BORANG PENGESAHAN STATUS TESIS υ

UNIVERSITI TEKNOLOGI MALAYSIA PSZ 19:16 (Pind. 1/97) BORANG PENGESAHAN STATUS TESIS υ JUDUL : JAR TEST: ONE-FACTOR-AT A-TIME VERSUS RESPONSE SURFACE DESIGN SESI PENGAJIAN: 2005/2006 Saya ALBRAHAM ENGGONG ANAK JIMBAT (HURUF BESAR) mengaku membenarkan tesis (PSM/Sarjana/Doktor Falsafah)* ini disimpan di Perpustakaan Universiti Teknologi Malaysia dengan syarat-syarat kegunaan seperti berikut: 1. Tesis adalah hakmilik Universiti Teknologi Malaysia. 2. Perpustakaan Universiti Teknologi Malaysia dibenarkan membuat salinan untuk tujuan pengajian sahaja. 3. Perpustakaan dibenarkan membuat salinan tesis ini sebagai bahan pertukaran antara institusi pengajian tinggi. 4. **Sila tandakan ( ) SULIT TERHAD (Mengandungi maklumat yang berdarjah keselamatan atau kepentingan Malaysia seperti yang termaktub di dalam AKTA RAHSIA RASMI 1972) (Mengandungi maklumat TERHAD yang telah ditentukan oleh organisasi/badan di mana penyelidikan dijalankan) TIDAK TERHAD Disahkan oleh (TANDATANGAN PENULIS) Alamat Tetap: 140F,Tingkat 3,J.K.R Flat,Batu 2 ½,Jalan Sultan Iskandar,97000 Bintulu,Sarawak. Tarikh: 17 APRIL 2006 (TANDATANGAN PENYELIA) DR. AZMI ARIS Nama Penyelia Tarikh: 17 APRIL 2006 CATATAN * Potong yang tidak berkenaan ** Jika tesis ini SULIT atau TERHAD, sila lampirkan surat daripada pihak berkuasa/organisasi berkenaan dengan menyatakan sekali sebab dan tempoh tesis ini perlu dikelaskan sebagai SULIT atau TERHAD. υ Tesis dimaksudkan sebagai tesis bagi Ijazah Doktor Falsafah dan Sarjana secara Penyelidikan, atau disertasi bagi pengajian secara kerja kursus dan penyelidikan, atau Laporan Projek Sarjana Muda (PSM).

Saya akui bahawa saya telah membaca karya ini dan pada pandangan saya karya ini adalah memadai dari segi skop dan kualiti untuk tujuan penganugerahan ijazah Sarjana Muda Kejuruteraan Awam (Alam Sekitar). Tandatangan: Nama Penyelia: Tarikh:

i JAR TEST: ONE-FACTOR-AT A-TIME VERSUS RESPONSE SURFACE DESIGN ALBRAHAM ENGGONG ANAK JIMBAT This thesis is submitted as a partial fulfilment of the requirement for the award of the Bachelor Degree in Civil Engineering (Environmental) Faculty of Civil Engineering Universiti Teknologi Malaysia April 2006

ii UJIAN BALANG: SATU-FAKTOR-PADA SATU-MASA (OFAT) MELAWAN RESPONSE SURFACE DESIGN ALBRAHAM ENGGONG ANAK JIMBAT Laporan projek ini dikemuka Sebagai memenuhi sebahagian daripada syarat Penganugerahan ijazah Sarjana Muda Kejuruteraan Awam (Alam Sekitar) Fakulti Kejuruteraan Awam Universiti Teknologi Malaysia April 2006

iii Saya akui karya ini adalah hasil kerja saya sendiri kecuali nukilan dan ringkasan yang tiap-tiap satunya telah saya jelaskan sumbernya. Tandatangan: Nama Penuh: Tarikh:....

iv Specially dedicated to my loving Dad & Mom Frederick Jimbat Galan and Audrey Sani Liam Brothers and sister Wilfred Jubin, Petrus Dunggat and Victoria Lilit All the KENYALANG members and GIFT members

v ACKNOWLEDGEMENT First of all, I thank God for answering my prayers and giving me strength to move on even during the hard times. Next I would like to take this opportunity to thank my supervisor, Dr. Azmi Aris who had given me the full support, guidance and encouragement to complete this project. Lastly, my heartiest thanks to my wonderful family and friends who played the major role in giving me all the guidance and support I need.you all the light of my light. Thank you.

vi ABSTRACT Presently, Jar Test experiment was conducted using one-factor-at a-time (OFAT) approach, which manipulates one factor at a time looking for the best value of each factor. Response surface design is an alternative way to determine the impact of two or more factors on a response. A study was conducted in laboratory to determine the final turbidity as a response by using Jar Test procedure. Two independent variables tested were ph and coagulant dosage. Experiment using OFAT approach was conducted in 18 runs, while experiment using response surface approach was conducted in 12 runs. Using response surface analysis, it was found that the variables were significant in affecting the response. The lowest turbidity obtained in OFAT experiments was 0.2 NTU at ph 6.5 and coagulant dosage of 100 mg/l. Using response surface design, the lowest turbidity obtained was 0.02 at ph 7.4 and dosage of 100 mg/l. It was found that a better water quality could be achieved using response surface approach rather than the traditional OFAT. A statistical relationship was also developed in the response surface analysis. Keywords: Jar Test, OFAT, Response Surface, Turbidity.

vii ABSTRAK Pada masa kini, Ujian Balang (Jar Test) hanya melibatkan penggunaan kaedah satu-faktor-pada-satu-masa(ofat), dimana hanya satu faktor ubahsuai pada sesuatu masa tertentu untuk mencari nilai terbaik untuk setiap faktor yang dikaji. Kaedah response surface merupakan satu kaedah alternatif untuk mengenalpasti kesan dua atau lebih faktor terhadap respon. Ujikaji telah dijalankan di makmal untuk mengenal pasti kekeruhan sebagai respon dengan menggunakan prosedur Ujian Balang. Dua pembolehubah iaitu dos pengental (coagulant) dan nilai ph diuji. Ujikaji untuk OFAT dijalankan sebanyak 18 run manakala ujikaji untuk response surface dijalankan sebanyak 13 run. Menggunakan kaedah analisis response surface, pembolehubah memainkan peranan yang penting dalam mempengaruhi respon. Nilai kekeruhan paling rendah didapati dari ujikaji OFAT adalah 0.20 NTU pada ph 6.5 dan 100 mg/l dos pengental. Sementara itu, response surface pula menghasilkan kekeruhan yang terendah sebanyak 0.02 NTU pada ph 7.4 dan dos pengental pada 106 mg/l. Kualiti air yang terbaik dapat dicapai dengan menggunakan kaedah response surface jika dibandingkan dengan kaedah tradisional OFAT. Analisis response surface juga dapat memberikan hubungkait statistic diantara pembolehubah dan respon. Kata Kunci: Ujian Balang, OFAT, Response Surface, Kekeruhan

viii CONTENTS SUBJECTS Title Tajuk Pengakuan Dedication Acknowledgement Abstract Abstrak Main Content Content List of Tables List of Figures PAGE i ii iii iv v vi vii viii ix xi xii

ix CONTENT CHAPTER I: INTRODUCTION 1 1.1 Preamble 1 1.2 Aims of Study 3 1.3 Objectives of Study 4 1.4 Scope of Study 4 CHAPTER II: LITERATURE REVIEWS 5 2.1 Introduction 5 2.2 Coagulation 6 2.2.1 Coagulants 8 2.3 Flocculation 9 2.4 One-Factor-At a-time (OFAT) 11 2.5 Factorial Design 12 2.5.1 Two-level (2 k ) Full Factorial 13 2.6 Response Surface Method (RSM) 15 2.6.1 Central Composite Design 17 CHAPTER III: METHODOLOGY 20 3.1 Material and Equipments 20 3.2 Experimental Procedure 22

x 3.2.1 Jar Test Procedure 22 3.2.2 One-Factor-At a-time 22 3.2.3 Response Surface Design 26 CHAPTER IV: RESULTS AND ANALYSIS 27 4.1 One-factor-at a-time 27 4.2 Response Surface Design 30 CHAPTER V: CONCLUSION AND RECOMMENDATION 36 5.1Conclusion 36 5.2Advantages of Response Surface Method Over OFAT Experiment 38 REFERENCES 40

xi LIST OF TABLES TABLE NO. TITLE PAGE 3.1 Three Sets of OFAT Experiments With a Total of 18 Runs 25 3.2 Experimental Runs Conducted Using Response Surface Method 26 4.1 The Results of OFAT Experiments 29 4.2 Turbidity Results for RSM Experiment 31 4.3 Result of the Analysis for Response Surface Using MINITAB TM Software. 31 4.4 Analysis of Variance for Turbidity 32

xii LIST OF FIGURES FIGURE NO. TITLE PAGE 2.1 Derivation of Zeta Potential In Diffuse double-layer Theory (Vesilind et al, 1994) 7 2.2 Two-level Full Factorial Designs 14 2.3 Two Factors 14 2.4 Three Factors 15 2.5 Central Composite Design 18 2.6 2 2 Factorial Designs 19 2.7 Central Composite Designs α = 2 19 3.1 Jar Test Apparatus (Philip and Bird) 21 3.2 The Soil Sample for OFAT and Response Surface Experiment 21 3.3 Raw Water Sample Before and After Experiment 23 3.4 Flowchart for OFAT Experiments 24 4.1 The Final NTU at ph 4.0 and ph 7.0 28 4.1 Contour Plot of the Response Surface for the Central Composite Design Experiment. 34 4.2 3D Response surface for Turbidity 35

CHAPTER 1 INTRODUCTION 1.1 Preamble Surface water generally contains a wide variety of colloidal impurities that may cause the water to appear turbid coloured. Turbidity is most often caused by colloidal clay particles produced by soil erosion. Colour may result from colloidal forms of iron and manganese or more commonly from organic compounds contributed by decaying vegetation. Colloidal particles that cause colour and turbidity are difficult to separate from the polluted water because the particles will not settle by gravity and are so small that they pass through pores of most common filtration methods. To be removed, the individual s colloids must aggregate and grow in size. Aggregation is complicated not only by the small size of the particles but more importantly by the fact that physical and electrical forces keep the particles separated from each

2 other and prevent the collision that would be necessary for aggregation to occur. Coagulation and flocculation are important processes in water and wastewater treatment plant. The purpose of coagulation and flocculation is to remove suspended matter, turbidity, colour, microorganisms, and odour producing substances. Coagulation involves the addition of chemical to destabilise (allow them to agglomerate) the suspended particles, colloidal materials, and macromolecules. Some common coagulants used are aluminum sulphate (alum) and ferric sulphate. Flocculation is usually defined as the aggregation of destabilized particles into larger flocs under slow mixing conditions.the flocs formed are subsequently removed by sedimentation and filtration. A useful laboratory experiment for the evaluation of coagulation and flocculation of untreated water is called Jar Test. Jar Test is an experimental method where optimal conditions are determined empirically rather than theoretically. Jar test are meant to mimic the conditions and processes that take place in the coagulationflocculation clarification portions of water and wastewater treatment plants. The values that are obtained through the experiment are correlated and adjusted in order to account for the actual treatment system requirement. Presently, Jar Test experiment is conducted using the traditional one-factor-at a-time (OFAT) approach. The first factor is

3 fixed at a "good" value while the other factors are held constant. Then, the next factor is examined using the best value of the previous factor with other factors remained fixed. The process continue until all factors are examined. Since each series of experiment considers only one factor, many runs are usually needed to get sufficient information. Because of this, a detailed study of all factors is generally prohibitive. Beside, the results of the approach are rather qualitative than quantitative and do not provide any information on the interactive effect that could cause by the factors. Response surface design is an alternative approach to OFAT. In response surface, all factors are considered simultaneously covering wide range of values, without requiring that they all be directly tested. Furthermore, it also reveals any interaction effect contributed by the factors to the response. 1.2 Aim of Study The main aim of the study is to determine the appropriate approach in optimizing the coagulation and flocculation process.

4 1.3 Objectives of Study The objectives of the study are, To investigate the differences of outcome between OFAT and response surface approaches. To identify the appropriate approach in conducting Jar Test. 1.4 Scope of the Study The study mainly consists of experiment conducted using Jar Test approach. Synthetic water was used in the experiments and only two factors were considered, namely ph and coagulant dosage. Turbidity was used as the response. MINITAB TM software was used as an aid in Response Surface approach.

5 CHAPTER II LITERATURE REVIEWS This chapter describes briefly on coagulation, flocculation, OFAT and response surface design. The review of coagulation is based on Vesilind et al (1994), Benefield et al (1982) and Reynolds and Richard (1996), while the review on flocculation is based on Moss and Dymond (2004). For the experimental approach, the review is based on Guan and Melchers(2000) and Czitrom (1999). 2.1 INTRODUCTION Removal of turbidity by coagulation depends on the nature and concentration of the colloidal contaminants, type and dosage of chemical coagulant, use of coagulant aids, and chemical characteristic on the water, such as ph, temperature, and ionic character. There are two types of colloids, hydrophilic and hydrophobic. Hydrophilic colloids are readily dispersed in water and their stability (lack of tendency to agglomerate) depends on a marked affinity for water rather than on the slight charge (usually negative) that they possess.

6 Hydrophobic colloids possess no affinity for water and owe their stability to the electric charge they possess. A charge on the colloids is gained by absorbing positive ions from water solution. In destabilising colloids, two basic mechanisms have been described in forming sufficiently large aggregates to settle from the suspension. The first mechanism, referred to as coagulation, reduces the net electrical repulsive forces at particle surfaces by electrolytes in solution and the second mechanism, known as flocculation is aggregation by chemical bridging between particles. 2.2 COAGULATION Coagulation, used in conjunction with flocculation and sedimentation is one of the important processes of potable water treatment. The objective of the process is to destabilise particles and enable them to become attached to other particles so that they may be removed in subsequent process. The suspended solids which comprise the colloidal particles, such as algae, bacteria, virus, organic and inorganic substances and clay particles cause the water unsuitable for domestic use. The mechanisms of chemical coagulation involved the reduction of zeta-potential compression, neutralisation by the opposite

7 charge, inter-particle bridging and precipitation (Vesilind et al, 1994). The zeta potential is as measure of the stability of particles and indicates the particles that would be required to penetrate the layer of ions surrounding the particles for destabilisation. The concept of zetapotential is derived from the diffuse double-layer theory applied to hydrophobic colloids as shown at Figure 2.1. Figure 2.1: Derivation of zeta potential in diffuse double-layer theory. (Vesilind et al, 1994) A fixed covering of positive ions is attracted to the negatively charged particle by electrostatic attraction. This stationery zone positively ions are referred to as the astern layer, which is surrounded by moveable, diffuse layer of counter ions. The concentration of the

8 positive ions in the diffuse zone decreases as it extents into the surrounding bulk of electro-neutral solution. Zeta potential is the magnitude of the charge at the surface of shear. The boundary surface between the fixed ion layer and the solution serves as a shear plane when the particle undergoes movement relative to the solution. The zeta-potential magnitude can be estimated from electrophoresis measurement of particle mobility in an electric field. The purpose of coagulation is to reduce the zeta-potential by adding specific ions and then induce motion for the destabilised particles to agglomerates. 2.2.1 Coagulants There are many coagulants available for coagulation process. The most widely used coagulants in water or waste water treatment are aluminium sulphate and iron salts. Aluminium sulphate (filter alum) is employed more frequently than iron salts because it is usually cheaper. However, sufficient alkalinity must be present in the water to react with the aluminium sulphate to produce the hydroxide floc. Usually, for the ph ranges involved, the alkalinity is in the form of the bicarbonate ion or carbonate ion. The optimum ph range for alum is from about 4.5 to 8.0. Iron salts have an advantage over filter alum because they are effective over a wider ph range. Ferrous sulphate requires alkalinity in the form of the hydroxide ion in order to produce a rapid reaction.

9 Consequently, slaked or hydrated lime, Ca (OH) 2, is usually added to raise the ph to a level where the ferrous ions are precipitated as ferric hydroxide. This reaction is an oxidation reduction reaction requiring some dissolved oxygen in the water. In the coagulation reaction, the oxygen is reduced and the ferrous ion is oxidized to the ferric state, where it precipitates as ferric hydroxide. For this reaction to occur, the ph must be raised to about 9.5, and some times stabilisation is required for the excess lime employed. Ferrous sulphate and lime coagulation is usually more expensive than alum. In the lime soda softening process, the lime serves as a coagulant since it produces a heavy flocs or precipitate consisting of calcium carbonate and magnesium hydroxide. This precipitate has coagulating and flocculating properties. Sometimes coagulants aids, such as recycled sludge or polyelectrolyte are required to produce a rapid- settling flocs (Reynold and Richard, 1996). 2.3 FLOCCULATION Most flocculants and filter aids in use today are synthetic polymers based on repeating units of acrylamide and its derivatives, which may contain either cationic or anionic charges and are available in a wide range of molecular weights and ionic charge. The term flocculation is often confused with coagulation, although the two refer

10 to quite different processes. Coagulation is basically electrostatic in that it is brought about by a reduction of the repulsive potential of the electrical double layer. According to Moss and Dymond (2004), the term flocculation is derived from the Latin, "flocculus", literally a small tuft of wool, or a loosely fibrous structure. Flocculation is the action of polymers to form bridges between the flocs and bind the particles into large agglomerates or clumps. Bridging occurs when segments of the polymer chain adsorb on different particles and help particles aggregate. An anionic flocculants will react against a positively charged suspension, adsorbing on the particles and causing destabilisation either by bridging or charge neutralisation (Moss and Dymond, 2004). In this process it is essential that the flocculating agent be added by slow and gentle mixing to allow for contact between the small flocs and to agglomerate them into larger particles. The newly formed agglomerated particles are quite fragile and can be broken apart by shear forces during mixing. Care must also be taken to not overdose the polymer as doing so will cause settling or clarification problems. Anionic polymers themselves are lighter than water. As a result, increasing the dosage will increase the tendency of the floc to float and not settle. Once suspended particles are flocculated into larger particles, they can usually be removed from the liquid by sedimentation, provided that a sufficient density difference exists between the suspended matter and the liquid. Such particles can also be removed or separated by media filtration, straining or floatation. When a filtering process is used, the addition of a flocculants may not be

11 required since the particles formed by the coagulation reaction may be of sufficient size to allow removal. The flocculation reaction not only increases the size of the floc particles to settle them faster, but also affects the physical nature of the floc, making these particles less gelatinous and thereby easier to dewater. 2.4 ONE-FACTOR-AT A-TIME (OFAT) One-factor-at a-time is a simple experiment design which involves testing the factors one-at-a-time. The first factor is fixed at a good value while the others factors are held constant. Then, the next factor is examined using the best value of the previous factor with other factors remained fixed. The process continues until all factors are examined. Since each series of experiment considers only one factor, many runs are usually needed to get sufficient information. Because of this, a detailed study of all factors is generally prohibitive. Besides, the results of the approach are rather qualitative than quantitative and do not provide any information on the interactive effect that caused by the factors. OFAT is the most common approaches used in experimental work. It is relatively simple and straight forward and does not require advanced statistical knowledge. Nevertheless, there are many shortcomings acquainted with OFAT. These include:

12 Requiring more experimental runs for the same precision in factor effect estimation Incapable of estimating interaction effect Possible leading to incorrect conclusion 2.5 FACTORIAL DESIGN Factorial design allows for the simultaneous study of the effects that several factors may have on a process. In experiment factorial design, the level of factors is simultaneously changed and determined prior to the experiments. It is capable of identifying the significant factor quantitatively and allow for detection of interaction effect. While different levels of factor can be used in factorial design, the most common approach is to use two level factorial designs.

13 2.5.1 Two- level (2 k ) Full Factorial A full factorial design contains all combinations of the levels of the factors. Since the number of experiments is based on 2 k, the number of experiment will increase double as the number of factor increase. Combinations of factor levels represent the conditions at which responses will be measured. Each experimental condition is called a "run" and the response refers to the measurement of observation. The entire set of runs is called the "design." Full factorials design allows the estimation of interactions of all orders up to the number of factors as shown in Figure 2.2. Most empirical modelling involves first or second order approximations to the true functional relationship between the factors and the responses, - 1-1 -1. The following diagrams (Figure 2.3 and Figure 2.4) show two and three factor designs. The points represent a unique combination of factor levels. For example, in the two-factor design, the point on the lower left corner represents the experimental run when Factor A is set at its low level and Factor B is also set at its low level.

14 Figure 2.2 Two-level full factorial designs Figure 2.3 Two Factors

15 Figure 2.4 Three Factors 2.6 R.S.M RESPONSE SURFACE METHOD The Response Surface method(rsm) defined as a broad category of experimental design and analysis methods based on fitting models which are linear and quadratic equations in the experimental factors (this includes cross-terms for interactions). Such purely empirical models are useful for describing systems behaviour, process improvement, and often increasing understanding so that more detailed conceptual (mechanistic) models can be developed.

16 The uses of RSM are stated as below: To determine the factor levels that will simultaneously satisfy a set of desired specifications. To determine the optimum combination of factors those yield a desired response and describe the response near the optimum. To determine the specific response is affected by changes in the level of the factors over the specified levels of interest. To achieve a quantitative understanding of the system behaviour over the region tested. To product properties throughout the region, even at factor combinations are not actually run. To find conditions for process stability, insensitive spot.

17 2.6.1 Central Composite Design The most popular RSM is the central composite design as illustrated in Figure 2.5. It combines a two-level fractional factorial and other kinds of points defined as follow: Centre points, for which all the factors values are at the mid range value. Axial (or star) points, for which all but one factor are asset at mid range and one factors is asset at outer (axial) values. Central composite response surface designs are two level full or fractional designs that have been augmented with a small number of carefully chosen treatments to persist estimation of the second-order response surface model.

18 FRACTIONAL FACTORIAL POINTS CENTRE POINTS AXIAL POINTS Figure 2.5 Central Composite Designs While Figure 2.6 illustrate a 2 2 factorial design with low and high factor level coded as (-1) and (+1), Figure 2.7 illustrate a central composite design developed from the factorial design. The additional points of central composite design on compared to factorial design in the center point (coded as 0) and star or axial points. The four star points are located at the centers of each of the four edges of the experimental region but position of the center point depend on the number of factors involved in the experiments. A more detail discussion is available in Design and Analysis of Experiments, Montgomery et al (1996).

19 Figure 2.6 2 2 Factorial Designs Figure 2.7 Central Composite Designs

20 CHAPTER III METHODOLOGY 3.1 MATERIALS AND EQUIPMENTS The Jar Test is a common laboratory procedure used to determine the optimum operating conditions for coagulation process. This method allows adjustments in ph, variations in coagulants or polymer dose, alternating mixing speeds, or testing of different coagulant or polymer types, on small scale in order to predict the functioning of a large scale treatment operation. The experiments were conducted using the Jar Test apparatus (Philips and Bird) as shown in Figure 3.1. Aluminate sulphate (alum) was used as the coagulant and the ph of the water was adjusted using 0.02 NaCl. Synthetic raw water was prepared using natural soil (Figure 3.2) which available near the laboratory. The ph of the solution was measured using ph meter and the turbidity of the water was measured using Hach turbid meter.

21 Figure 3.1: Jar Test apparatus (Philips and Bird) Figure 3.2: The soil sample for OFAT and response surface experiment

22 3.2 EXPERIMENTAL PROCEDURES 3.2.1 Jar Test Procedure The experiments (OFAT or RSM) were conducted according to the typical Jar Test procedure. It consists of rapid stirring (60 80 rpm) of the samples after chemical addition for one minute, followed by slow mixing (10 to 30 rpm) for about 15 minutes. After the stirring period is over, stop the stirrer and allow the flocs to settle for about 30 minutes. Pipette 10 ml of cleared sample from each beaker and measure its turbidity using the turbid meter. 3.2.2 One-Factor-At a-time One-Factor-At a-time experiment consists of 18 experimental runs. The flow chart of the OFAT experiment is shown in Figure 3.4, while the experimental conditions are given in Table 3.1. A volume of 500 ml of distilled water was added in each of six beakers and mixed with 20 mg of soil sample. Alum solutions at different dose (20 mg/l, 40 mg/l, 60 mg/l, etc.) were added into each beaker. For the first set of six runs, the ph value was fixed to 7.0. The dose of alum was varied in each beaker for 20 mg/l to 120 mg/l. The water was rapid

23 mixed at approximated 80 rpm for one minute followed by slow mixing at 30 rpm for approximates 15 minutes. At the end of mixing period, the floc was let settled for about 30 minutes. The turbidity in each sample was immediately analysed. The best dosage was determined based on the lowest final turbidity. Figure 3.3 shows the raw water sample before and after experiment conducted. The second set of experiments was conducted using the best dosage of the first at different ph following the procedures as detailed before. The best ph was obtained based on the lowest turbidity value. The third set was runs at ph4.0 as a comparison to the first set of experiment. Figure 3.3: Raw water sample before and after experiment

24 SET ONE The best dosage was determined based on the lowest final turbidity SET TWO Using the best coagulant dosage of set one. SET THREE As a comparison to set one with ph4 Figure 3.4: Flow chart for OFAT experiment

25 Table 3.1: Three sets of OFAT experiment with a total of 18 runs. RUN ph Coagulant (mg/l) SET 1 OFAT 1 OFAT 2 OFAT 3 OFAT 4 OFAT 5 OFAT 6 7.0 7.0 7.0 7.0 7.0 7.0 20 40 60 80 100 120 SET 2 OFAT 7 OFAT 8 OFAT 9 OFAT 10 OFAT 11 OFAT 12 5.0 5.5 6.0 6.5 7.0 7.5 100 100 100 100 100 100 SET 3 OFAT 13 OFAT 14 OFAT 15 OFAT 16 OFAT 17 OFAT 18 4.0 4.0 4.0 4.0 4.0 4.0 20 40 60 80 100 120

26 3.2.3 Response Surface Design The experimental work for response surface was statically designed using Central Composite Rotatable Design with the aid of MINITAB TM software. The experiments were divided into 13 runs as shown in Table 3.2. The ranges of ph and coagulant dosage follow those conducted in OFAT approach. Run of RSM 9 to RSM 13 are repetition for the calculation of the residual error in the analysis to avoid bias, the experiment were conducted in a random manner. The experiments procedures are similar to those in previous OFAT experiment. Table 3.2: Experimental runs conducted using Response Surface Method RUN ph Coagulant (mg/l) RSM 1 4.6 34 RSM 2 4.6 106 RSM 3 7.4 34 RSM 4 6.0 20 RSM 5 7.4 106 RSM 6 6.0 120 RSM 7 4.0 70 RSM 8 8.0 70 RSM 9 6.0 70 RSM 10 6.0 70 RSM 11 6.0 70 RSM 12 6.0 70 RSM 13 6.0 70

27 CHAPTER IV RESULTS AND DISCUSSION 4.1 ONE-FACTOR-AT A-TIME The results of the experiments are shown in Table 4.1. The final turbidity ranges from 0.20 NTU to 1.30 NTU and the ranges of coagulant dosage between 20 mg/l to 120 mg/l. The lowest final achieved turbidity in set one is 0.50 NTU with coagulant dosage of 100 mg/l of ph7.0. Using alum dosage of 100 mg/l, the lowest turbidity achieved in set two was 0.20 NTU when the ph equal to 6.5. However, the lower final turbidity of 0.25 NTU was obtained when the coagulant dosage of 100 mg/l when the ph was set to 4.0. Figure 4.1 illustrate the final NTU at ph 4 and ph 7 as a function of coagulant dosage. The final turbidity at ph 4.0 (0.25 NTU) is much lower than the final turbidity achieved at ph 7.0 (0.50 NTU) although both coagulant dosage is equal. It shows that if only two sets of experiment were conducted as in normal procedure, the final turbidity obtained was not the lowest one. There seem to be found that the

28 interaction affect in the process which is not disclosed by the OFAT approach. Final Turbidity (NTU) versus Coagulant Dosage (mg/l) 1.4 1.2 1 Final Turbidity (NTU) 0.8 0.6 0.4 0.2 0 20 40 60 80 100 120 Coagulant Dosage (mg/l) ph 7.0 ph 4.0 Figure 4.1: The final NTU at ph 4.0 and ph7.0.

29 Table 4.1: The results of the OFAT experiments SET RUN ph Coagulant (mg/l) Final Turbidity (NTU) 1 OFAT1 OFAT2 OFAT3 OFAT4 OFAT5 OFAT6 7.0 7.0 7.0 7.0 7.0 7.0 20 40 60 80 100 120 1.20 1.03 0.86 0.70 0.50 0.54 OFAT7 5.0 100 0.85 OFAT8 5.5 100 0.53 OFAT9 6.0 100 0.32 2 OFAT10 6.5 100 0.20 OFAT11 7.0 100 0.22 OFAT12 7.5 100 0.24 OFAT13 4.0 20 1.30 OFAT14 4.0 40 1.26 OFAT15 4.0 60 0.90 3 OFAT16 4.0 80 0.56 OFAT17 4.0 100 0.25 OFAT18 4.0 120 0.26

30 4.2 RESPONSE SURFACE DESIGN The result of response surface method is given in Table 4.2. The ranges of final turbidity achieved in response surface experiments is 0.02 NTU to 3.21 NTU with coagulant dosage ranges from 20 mg/l to 120 mg/l and ph from 4.0 to 8.0. The lowest final turbidity for response surface experiments in Table 3 was 0.02 NTU in RSM 5, when the coagulant dosage was 106 mg/l and the ph is 7.4. The final turbidity achieved in response surface experiment (0.02 NTU) is much lower than OFAT experiment (0.20 NTU). The percentage differences between both lowest final turbidity OFAT and response surface experiment are 90 %. Table 4.3 summarizes the results of the analysis using MINITAB TM software. It shows that the dosage of the coagulant and ph were significant factors in the coagulation process at confidence level of 95%. The two ways interaction, coagulant dosage * ph (p = 0.0020), and two main effects coagulant dosage (p = 0.0000) and ph (p = 0.0000) is significant, and the square term (coagulant dosage * coagulant dosage) and (ph * ph) also significant, p=0.0000. The square term indicate the non-linear characteristic of the coagulation behaviour.

31 Table 4.2: Turbidity results for RSM experiments RUN ph Coagulant (mg/l) Final Turbidity (NTU) RSM1 RSM2 RSM3 RSM4 RSM5 RSM6 4.6 4.6 7.4 6.0 7.4 6.0 34 106 34 20 106 120 3.21 1.10 1.15 2.14 0.02 0.087 RSM7 RSM8 RSM9 RSM10 RSM11 RSM12 4.0 8.0 6.0 6.0 6.0 6.0 70 70 70 70 70 70 2.60 0.10 0.50 0.50 0.50 0.50 RSM13 6.0 70 0.50 Table 4.3: Result of the analysis for response surface using MINITAB TM software Term Coefficient Sum of Error T Test P Coefficient Constant 0.5000 0.04539 11.016 0.0000 Dosage of -0.7692 0.03588-21.436 0.0000 Coagulant ph -0.8344 0.03588-23.255 0.0000 Coagulant x 0.3400 0.03848 8.836 0.0000 Coagulant ph x ph 0.4600 0.03848 11.955 0.0000 Coagulant x ph 0.2450 0.05074 4.828 0.0020

32 Table 4.4: Analysis of variance for turbidity Source DF Seq. SS Adj. MS F P *Linear 2 10.3032 5.15160 500.16 0.000 *Square 2 2.0268 1.01342 98.39 0.000 *Interaction 1 0.2401 0.24010 23.31 0.002 Residual Error 7 0.0721 0.01030 Total 12 12.6422 *Linear: coagulant dose and ph, *square: coagulant 2 and ph 2, *Interaction: Coagulant dose x ph The analysis of variance in Table 4.4 gives a summary of the main effect (linear and square) and interaction. The total sum of squares is 12.6422 (linear=10.3032, square=2.0268, interaction=0.2401) with total of 12 degree of freedom. A relationship between the dose and ph and the final turbidity could be described as Final Turbidity = 17.35-0.17 Coagulant - 3.75 ph + 0.001 Coagulant 2 + 0.23 ph 2 + 0.01 Coagulant x ph

33 Figure 4.2 and Figure 4.3 illustrate the contour and response surface plots described by the above equation. In the contour plot, the minimum final turbidity is 0.01 NTU when coagulant dosage is 106 mg/l of ph 7.0. This predicted final turbidity of 0.01 mg/l is almost 95 % lower than the lowest final turbidity obtained from OFAT experiment. The lower turbidity value was found by searching the entire area inside the square using the response surface design. The lowest turbidity, 0.01 NTU are found where ph and coagulant dosage both at their high setting. As the ph and coagulant dosage move backward on their lower setting, the value of turbidity increased. It was shows that the better water quality could be achieved using response surface method that the traditional OFAT approach. The interaction between the factors (coagulant dosage and ph) and response (turbidity) can be estimated for the response surface design, but it cannot be estimated for the OFAT experiment.

34 Figure4.2: contour plot of the response surface for the central composite design experiment

Figure 4.3: 3-Dimensional response surface for turbidity 35

36 CHAPTER V CONCLUSION AND RECOMMENDTATION 5.1 CONCLUSION Response surface design is a more effective way to determine the impact of two or more factors on a response than a One-Factor-At a-time (OFAT) experiment, where only one factor is changed at one time while the other factors are kept fixed. So, basically the conclusions that can be drawn from this study are as follow: 1. In OFAT experiment, the lowest final turbidity achieved in set one is 0.50 NTU with coagulant dosage of 100 mg/l at ph 7.0 and final turbidity achieved in set two was 0.30 NTU when the ph equal to 6.5 and the coagulant dosage of 100 mg/l. If only two sets of experiment were conducted as in normal procedure, the final turbidity obtained was not the lowest one.

37 2. In response surface method experiment, the lowest final turbidity achieved was 0.02 NTU in RSM 5 (Table 4.2), when the coagulant dosage was 106 mg/l and the ph is 7.4. 3. For response surface analysis, the coagulant dosage and ph were significant factors in the coagulation process at the confidence level of 95%. 4. The two ways interaction, coagulant dosage * ph (p = 0.0020), and two main effects coagulant dosage (p = 0.0000) and ph (p = 0.0000) is significant, and the square term (coagulant dosage * coagulant dosage) and (ph * ph) also significant, p=0.0000. The square term indicate the non-linear characteristic of the coagulation behaviour. 5. A relationship between the coagulant dose and ph and the final turbidity could be described as, 17.35-0.17 Coagulant - 3.75 ph + 0.001 Coagulant 2 + 0.23 ph 2 + 0.01 Coagulant x ph 6. The lowest turbidity, 0.01 NTU (Figure 4.1 and Figure 5.2) are found where ph and coagulant dosage both at their high

38 setting. As the ph and coagulant dosage move backward on their lower setting, the value of turbidity increased. 7. As overall conclusion, the better water quality could be achieved using response surface design rather than OFAT experiment. The interaction between the factors and response can be estimated for the response surface design, but it cannot be estimated for the OFAT experiment. 5.2 ADVANTAGES OF RESPONSE SURFACE METHOD OVER OFAT RXPERIMENT Response surface method provides a powerful means to achieve quality and efficiency in Jar Test experimental process because; a) It requires less resource such as experiments, time, material, etc. for amount of information obtained. This can be important in finding the optimum ph, optimum coagulant dosages and lowest turbidity, where time is consuming. b) There is experimental information in a larger region of the factor space. This improves the prediction of the response in factor space by reducing the variability of the estimates of the

39 response in the factor space, and makes process optimization more efficient because the optimal solution is searched for over the entire factor space. c) The estimates of the effect of each factor are more precise. Using more observations to estimate an effect results in higher precision (reduced variability). As an example, for the response surface, all the observations are used to estimate the effect of each factor and each interaction, while in OFAT experiment, typically only two of observations are used to estimates the effect of each factor.

40 REFERENCES 1. Benefield, L.D, Judkins, J.F and Weand, B.L (1982), Process Chemistry for Water and Wastewater Treatment, Prentice- Hall, Inc. (pp211-235) 2. Montgomery, D.C and Wiley, John (1996), Design and Analysis of Experiments, 4th edition, Ch. 14. 3. Vesilind, P.A, Peirce, J.J and Weiner, R. (1994), Kejuruteraan Alam Sekitar, Edisi Kedua, Noraini Jaafar (Translator), Unit Penerbitan Akademik, U.T.M, Skudai (pp97-98) 4. SAM4501, Laboratory manual (2004/2005), Faculty of Civil Engineering, Universiti Teknologi Malaysia. 5. Reynolds T.D and Richard P. (1996), Coagulation and Flocculation, PWS publishing Company. 6. Czitrom, Veronica (1999), One-Factor-at a Time versus Designed Experiment, The American Statistician, Vol. 53, No. 2,(pp126-13) 7. Guan,X.L. and Melchers, R.E (2000), A parametric Study on the Response Surface Method, 8 th ASCE Specialty Conference on Probabilistic Mechanics and Structural Reliability, PMC 2000-02