pressure. The intermediate values of the transmembrane
pressure throughout the operation of the membrane were not considered.

22 | P a g e
Conclusion

In conclusion, this chapter presented a simulation and modeling approach for predicting
transmembrane pressure during the filtration process. The developed MATLAB code,
utilizing the ODE45 function, successfully estimated the deposition mass and transmembrane
pressure. The experiments revealed that the specific cake resistance parameter had a
substantial impact on transmembrane pressure variation, while the shear force parameter
showed less influence. By selecting appropriate parameter values and considering
backwashing cycles, the simulation results closely matched the experimental data.

However, it is important to acknowledge the limitations of this approach. The current


simulation only compared the initial and final transmembrane pressure values, overlooking
the intermediate pressure fluctuations during the membrane operation. Additionally, the
model did not account for the effect of backwashing on transmembrane pressure accurately.
These limitations suggest the need for further refinements in the modeling approach to capture
the dynamic behavior and account for additional factors such as backwashing. Nonetheless,
the presented results provide valuable insights into the behavior of transmembrane pressure in
the filtration process and serve as a foundation for future investigations.

23 | P a g e
V. Modeling of Transmembrane Pressure

Introduction

This chapter focuses on achieving a more precise and efficient optimization of the cake
resistance, shear force, and backwashing coefficient efficiency parameters. The ultimate
objective is to obtain modeled transmembrane pressure results that closely align with
experimental data.

To achieve this, a two-step approach is employed. First, the squared error between the
simulated transmembrane pressure values and the corresponding experimental data is
analyzed. The Fmincon function in MATLAB, along with ODE45, is utilized to minimize this
squared error through an iterative process. Through this iterative optimization, the model is
refined and transmembrane pressure values that closely align with the experimental data are
generated.

To further enhance the optimization process, the total operational duration is divided into
smaller time intervals. Within each interval, the parameter values are adjusted specifically
tailored to that duration. This finer-grained optimization takes into consideration the dynamic
variations in transmembrane pressure that occur during different stages of the process. By
optimizing the parameter values for each interval, the temporal changes can be accurately
captured, leading to a more precise representation of transmembrane pressure behavior.

V.1 Error Analysis

The objective is to optimize the assessment of the correspondence between modeled and
experimental transmembrane pressure values. This will be achieved by calculating the
squared error and coefficient of determination to quantify the unexplained variation and
determine the extent to which the observed data is accurately represented by the model.

A thorough error analysis will be conducted in this initial phase to accurately measure the
level of agreement between the modeled and experimental transmembrane pressure values.
The effectiveness and reliability of our model for predicting and comprehending the filtration
process will be determined through this evaluation.

24 | P a g e
- Squared Error

The squared error is a quantitative measure used to assess the dissimilarity between
theoretical and experimental values. A smaller squared error indicates a closer match between
the two sets of data, implying a higher level of accuracy in the model's predictions. On the
other hand, a larger squared error signifies a greater deviation between the theoretical and
experimental values, indicating a lower level of accuracy.

To calculate the squared error between the modeled and experimental transmembrane
pressure values, a straightforward process is followed. First, the difference is computed
between each corresponding pair of values. Then, each difference is squared to eliminate
negative signs and emphasize the magnitude of the disparities. Finally, all the squared
differences are summed up to obtain the overall squared error as expressed in the following
Equation:
𝑛

∑(𝑌𝑖 − 𝑌̂𝑖 )2
𝑖=1

Where:

𝑌𝑖 : respresents the modeled transmembrane pressure

𝑌̂𝑖 : respresents the experimental transmembrane pressure

V.2 Global ParameterOptimization

V.2.1 Coding Approach

In this part, a MATLAB code was developed to optimize the specific cake resistance, shear
force, and backwashing coefficient efficiency by uniformly adjusting their values throughout
the filtration process. The objective is to minimize the squared error between the theoretical
and experimental data.

For this optimization task, the Fmincon function in MATLAB is utilized. The code iteratively
adjusts the parameter values in order to find the values that minimize the squared error, which
serves as the objective function in this case.

This approach refines the model and determines optimized parameter values that accurately
represent the filtration process. By minimizing the squared error, the precision and reliability
of the simulation results are enhanced.

25 | P a g e
V.2.2 Resultsand Discussion

By providing the Fmincon function with these intervals and initial values, we can effectively
optimize the parameters and achieve maximum optimization for the filtration process.

The optimal range of values that result in maximum optimization, along with the
corresponding intervals and initial values, are presented in the following table:

Table V-1Fmincon Function Input Arguments for Global Optimisation.

Parameter Lower bound Upper bound Initial value Unit

Specific cake
1014 2 × 1014 1014 𝑚. 𝑘𝑔−1
resistance

Shear force 0 7 0.1 𝑚2 . 𝑘𝑔−1

Backwashing
coefficient 0 0.1 0.01 𝑚−2 . 𝑠 −1
efficiency

In the figure below, the experimental and modeled transmembrane pressure values are
represented and superimposed on the same graph:

Figure V-1 Evolution of ModeledTransmembrane Pressure over Time Throughout the Entire Process.

26 | P a g e
The results of the modeling process have been collected and presented in the following table:

Table V-2 Global Parameters OptimisationResults.

Parameter Value Unit

Specific cake resistance 1014 𝑚. 𝑘𝑔−1

Shear force 0.1 𝑚2 . 𝑘𝑔−1

Backwashing coefficient
0.001 𝑠 −1
efficiency

Squared error 112.541 −

Coefficient of determination 0.8201 −

These results demonstrate that the optimization of the parameter values has resulted in a
significant reduction in the squared error, indicating a closer alignment between the
theoretical and experimental transmembrane pressures. However, it is worth noting that the
squared error is still relatively high, indicating that there is room for further improvement to
enhance the accuracy of the model's predictions.

V.3 Time-DividedParameterOptimization

V.3.1 Coding Approach

During the initial optimization attempts, a significant squared error was encountered,
suggesting that the complexities of the filtration process were not adequately captured by the
chosen parameter values.

To address this limitation, a more refined approach will be employed by optimizing the
parameter values for each individual time interval. By dividing the process into smaller
durations, the parameters can be tailored to each interval, taking into account the temporal
dynamics of the filtration process and adjusting the parameter values accordingly.

By optimizing the parameters for each interval and utilizing the optimized values as the
starting point for the next duration, the model's representation can be refined, improving the
agreement between the model and experimental data.

27 | P a g e
Through the implementation of this iterative optimization strategy, a gradual improvement in
the squared error is expected as the model parameters become more finely tuned to the
specific characteristics of each duration. This iterative approach enables the capture of the
nuances of the filtration process at different time intervals, resulting in a more accurate
representation of the transmembrane pressure.

V.3.2 Results and Discussion

By providing the Fmincon function with these intervals and initial values, the simulation is
able to optimize the parameters and achieve maximum optimization for the filtration
process.The optimal range of values that result in maximum optimization, along with the
corresponding intervals and initial values for the parameters, are summarized in the table
below:

Table V-3Fmincon Function Input Arguments for Time-divided Optimisation.

Parameter Lower bound Upper bound Initial value Unit

Specific cake
1014 15. 1014 8. 1014 𝑚. 𝑘𝑔−1
resistance

Shear force 10 40 25 𝑚2 . 𝑘𝑔−1

Backwashing
coefficient 0,0014 0,1 0,0014 𝑚−2 . 𝑠 −1
efficiency

28 | P a g e
The simulation results are presented in the following tables and figure:

Figure V-2 Time-divided Optimization of Modeled and experimental Transmembrane Pressure over Time.

The analysis of the plotted curve reveals that, in this instance, the theoretical values of the
transmembrane pressure closely match the experimental values for most of the duration. This
indicates a strong correspondence between the model predictions and the actual observations.

29 | P a g e
The simulation results are presented in the following table:

Table V-4 Time-divided Parameters Optimisation Results.

Backwashing
Specific Cake
Squared Shear Force Coefficient
Duration Resistance
Error Efficiency
𝒎𝟐 . 𝒌𝒈−𝟏 −𝟏
𝒎. 𝒌𝒈
𝒎−𝟐 . 𝒔−𝟏

1 0,1869 39,99 6,8254 0,0174

2 0,1153 40 6,6900 0,0153

3 0,1762 39,99 6,6152 0,0126

4 0,1535 40 6,4481 0,0111

5 0,1845 39,99 6,2668 0,0094

6 0,5957 39,99 6,2463 0,066

7 0,1693 10 6,1580 0,008

8 0,1756 10,399 6,0859 0,007

9 0,1960 10,487 5,8997 0,0069

10 0,1775 10 5,8818 0,007

11 0,1525 10 5,9915 0,0068

12 0,1810 10 5,9276 0,0069

13 0,1413 10 5,8289 0,0069

For the time-divided approach, the calculated squared error values range from 0.1153 to
0.5957, with a sum of 2.2403. These values indicate that the theoretical predictions closely
align with the experimental observations for most of the durations, resulting in an average
squared error value of approximately 0.1614.

30 | P a g e
Analyzing the parameter values, it is observed that the shear force stabilizes at a value of
39.99 m².kg-1during the first half of the total duration of the process. However, in the second
half of the process, the shear force stabilizes at a value of 10 m².kg-1.This alignment between
the modeled shear force and the observed data captures the change observed during the
filtration process.

The specific cake resistance, ranging from 6.8254 to 5.8289.1014 m.kg-1, does not vary
significantly due to the relatively consistent transmembrane pressure values observed in the
experimental data. Considering that the specific cake resistance has the greatest influence on
transmembrane pressure values, the close proximity of these values is logical.

Regarding the backwashing coefficient efficiency, ranging from 0.0174 to 0.0069 s-1, it
indicates a decrease as the cake formation progresses, which is expected due to the influence
of cake buildup. These parameter values are obtained through optimization for each duration,
reflecting the unique characteristics of the process at different time intervals.

Conclusion

In conclusion, the utilization of the time-divided optimization approach has proven to yield
superior results for the shear force, specific cake resistance, and backwashing coefficient
efficiency parameters. This approach acknowledges the distinctive characteristics of the
filtration process during various time intervals, resulting in enhanced optimization outcomes.

To further improve the optimization process, additional measures can be taken. One such
measure involves reducing the duration of the intervals, allowing for a finer-grained analysis
of the filtration process. By dividing the process into smaller time intervals, the optimization
algorithm can more accurately capture the temporal dynamics and variations in the parameter
values, leading to improved optimization results.Additionally, exploring different values for
the upper and lower intervals of the parameters as arguments in the fmincon function can
contribute to further optimization. This would allow for a broader exploration of the
parameter space and potentially uncover more optimal values that lead to even better
alignment between the theoretical predictions and experimental data.

31 | P a g e
General Conclusion and Perspectives

In conclusion, the objectives of this project were successfully achieved in terms of exploring
Aerobic Membrane Bioreactors (AMRs), understanding fouling mechanisms, and optimizing
transmembrane pressure. Valuable insights were gained through the literature review, which
provided an understanding of AMR technology, fouling mechanisms, and the factors
influencing membrane fouling. The process of mathematical modeling of membrane filtration,
coupled with the utilization of MATLAB simulation and computational tools, facilitated a
comprehensive understanding of filtration dynamics, energy consumption, and
transmembrane pressure. Through the simulation of transmembrane pressure, a deeper
understanding of the filtration process and its key parameters was obtained. Promising results
were obtained through the optimization of transmembrane pressure using both global
parameter optimization and time-divided parameter optimization, demonstrating the potential
for improving AMR performance.

However, there is scope for further improvement and enhancement. Future research should
focus on the development of advanced fouling control strategies, the integration of sensor
technologies for real-time monitoring and management, and the exploration of sustainable
energy recovery options. Additionally, attention should be given to addressing scale-up
challenges and investigating new applications beyond wastewater treatment, contributing to
the broader adoption and commercial viability of AMR technology. By continually advancing
our knowledge and refining operational parameters, the efficiency, sustainability, and
versatility of AMRs in water treatment and related fields can be further enhanced.

32 | P a g e
Appendix
Appendix 1: Summary table of BRMIe characteristics and conventional treatment [6]

33 | P a g e
Bibliography

[1] ScienceDirect. [en ligne] [consulté: 03 Juin 2023.]


https://www.sciencedirect.com/science/article/pii/S1385894712006420

[2] ResearchGate. [en ligne] [consulté: 03 Juin 2023.] https://www.researchgate.net/figure/10-


Mechanisms-of-fouling-in-membrane-filtration-a-cake-layer-b-pore-
constriction_fig10_267682502

[3] Gaetano Di Bella, Daniele Di Trapani, 2019. A Brief Review on the Resistance-in-Series
Model in Membrane Bioreactors (MBRs). NIH, 2019 Feb; 9(2): 24

[4] ScienceDirect. [en ligne] [consulté: 03 Juin 2023.]


https://www.sciencedirect.com/science/article/abs/pii/S0376738820314113

[5] MathWorks. [en ligne] [consulté: 03 Juin 2023.]


https://www.mathworks.com/products/matlab.html

[6] Aymen Chaaban. (2021): Traitement des eaux usées municipales par Bioréacteur à
Membrane (BRM): Optimisation du nettoyage physique par contrôle optimal. Projet de Fin
d’Etudes, Institut national des Sciences Appliquées et de Technologie

34 | P a g e

You might also like