EFFECT OF RAW WATER QUALITY ON COAGULANT DOSAGE AND OPTIMUM pH YANNIE ANAK BENSON A project report submitted in partial fulfillment of the requirements for the award of the degree of Master of Engineering (Civil-Environmental Management) Faculty of Civil Engineering Universiti Teknologi Malaysia 16 NOVEMBER 2006 iii Praise the LORD GOD because of the blessed and strength HE gave unto me, Sincere thanks to my lovely husband, Johnny and daughter, Jwelyn Ystefanie because of your moral support, sacrifice and became my backbone, Thanks Apak and Mama for your love all these years, In loving memory brother, Jeruslavin Benson (18th June 2003); You always in our heart and this is a gift for you.. iv ACKNOWLEDGEMENT I would like to thank Dr. Azmi Aris for his patience, dedication and excellent supervision. Without him, my Master Project would not excellent and succesful completed. I also would like to extend my gratitude to all Environmetal Lab’s technician especially Mdm. Rosmawati because she always be there when I need the assist in laboratory work. Special thanks to Ooi Boon Siew because of her dedication in teaching me the basic to explore MINITABTM statistical software. Not forget to Jini Anak Gilbert Malandang, Nadiah and Zul Said, thank you for their companionship and advice. My sincere thanks to all who involved in this project that I did not mentioned their name. Lastly to all my family members; thank you so much for their companionship and sporting all these years. v ABSTRACT Removal of turbidity, suspended solids (SS) and natural organic matter (NOM) using coagulation are well known because of the ability of the process in destabilizing the colloids particles and reducing the repulsion force between the particles. The objectives of the study are to explore the effect of the selected water quality parameters (i.e initial pH, initial temperature and SS) and to develop a statistical relationship between the water quality parameters and the optimum dosage and pH. The study was conducted using jar test procedures using synthetic water prepared using kaolin as the source of SS. The experiments were designed using Response Surface Method (RSM) with final turbidity as the response. RSM was found to be better approach than one-factor-at-a-time (OFAT) in determining the optimum dose and pH. Initial pH and SS was found to have significant effect to optimum dose at 90% confidence level (α = 0.1) and temperature was the only factor having significant effect on optimum pH at 80% confidence level (α = 0.2). Probably due to the complexity of the nature of the coagulation process, the relationship between the parameters and the response was only developed for optimum pH. vi ABSTRAK Penggunaan proses pengentalan untuk menyingkirkan kekeruhan, bendasing terampai dan jirim semulajadi organik didalam air sangat popular kerana kebolehannya dalam menidakstabilkan zarah-zarah koloid dan mengurangkan daya tolakan di antara zarah. Objektif utama kajian ini adalah untuk melihat kesan daripada parameter kualiti air yang terpilih (pH awal, suhu awal dan kepekatan pepejal) dan juga untuk menghasilkan hubungkait statistik antara parameter kualiti air dengan dos optimum dan pH optimum. Kajian telah dijalankan menggunakan prosedur Ujian Balang dengan penggunaan air sintetik yang telah disediakan menggunakan kaolin sebagai pepejal terampai air. Ujikaji telah direkabentuk menggunakan Kaedah Response Surface (RSM) dengan mengambilkira kekeruhan adalah sebagai hasil tindakbalasnya. Didapati bahawa RSM adalah jauh lebih bagus berbanding dengan pendekatan menggunakan Satu-Faktor-Pada-Satu-Masa (OFAT) dalam menentukan pH optimum dan dos optimum. Adalah didapati bahawa pada 90% tahap keyakinan (α = 0.1), pH awal dan nilai awal pepejal terampai mempunyai kesan yang penting terhadap dos optimum manakala hanya suhu sahaja didapati memberi kesan penting terhadap pH optimum pada 80% tahap keyakian (α = 0.2). Hubungkait antara parameter dan hasil tindakbalas hanya boleh dibangunkan untuk pH optimum, kemungkinan besar disebabkan oleh tindakbalas pengentalan yang agak kompleks. vii TABLE OF CONTENTS CHAPTER 1 2 TITLE PAGE TITLE PAGE i DECLARATION ii DEDICATION iii ACKNOWLEDGEMENTS iv ABSTRACT v ABSTRAK vi TABLE OF CONTENTS vii LIST OF TABLES x LIST OF FIGURES xiii LIST OF SYMBOL xvi INTRODUCTION 1.1 Preamble 1 1.2 Problem Statement 2 1.3 Aim 3 1.4 Objectives 3 1.5 Scope and Limitation of The Study 3 LITERATURE REVIEW 2.1 Introduction 5 viii 2.2 Colloidal Stability 5 2.3 Theory of Coagulation 8 2.4 Common Chemical Used As Coagulant 10 2.5 Factor Affecting Coagulation 11 2.5.1 Coagulant Dose 12 2.5.2 Turbidity 13 2.5.3 Natural Organi Matter 14 2.5.4 Alkalinity and pH 15 2.5.5 Temperature 15 2.5.6 Mixing Speed 16 2.5.7 Treatmnet process 17 2.6 Common Experiment For Coagulation 17 2.7 Experimental Design 18 2.7.1 One-Factor-At-A-Time and Matrix 18 2.7.2 Response Surface Method 20 2.7.2.1 21 Factorial / Fractional Factorial Design 2.7.2.2 3 Central Composite Rotatable Design 25 METHODOLOGY 3.1 Equipments and Materials 29 3.2 Experimental Procedure 30 3.2.1 Preliminary Study 30 3.2.1.1 One-Factor-At-A-Time 30 3.2.1.2 Response Surface Method 32 ix 3.2.2 Raw Water Quality - Alum Dosage and pH 33 Relationship 4 5 RESULTS AND DISCUSSION 4.1 Preliminary Study 37 4.2 Preliminary Response Surface Method 39 4.3 Water Quality - Optimum Dose And pH Relationship 40 4.3.1 Response Surface Analysis 41 4.3.2 Lowest FTU Approach 42 4.3.2.1 Factorial Analysis 44 4.3.2.2 Response Surface Analysis 47 CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions 56 5.2 Future Study Recommendations 57 REFERENCES 58 APPENDIX A (DESIGN TABLE FOR 20 SET) 63 APPENDIX B (DETAILS FOR FACTORIAL ANALYSIS) 74 APPENDIX C (RESPONSE SURFACE ANALYSIS FOR FACTORS THAT AFFECT COAGULATION - FULL 76 QUADRATIC TERMS) APPENDIX D (RESPONSE SURFACE ANALYSIS FOR FACTORS THAT AFFECT COAGULATION - LINEAR + SQUARED TERMS) 79 x LIST OF TABLES TABLE NO. TITLE PAGE CHAPTER 1 INTRODUCTION CHAPTER 2 LITERATURE RIVIEW 2.1 Settling Velocity of various sizes of colloidal particles* (Source: Peavy et. al., 1985) 6 2.2 Guideline of the nature of NOM and expected TOC removals (Source: Edzwald and Tobiason, 1999) 14 2.3 Typical pattern of a 2-level, 3-factor full factorial design 24 2.4 An example of fully expanded 23 factorial 24 2.5 Guide to the Central Composite Rotatable Design and as for a 2k Full Factorial (Diamond, 2001) 25 CHAPTER 3 METHODOLOGY 3.1 Experimental run for OFAT approach 31 3.2 Experimental run in RSM approach 33 3.3 Setting of water characteristics used in the study 35 3.4 Design table for range 20 mg/L to 120 mg/L of alum dosage 36 3.5 Design table for range 40 mg/L to 200 mg/L of alum dosage 36 xi CHAPTER 4 RESULTS AND DISCUSSIONS 4.1 Results of the OFAT experiment on the turbidity removal (Initial turbidity = 31 FTU) 38 4.2 Turbidity results from the RSM experiment in preliminary works (Initial turbidity = 31 FTU) 40 4.3 The summary of response surface regression for 20 sets of experiment 42 4.4 Optimum pH and dosage for the experiments based on Lowest FTU approach 43 4.5 ANOVA for turbidity removal at optimum dose and optimum pH 44 4.6 The summary of the response surface analysis for the water quality effect on optimum dose 48 4.7 The summary of the response surface analysis for the water quality effect on optimum pH 48 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS APPENDICES A-1 Design table of RSM for Set 1 64 A-2 Design table of RSM for Set 2 64 A-3 Design table of RSM for Set 3 65 A-4 Design table of RSM for Set 4 65 A-5 Design table of RSM for Set 5 66 A-6 Design table of RSM for Set 6 66 A-7 Design table of RSM for Set 7 67 A-8 Design table of RSM for Set 8 67 xii A-9 Design table of RSM for Set 9 68 A-10 Design table of RSM for Set 10 68 A-11 Design table of RSM for Set 11 69 A-12 Design table of RSM for Set 12 69 A-13 Design table of RSM for Set 13 70 A-14 Design table of RSM for Set 14 70 A-15 Design table of RSM for Set 15 71 A-16 Design table of RSM for Set 16 71 A-17 Design table of RSM for Set 17 72 A-18 Design table of RSM for Set 18 72 A-19 Design table of RSM for Set 19 73 A-20 Design table of RSM for Set 20 73 xiii LIST OF FIGURES FIGURE NO. TITLE PAGE CHAPTER 1 INTRODUCTION CHAPTER 2 LITERATURE RIVIEW 2.1 A negative colloid particle with its electrostatic field (Source: Reynolds and Richards 1996; McGhee, 1991) 7 2.2 Ionic compression or repulsion force: (a) Charge system in a colloidal suspension which shows the reduction of thickness in diffused layer (b) Reduction of net force 8 2.3 Schematic actions of forces acting on hydrophobic colloids in stable suspension 9 2.4 Conceptual view of reaction on coagulation mechanism (Source: Pernitsky, 2003; Pernitsky, 2001; Dennett et. al., 1996) 10 2.5 The alum coagulation diagram and its relationship to zeta potential (Source: AWWA and ASCE, 1990) 13 2.6 Diagram of One-Factor-At-A-Time shows of the turbidity versus alum dose (Source: Adopted from Czitrom, 1999) 19 2.7 Matrix in 3D view. (Source: Aris, 2004) 20 2.8 Full Factorial Design with 2-level and 3 factors in 3D view 22 2.9 One-factor-at-a-time vs factorial (Source: Czitrom, 1999) 22 2.10 Central composite rotatable design in 3D view 26 xiv 2.11 The view of 3D CCRD with 2 factors (F: 4 runs of full factorial points or cube; S: 4 axial points or star points; C: 1 centre point). (Source: Czitrom, 1999) 26 2.12 Central composite rotatable design for three factors used in effect of raw water quality on coagulant dosage study 28 CHAPTER 3 METHODOLOGY 3.1 Flow chart of OFAT experiment 31 3.2 Flow chart of experimental design 34 CHAPTER 4 RESULTS AND DISCUSSIONS 4.1 Effect of alum dose on final turbidity (pH 7) 38 4.2 Effect of pH on final turbidity (Alum dose = 40 mg/L) 39 4.3 Pareto chart for (a) Optimum Dose and (b) Optimum pH based on final turbidity (A: Temperature; B: pH; C: SS) 45 4.4 Main effect plot for water quality parameters at (a) Optimum Dose (b) Optimum pH 46 4.5 Interaction effect plot for water quality parameters at optimum dose 46 4.6 Interaction effect plot for water quality parameters at optimum pH 47 4.7 (a) Contour and (b) response surface plots representing relationship between pH and temperature at optimum pH 50 4.8 (a) Contour and (b) response surface plots representing relationship between temperature and SS at optimum pH 51 4.9 (a) Contour and (b) response surface plots representing relationship between pH and SS at optimum pH 52 4.10 (a) Contour and (b) response surface plots representing relationship between pH and temperature at optimum dose 53 xv 4.11 (a) Contour and (b) response surface plots representing relationship between temperature and SS at optimum dose 54 4.12 (a) Contour and (b) response surface plots representing relationship between pH and SS at optimum dose 55 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS APPENDICES xvi LIST OF SYMBOL mg - milligrams g - grams L - litre mL - millilitre °C - Degree Celsius FTU - Formazin Turbidity Unit NTU - Nephelometric Turbidity Unit p-value - level of confidence in percentage CaCO3 - calcium carbonate CHAPTER 1 INTRODUCTION 1.1 Preamble In drinking water treatment, coagulation process is use to destabilize colloidal materials or contaminants. Followed by solid-liquid separation processes such as flocculation, sedimentation, or dissolved air flotation (DAF) and filtration, the processes are capable to remove the colloidal particles from the water (Pernitsky, 2001). Chemicals that are used for coagulation process is called coagulant. Currently, there are many types of coagulant available in wastewater treatment but the most frequently used are alum or ferric sulphate. These chemical coagulants are in positive charged and it will react with colloidal suspension of organic and inorganic solids that are usually negatively charged. Besides the man-made coagulant, other traditional coagulants originated from the plant origin such as Moringa Olerfera seeds which can be found in India and Strychnos Potatorum are seldomly used. The uses of other traditional coagulant from soil origin include bentonite or clay, algae, chitosan and dough from millet bread are also reported (Anselme and Narasiah, 1998). 2 Many factors have been reported to affect the coagulation process. These include turbidity, organic matter and pH, ultraviolet (UV), alkalinity or acidity and temperature (Pernitsky, 2003). While some studies have been conducted to relate these parameters to coagulant dosage, the standard method commonly used to determine the coagulant dosage is by using Jar Test. 1.2 Problem Statement The effectiveness of the coagulation process is highly dependent on the dosage of the coagulant and the pH of the water during the process. However the effectiveness of this process and the relationship between the raw water quality and the coagulant dosage and optimum pH can hardly be predicted until today mainly due to the complexity of the chemistry of the coagulation process. Hence, the dose of coagulant and pH of the process mainly depend on the results of the Jar test which is conducted at the water treatment plant. Typically, the Jar Test will be carried out in a daily basis and also in the event of changes in raw surface water characteristics. Since Jar Test is a tedious experimental process and time consuming, this study intends to develop a relationship between raw water quality parameters and the optimum coagulant dosage and the pH based on statistical approach. Such relationship is anticipated to ease the operator in plant to determine the optimum dosage and pH in the coagulation process. 3 1.3 Aim The aim in this study is to ease the process of determining the optimum chemical dosage and pH for coagulation process in water treatment. 1.4 Objectives There are two main objectives of the study: a) To explore the effect of the selected water quality parameters on the optimum dosage of coagulant and pH. b) To develop a statistical relationship between the selected water quality parameters with the optimum coagulant dosage and pH. 1.5 Scope and Limitation of The Study The study covers a comprehensive experimental works at laboratory scale. Synthetic water prepared by using kaolin was used in the study. The experimental work was designed using Response Surface Method (RSM). Three independent water quality variables were chosen, namely initial turbidity, pH and temperature. Optimum coagulant dosage and pH were used as the response variables based on the 4 lowest turbidity achieved after the jar test. Aluminium sulphate (Alum) was used as the coagulant. CHAPTER 2 LITERATURE REVIEW 2.1 Introduction Coagulation is one of the important processes in water treatment. It is an effective method in removing natural organic matters (NOM) and colloidal particles in high and intermediate molecular weight ranges (Sinsaubaugh et. al., 1986). This process has a direct impact on the reliability of plant operations and final water quality and has significant contribution to the operational cost of the treatment plant (Hooge, 2000). This chapter discuss on coagulation process, the phenomenon of coagulation, the mechanism involved, important of water parameter in coagulation process and the coagulants. 2.2 Colloidal Stability A large number and variety of substances that produce turbidity can be found in the waters. Examples of these include clay particles, organic matter 6 from decaying vegetation and animal. These particles especially colloids, range in size from 1 millimicrons to 500 millimicrons (nanometers) and it is not visible when using an ordinary microscope and do not easily settle ion of solution (Hammer and Hammer Jr., 2005). The terms stable are used for colloidal suspensions that do not agglomerate naturally. It is stable as the individual particle has such a large surface area relative to its weight that gravity forces do not influence its suspension (Peavy et. al., 1985). Table 2.1 illustrates the setting velocity of different particles with respect to their diameter. Table 2.1: Settling Velocity of various sizes of colloidal particles* (Source: Peavy et. al., 1985) Particle diameter (mm) Size typical of Settling velocity 10 Pebble 0.73 m/s 1 Coarse Sand 0.23 m/s 0.1 Fine Sand 1.0 x 10-2 m/s (0.6 m/min) 0.01 Silt 1.0 x 10-4 m/s (8.6 m/d) 0.0001 Large colloid 1.0 x 10-8 m/s (0.3 m/yr) 0.000001 Small colloid * Spheres with specific gravity of 2.65 in water at 20°C. 1.0 x 10-13 m/s (3 m/million yr) There are two types of colloids, namely hydrophobic and hydrophilic. Hydrophobic colloids are those that do not have affinity with water while hydrophilic are colloids that have affinity with water. Its particles depend on electrical charge for their stability in suspension. For hydrophobic colloids, individual particles are held apart by electrostatic compression or repulsion forces by positive ions adsorbed onto their surfaces from solution (Reynolds and Richards 1996; McGhee, 1991). Figure 2.1 on the next page illustrates the theory of double layer of electrostatic. This 7 colloidal type may be found in the bulk of inorganic and also organic matter in turbid waters. Rigid or Fixed Layer / Stern Layer attached to particles Stern potential Surface/ Nerst potential Figure 2.1: A negative colloid particle with its electrostatic field. (Source: Reynolds and Richards 1996; McGhee, 1991) Hydrophilic colloids are stable because of their attraction for water molecules rather than because of the slight charge that they might posses. The typical examples of the hydrophilic colloid are as soap, soluble starch, synthetic detergents and blood serum. These colloids are not easy to remove from suspension and thus required 10 to 20 times more coagulant than the dosage normally the used in the conventional water treatment (Hammer and Hammer Jr., 2005). Natural forces of attraction also exist between two particles and are called Brownian movement and Van Der Waals’ force. Brownian movement is a random 8 motion of colloids caused by bombardment of water molecules which is tend to enhance this physical force of attraction in pulling the particles together (Pushkin, 2004). In both cases, if the repulsion is over the attraction, the particles are not allowed to contact each other and these hinder the flocculation process (Figure 2.2). Figure 2.2: Ionic compression or repulsion force: (a) Charge system in a colloidal suspension which shows the reduction of thickness in diffused layer (b) Reduction of net force 2.3 Theory of Coagulation Coagulation process in raw water treatment is the process of charge neutralization of colloidal particles using the addition of a chemical reagent or the process of conditioning suspended solids particles to promote their agglomeration thus produces larger particles that can be more readily removed in subsequent treatment processes (AWWA and ASCE, 1990; Jin, 2005). 9 The effectiveness of the coagulation process in water treatment practice depends on the interaction of the coagulant species with particles and dissolved organic material in the raw water. If both of the surface charges are similar, the molecules of the contaminants will repel with each other (Sawyer et. al., 1978). The basis examining coagulation can be referring by a conceptual overview of the reactions that occur during the coagulation process on Figure 2.3 (Pernitsky, 2001). Attraction Repulsion Figure 2.3: Schematic actions of forces acting on hydrophobic colloids in stable suspension There are four mechanisms that contribute the coagulation process, namely enmeshment or sweep-floc particles, charge neutralization or destabilization, complexation or precipitation and adsorption. Figure 2.4 illustrates the conceptual of possibly reaction occur during coagulation mechanism. All these mechanisms were categorizing as primary reaction mechanisms and they may exists either by themselves or they also may exists in combination due to the complexity of the nature of the coagulation process coagulation process when chemical coagulant was added (Edzwald and Van Benscoten, 1990). Details discussions on these mechanisms are given in Pernitsky, 2003; Amirtharajah and Mills, 1982; Dempsey et. al., 1985; Randtke, 1988; Edzwald and Van Benschoten, 1990; Dempsey, 1994. 10 ACTIVE COAGULANT SPECIES COAGULANT Alum Hydrolysis CONTAMINANTS PRODUCTS MECHANISM Al(OH)3 (am) + Colloids A: Enmeshment NOM + Al(OH)3 (am) B: Adsorption Al=Colloid C: Charge Neutralization/ Desztabilization Al=NOM Al+NOM(am) D: Complexation/ Precipitation Colloids Al(OH)3 (am) NOM 3+ Al , AlOH2Al(OH)4- Al3+, SO42- 3+ H+ Alkalinity consumption Al , Al(OH)2+ Colloids NOM Figure 2.4: Conceptual view of reaction on coagulation mechanism (Source: Pernitsky, 2003; Pernitsky, 2001; Dennett et. al., 1996) 2.4 Common Chemical Used As Coagulant Most widely used coagulant in drinking water treatment is based on aluminium, called aluminium sulphate (Al2 (SO4)3) or alum, filter alum or alumina sulphate. Coagulants react with available alkalinity such as carbonate, bicarbonate and hydroxide or phosphate to form insoluble aluminum salts. Equation 2.1 shows the reaction between aluminium and natural alkalinity. As metal salts, alum will react with alkalinity in the water to produce an insoluble metal hydroxide floc which incorporates the colloidal particles. The addition of alum was also found to remove a large proportion of the high molecule weight of NOM compounds with the decrease in the number-average of molecule weight values (Ho, 2005). Al2(SO4)3.14.3H2O + 3Ca(HCO3)2 = 2Al (OH)4 + 3CaCO4 + 14.3H2O + 6CO2 Equation 2.1 11 Iron-based coagulant in the form of ferrous sulphate or copperas (FeSO4.7H2O), ferric sulphate (Fe2(SO4)), ferric chloride (FeCl3) and the mixture of Fe2(SO4) and FeCl3 are also commonly used in water treatment. In commercial product, ferric sulphate is available in the form of a reddish-brown granular material and it is readily soluble in water (Tebbuti, 1973). While aluminium-based coagulant in the form of alum and it can be found in round-white granular. Alum not only becomes a favourite coagulant but also no need to add lime or soda ash along with its usage. Ferric sulphate is effective over a wide range of pH. It is successful on colour removal at low pH values and may be used for iron and manganese removal and as well as a coagulant in precipitation softening at high pH (Hammer and Hammer Jr., 2005). Equation 2.2 shows the reaction between ferric sulphate and the natural alkalinity of water. Fe2(SO4)3 + 3Ca(HCO3)2 = 2Fe(OH)3 ↓ + 3CaSO4 + 6CO2 2.5 Equation 2.2 Factor Affecting Coagulation There are seven factors found likely to affect the coagulation process in removing the NOM and colloid particles or turbidity. These are coagulant dose, turbidity or SS, natural organic carbon (NOC), pH and alkalinity, temperature, mixing speed and the treatment process. 12 2.5.1 Coagulant Dose Coagulant dose is a key of process-control factors because of the different water quality conditions can have an effect on different dose of coagulant (Budd et. al., 2004). It was found that coagulant doses are controlled by dissolved organic carbon (DOC) concentration rather than by turbidity in most surface waters (Edzwald and Van Benschoten, 1990). These makes an adjustment on coagulant dosage is important to ensure the effectiveness for maintaining settled raw water quality in plants. High concentrations of coagulant would reduce the time of particles to destabilise by increasing the collision-attachment efficiency which makes bulky precipitate enmeshes particles settles rapidly to form the ‘sweep-floc’ region of coagulation (Reed et. al., 1999; Sanks 1979). Usage of alum dose less than 5 mg/l is believed to result in charge neutralization (destabilization) would be the primary mechanism while at more than 5 mg/L, an entrapment mechanism would be predominant. In considering of turbidity and NOM removal, the optimum of alum coagulants required is 1 mg/L which it was absolute depend on both Specific Ultraviolet Absorbance (SUVA) and NOM concentration (Pernitsky, 2001). Figure 2.5 depicts some of the aluminum species involved in alum coagulation and the conditions of aluminum concentration and pH under which they occur (AWWA and ASCE, 1990). 13 Figure 2.5: The alum coagulation diagram and its relationship to zeta potential (Source: AWWA and ASCE, 1990) 2.5.2 Turbidity Turbidity is a very important factor in water treatment and it may control the coagulation process if low concentration of total organic carbon (TOC) in raw water. Higher coagulants doses, longer flocculation time and lower filtration rates are much required in high water turbidity (≥ 100 NTU) (Pernitsky, 2003). The optimum turbidity removal and solid-liquid separation often occurs in low solubility of coagulant’s pH (Eikebrokk, 1990; Edzwald and Van Benschoten 1990; Bell-Ajy et. al., 2000). However, turbidity has less effects on coagulant dose when the colloidal particles removal through adsorption onto or sweep-floc mechanism (Pernitsky, 2001). 14 2.5.3 Natural organic matter Natural organic matter also may contribute to poor water quality. It was characterizing by SUVA and the concentration was measures by TOC, DOC or ultraviolet (UV) (Pernitsky, 2001). Natural organic matter completely predominant in water supplies which has SUVA more and equal to 2 but have less effect on coagulation as the decreasing of SUVA. At this condition, coagulant dose seems to increase with increasing of TOC. Table 2.2 shows the guidelines for interpretation of SUVA value in NOM. Table 2.2: Guideline of the nature of NOM and expected TOC removals (Source: Edzwald and Tobiason, 1999) SUVA <2 Composition Coagulation DOC Removals Mostly non-humics NOM has little influence < 25% for alum, Low hydrophobicity, Poor DOC removals Little greater for ferric Mixture of aquatic humics and other NOM, NOM influences 25-50% for alum, Mixture of hydrophobic and hydrophilic NOM, DOC removals should be fair to good Little greater for ferric NOM controls > 50% for alum, Good DOC removals Little greater for ferric Low molecular weight 2–4 Mixture of molecular weights >4 Mostly aquatic humics, High hydrophobicity, High molecular weight 15 2.5.4 Alkalinity and pH Alkalinity and pH are two different things because alkalinity is a measured of water capacity to neutralize acids while pH is the unit in measuring the level of acidic in water. Addition of 1 mg/L of alum might neutralize 0.5 mg/L of hydrogen ions (Tang et al, 1998). Low concentration of alkalinity might destroy a buffering capacity and also makes rapidly dropping in water pH. In maintaining the optimum pH, it is required an artificial buffer because pH was emphasized as a critical process condition for coagulation (Peavy et. al., 1985; Amirtharajah & Mills, 1982). While a high alkalinity water would be highly buffered and exhibit a limited pH decrease after coagulant (Budd et. al., 2004). A level of response in water varies depending on alkalinity and it is dominated by the dissolve inorganic carbon species (carbonate and bicarbonate) in the range of pH in most natural waters. At 20°C, a pH for minimum polyaluminum solubility was in ranged 6.0 to 6.7 (Pernitsky, 2001). Coagulation at pH less than 7.5 with alum would decrease and become a significant problem for high alkalinity water and it is recommended to maximize TOC removal of alum when the pH is very low (less than pH 5.5) (Pernitsky, 2003). Although sweep coagulation occur when negatively-charged forms of alum predominate in range pH 6 to 8, an optimum pH for coagulation is within the range of 5.5 to 7.5, while 5.0 to 8.5 in the treated water with ferric, respectively. 2.5.5 Temperature Low temperature in water affects the chemical properties of water (reaction rates, solubility of solids and liquids), pH and hydrolysis species of coagulants and mostly in sedimentation. Reaction rates and reaction kinetics decrease with reducing 16 temperature which the relationship is expressed with Arrenius empirical rate law (Snoeyink and Jenkins 1980). Solubility may be considered as a function of temperature in which pH of minimum solubility for aluminum and of most solids decrease as the temperature decreases (Al-Laya and Middlebrooks, 1974; Odegard et. al., 1990; Dempsey, 1994; Edzwald et. al., 1994). However, an increase of temperature in endothermic reactions and a positive enthalpy data and also in equilibrium makes the solubility increase (Bagwell et al., 2001, Pernitsky, 2003). Decreasing in water temperature was accompanied by a decreasing in turbidity removal because of the deposition of Al(OH)3 on the particulates surface significantly reduced with decrease of temperature (Kang et al., 1995; Hooge, 2000). Recommendation on slow hydrolysis and precipitation reactions of metal coagulants in cold water was beneficial to flocculation loading rates, perhaps by permitting hydrolysis species to react more extensively with turbidity and with humic substances (Schultz et al,. 1984; Shea et al., 1971; Adhin et al., 1974). 2.5.6 Mixing Speed Rapid mixing is utilized as part of the coagulation process to distribute the coagulant chemicals throughout the water stream. When alum or ferric chloride is used to achieve destabilization through charge neutralization, it is extremely important that the chemical coagulant is efficiently distributed because the intermediate products of the coagulant reaction are the destabilizing agents. Thus, because of the short life of these intermediate species; coagulant must be contact with the solids particles if the destabilization is to be achieved. However, in some cases, excessive rapid mixing may serve to break-up coagulant molecules or floc particles, thereby reducing the effectiveness of subsequent solids removal processes (EPA, 1999). 17 Mixing intensity is typically quantified with a number known as the velocity gradient or G value and this value is a function of the power input into the rapid mixing process and the volume of the reaction basin. Typical G values for rapid mixing coagulation were in range from 300 to 8000 sec–1 (Hudson, 1981). Time required to achieve efficient coagulation varies were depending on the coagulation mechanism involved. When the charge neutralization is involved, the detention time required may be one second or less but if the sweep-floc or entrapment involved, a longer detention times (1 - 30 seconds) may be appropriate (Kawumara, 1991; AWWA and ASCE, 1998; Hudson, 1981) 2.5.7 Treatment Process A physical forces also to be believe, one of the affecting factor involved with the effectiveness of the raw water treatment for different of it treatment processes (Pernitsky, 2001). As for direct filters, it is needs or requires the effective charge of neutralization and the production of small floc particles. This element is important to minimize the headloss of the filters. On the other hand, sedimentation requires the precipitation of a dense, strong and rapidly settling floc. While treatment using dissolves air flotation (DAF) approach requires effective charge neutralization but at the same time, it may tolerate reasonably large amounts of precipitation floc. 2.6 Common Experiment For Coagulation Jar Test is a common laboratory procedure to determine the optimum coagulant dosages for water and wastewater treatment. This test simulates the coagulation and flocculation processes that encourage the removal of suspended 18 colloids and organic matter which can lead to turbidity, odor and taste problems. The range of mixing time are 1 minutes to ensure complete dispersion of the chemicals and then mixed slowly for 15 to 20 minutes to aid in the formation of flocs and also approximately 30 minutes to allow settlement or until clarification has occurred. The results from this test are used to determine the quantity of coagulant to be used in the water treatment plant. 2.7 Experimental Design Two methods has been preferable to solving problem in drinking water treatment are traditional method and the statistical method. Traditional methods were using one-factor-at-a-time (OFAT) and matrix while statistical method was using response surface method (RSM). The purposes of these approaches usage are to find the most effective, truly important and efficiency variables or winning combinations variables in coagulant dosage. 2.7.1 One-Factor-At-A-Time and Matrix One-factor-at-a-time (OFAT) approach is the most common traditional problems solving method used by researchers nowadays as alternatives to analyse and statistical experimental design (Figure 2.6). In this method, researcher conducts the tests in systematically changing the levels of one factor but in the same time keeping the others in constant and then study the behavior of the system at several levels of those factors (Haaland, 1989). 19 Turbidity (NTU) 4 3 2 1 40 60 80 Alum Dose (mg/L) Figure 2.6: Diagram of One-Factor-At-A-Time shows of the turbidity versus alum dose (Source: Adopted from Czitrom, 1999) This method will find the best value for each factor and the process of this finding value is repeated for the remaining factors until all factors have been considered. Although it is simple and straight forward method but it also ineffective because it is suffers from several shortcomings which may lead to wrong conclusions (Haaland, 1989; Hendrix, 1979). It also takes too many experiments or trial to come out with the answer. Eventhough it does not require many measurements in one space, in particular OFAT not capable to identifying the interactions effects from more than one factor because it is not fully explore the space of possible solutions and may miss the solution (Haaland, 1989; Adrion et al., 1984). This method can be simplify said that “A botched design gives very little information for a lot of work” (Box, 1965). It is often called botched because the design so poor. A second traditional approach to statistical and analyze experimental design is with laying out the matrix of all possible combinations of the important factors. Figure 2.7 illustrate the matrix in 3D view. If the OFAT is the simplest method, matrix show the most complicated method because it is fully explores the 20 experimental space until the solutions are identified. Eventhough this method overcomes the shortcomings that occur in OFAT approach and effective, however it is inefficient because it requires an unnecessarily large number of measurements which this costly the implementation (Haaland, 1989; Adrion et al., 1984). Figure 2.7: Matrix in 3D view. (Source: Aris, 2004) 2.7.2 Response Surface Method Response Surface Method is one of the statistical methods which are providing an advantage over traditional problem solving such as OFAT and matrix. Response Surface Method more focus on small, well-design experiments which solve problem on the process being studied (Haaland, 1989). Compare to OFAT and matrix design, this statistical design method also efficient and effective because it provides good coverage of the experimental space with as few measurements as possible (Adrion, et al., 1984). 21 Response Surface Method explores the relationships between several explanatory variables and one or more response variables (Box and Wilson, 1951). The main idea of RSM is to use a sequential experimental procedure to obtain an optimal response. Box and Wilson has suggested two of polynomial model to do this experimental procedure. A first model is a first-degree polynomial model and the second is a second-degree polynomial model. 2.7.2.1 Factorial / Fractional Factorial Design First step to obtain an optimal response in sequence experimental procedure is by using a first-degree polynomial model. This model actually is only an approximation but it is useful because it is easy to estimate and apply, even easy for those who have little knowledge in the process to use it. Easiest way to estimate a first-degree polynomial model is by using a factorial design or a fractional factorial in fully design. This model is sufficient to determine which of the explanatory variables have an impact on the response variables of interest. Once it is suspected that only significant explanatory variables are left, then this solution need more complicated design that is a second-degree polynomial model. Eventhough full factorial design or a full fractional factorial design are in same field but full factorial design actually allow the researcher to explore multiple factors simultaneously while fractional means that we do a fraction or a part of the full factorial design. Figure 2.8 illustrate the full factorial design. It’s enable researcher to do half-fraction, a quarter- fraction or an eighth-fraction (Box and Hunter, 1961a; Box and Hunter, 1961b; Tiao and George, 2000; Box et al., 1978). For example a half-fraction is to do half of the full factorial design, or (1/2)24 = 22 (1/2)16 = 8 runs to investigate four factors; (1/2)(25) = (1/2)32 = 16 runs to investigate five factors; and so on. For details, Figure 2.9 shows a graphical Factor 2 demonstration of why factorial design is better than OFAT experiments. Factor 3 Factor 1 Figure 2.8: Full Factorial Design with 2-level and 3 factors in 3D view Figure 2.9: One-factor-at-a-time versus factorial (Source: Czitrom, 1999) 23 A design with all possible high and low combinations of all the input factors in different experimental conditions is called a full factorial design with runs at nk. If an experiment with k factors, each set at two levels, it is called a two-level factorial design. It is also called a saturated design. k is the independent variables whose possible influence on a response variable is to be assessed are referred to as factors and n is refer to the number of the response level. The high and low levels are conveniently denoted by + and − or by + 1 and − 1. The factors can be continuous (pressure, temperature, concentration, etc.) or discrete (additive present, source of raw material, stirring used, etc.). The block with the 1's and -1's is called the Model Matrix or the Analysis Matrix. The table or tabular formed by the columns factor or response is called the Design Table or Design Matrix. For example, if there are k factors, each at 2 levels, a full factorial design has k 2 runs and if the factor used 3, a full factorial design has 23 runs which is equal to 8 runs (Table 2.3) and the Table Matrix may be expanding to observed the presence of interaction effect by two or more factors with few more columns as shown in Table 2.4.. The example of the run in the design always used the Standard Order which is not in randomized order and wrote from number 1 until the run’s number will be conducted. 24 Table 2.3: Typical pattern of a 2-level, 3-factor full factorial design Factor Run X Y Z 1 + + + 2 + + - 3 + - + 4 + - - 5 - + + 6 - + - 7 - - + 8 - - - Table 2.4: An example of fully expanded 23 factorial Factor Interaction effect or additional factor X Y Z XY YZ XZ XYZ Run (1) (2) (3) (4) (5) (6) (7) 1 + + + + + + + 2 + + - + - - - 3 + - + - - + - 4 + - - - + - + 5 - + + - + - - 6 - + - - - + + 7 - - + + - - + 8 - - - + + + - ( ) denotes column no. As shown by Table 2.5, when the number of factors is greater or 5, a full factorial design requires a large number of runs and is not very efficient. It is recommended that in that number of factors, applied a fractional factorial design or a Plackett-Burman design is a better choice for 5 or more factors. 25 Table 2.5: Guide to the Central Composite Rotatable Design and as for a 2k Full Factorial (Diamond, 2001) a No. of variables/no. of factor No. of Hadamard matrix trials/no. of runs No. of star trials No. of centre trials ψ 2 4 4 5 1.4142 3 8 6 6 1.6820 4 16 8 7 2.0000 5 16 10 6 2.0000 6 32 12 9 2.3780 7 64 14 14 2.8280 8 128 16 20 3.3640 a Runs with no. of variable greater than 4 are half-fraction ψ = F0.25 where F is the number of factorial points in the design; also equal to the ratio between the difference of star points value divided by the difference of factorial points value 2.7.2.2 Central Composite Rotatable Design To overcome and to estimates this advance design, a second-degree polynomial model is neede by implementd a central composite design or central composite rotatable design (CCRD) (Figure 2.10). However, this second model is still only an approximation but the approximation results is best than a first model. Moreover, the model absolutely can be used to optimize which it can maximize, minimize, or attain a specific target for a response or factor (Box and Wilson, 1951). 26 Figure 2.10: Central composite rotatable design in 3D view As shown in Figure 2.11, CCRD in two factors in nine different runs consists of four runs of full factorial, four axial points and one centre point. Other term for full factorial is cube and other term for axial points is star points (Czitrom, 1999; Aris, 2004). It has the advantage of incorporating information from a properly planned factorial experiment. The factorial portion and centre points may serve as a preliminary stage to fit a first order (linear) model and still provide evidence regarding the existence of second-order contribution or curvature. S F F S S C F F S Figure 2.11: The view of 3D CCRD with 2 factors (F: 4 runs of full factorial points or cube; S: 4 axial points or star points; C: 1 centre point). (Source: Czitrom, 1999) 27 A complete design enable the generation of a mathematical model which describe the process and may be used to determine the best setting of the process (Aris, 2004; Diamond, 2001). Table 2.5 in Section 2.7.2.1 provides a guideline for the CCRD as a function of the number of variables involved in the study (Diamond, 2001). The example of CCRD for the three factors such as suspended solid (SS), pH and temperature in effect of raw water quality on coagulant dosage is shown in Table 2.7. As mentioned in Section 2.7.2.1, the number of the factorial run is determined by 2k where k represents the number of variables involved. Any values may be assigned to Ψ and number of replicates of the centre point run. However, the values given in Table 2.6 provide only orthogonally blocked and rotatable designs which improve the quality of the model prediction (Diamond, 2001; Aris, 2004). 28 Table 2.6: Central composite rotatable design for three factors used in effect of raw water quality on coagulant dosage study SS (mg/L) pH Temperature (°C) SS (mg/L) Coded Termc Runb pH Temperature (°C) Uncoded Termd RSM101 -1 -1 -1 54.3 5.8 27.0 RSM 102 +1 -1 -1 54.3 5.8 33.0 RSM 103 -1 +1 -1 54.3 8.2 27.0 RSM 104 +1 +1 -1 54.3 8.2 33.0 RSM 105 -1 -1 +1 125.7 5.8 27.0 RSM 106 +1 -1 +1 125.7 5.8 33.0 RSM 107 -1 +1 +1 125.7 8.2 27.0 RSM 108 +1 +1 +1 125.7 8.2 33.0 RSM 109 -1.682 0 0 90.0 7.0 25.0 RSM 110 +1.682 0 0 90.0 7.0 35.0 RSM 111 0 -1.682 0 90.0 5.0 30.0 RSM 112 0 +1.682 0 90.0 9.0 30.0 RSM 113 0 0 -1.682 30.0 7.0 30.0 RSM 114 0 0 +1.682 150.0 7.0 30.0 RSM 115 0 0 0 90.0 7.0 30.0 RSM 116 0 0 0 90.0 7.0 30.0 RSM 117 0 0 0 90.0 7.0 30.0 RSM 118 0 0 0 90.0 7.0 30.0 RSM 119 0 0 0 90.0 7.0 30.0 RSM 120 0 0 0 90.0 7.0 30.0 RSM 101-108: factorial point run; RSM 109- RSM 114: star point run; RSM 115-120: centre point run b c (-1) and (+1) represent the low- and high- level values of the factorial points; (-1.682) and (+1.682) represent the minimum and maximum values of the star points; (0) represents the centre point value. d The actual value assigned to the variables translated from the coded value. CHAPTER 3 METHODOLOGY 3.1 Equipments and Materials Main equipments used in this study include jar test apparatus (Chemix Floctester Model CL6), turbidity meter (HI 93703 Portable Microprocessor Turbidity Meter) (HANNA Instruments, 1998) and pH meter (Thermo Orion Model 420A+). Chemicals used in this study include alum or aluminium sulphate (Al2(SO4)3.14.3H2O), hydrochloride acid (HCl), natrium hydroxide (NaOH) and the pH buffer for pH 7, pH 4 and pH 9. There are of analytical type and used without further treatment and chemicals were used as received. Hydrochloric acid of 0.1N concentration and NAOH of 0.1N concentration were used. Distilled water were prepared by (Bibbly Merit W4000) were used in the study. Raw water used in preliminary work were taken from middle part of Sungai UTM in the Universiti Teknologi Malaysia’s campus and kaolin were taken from UTM’s Soil and Geology Laboratory. 30 3.2 Experimental Procedure 3.2.1 Preliminary Study The purpose of preliminary study is to confirm the findings of previous study conducted by Jimbat (2006). In his study, Jimbat used synthetic water to compare between OFAT and RSM experimental approach in determining the optimum coagulant dosage and pH. While Jimbat study used synthetic water, this preliminary work used raw surface water collected from one of the river at Universiti Teknologi Malaysia. 3.2.1.1 One-Factor-At-A-Time In OFAT experimental approach, the experiments were carried out by varying one factor at a time with another factor being fixed. In this study, the approach followed the common procedures used in Jar Test study. The flow chart of the OFAT is shown in Figure 3.1 and the experimental design is shown in Table 3.1. A volume of 500 mL of raw surface water were filled in six beakers (at a time). The pH of the solutions was fixed at 7, while the alum solutions at six different dose (i.e. 20 mg/L, 40 mg/L, 60 mg/L, 80 mg/L, 100 mg/L and for 120 mg/L) were used and added into each beaker. 31 SET I Alum dosage varied from 20 mg/L - 120 mg/L; pH set at 7. SET II pH varied from 4.5 – 7.0; Alum set at best dosage from SET I. Figure 3.1: Flow chart of OFAT experiment Table 3.1: Experimental run for OFAT approach Run pH Alum Dose (mg/L) OFAT11 7.0 20 OFAT12 7.0 40 OFAT13 7.0 60 OFAT14 7.0 80 OFAT15 7.0 100 OFAT16 7.0 120 OFAT27 4.5 40 OFAT28 5.0 40 OFAT29 5.5 40 OFAT210 6.0 40 OFAT211 6.5 40 OFAT212 7.5 40 * OFAT11 – OFAT 212; refer to beaker number 1 in first run of the jar test and so on. The water was initially rapid mixed at approximately 80 rpm for one minute and then followed by slow mixing at 30 rpm for approximatly 15 minutes. After 15 32 minutes of slow mixing, the mixer was turn-off and the floc was allowed to settle for 30 minutes. Samples from the supernatant for each beaker were taken and immediately analyzed for turbidity. The best dosage was determined based on the lowest turbidity achieved. The second set of experiments was conducted using the best alum dosage of the first jar test run. The pH of the water in each beaker was set at 4.5, 5.0, 5.5, 6.0, 6.5 and 7.0. The same procedures as explained in previous paragraph were repeated and the best pH was determined best on the lowest turbidity achieved. 3.2.1.2 Response Surface Method As mentioned earlier, RSM consists of a collection of experimental runs made at predetermined settings of the experimental condition. In this approach, pH was set at 4.5, 5, 6.25, 7.5 and 8 while the alum dosage was set at 20, 35, 70, 105 and 120 mg/L. The combinations of these settings are shown in Table 3.2. Except for the combination of alum dosage and pH setting, the rest of the experimental procedures followed those as described in previous section of OFAT. 33 Table 3.2: Experimental run in RSM approach Run Alum Dose (mg/L) pH RSM11 35 5.00 RSM 12 105 5.00 RSM 13 35 7.50 RSM 14 105 7.50 RSM 15 20 6.25 RSM 16 120 6.25 RSM 27 70 4.50 RSM 28 70 8.00 RSM 29 70 6.25 RSM 210 70 6.25 RSM 211 70 6.25 RSM 212 70 6.25 RSM213 70 6.25 * RSM11 - RSM27; refer to beaker number 1 in first run of the jar test and so on. 3.2.2 Raw Water Quality - Alum Dosage and pH Relationship Response Surface Method approach was used to explore the relationship between water quality variables and optimum dosage and also pH. Three raw water quality variables were selected in this study. They are suspended solid (SS), temperature and pH. Synthetic water using distilled water was used with kaolin added as the source of SS. The characteristics of the water in term of SS, temperature and pH were predetermined accordingly prior to the experiments and are shown in Table 3.3. For each water characteristic, jar test procedures as described in Section 3.2.1 (RSM approach) were conducted to determine the optimum alum dosage and pH. 34 The range of alum dosage used was depended on the initial SS of the water. For initial SS concentration of 54.3 mg/L in the initial four sets, alum dosages between 20 mg/L to 120 mg/L were used while for SS concentration of 30.0 mg/L, 90.0 mg/L, 125.7 mg/L and 150.0 mg/L in the following sets, alum dosages between 40 mg/L to 200 mg/L were used. The flow chart of the experiments is given in Figure 3.2. The details of the experimental design with respect to the alum dosage and pH used in the study are given in Tables 3.3, 3.4 and 3.5. Water Characteristics (Set 1 to 20) Jar Test procedures according to RSM approach Repeat from each set of water characteristics Determination of optimum dosage and pH Result recorded Figure 3.2: Flow chart of experimental design 35 Table 3.3: Setting of water characteristics used in the study Set Temperature (°C) pH SS (mg/L) 1 27.0 5.8 54.3 2 33.0 5.8 54.3 3 27.0 8.2 54.3 4 33.0 8.2 54.3 5 27.0 5.8 125.7 6 33.0 5.8 125.7 7 27.0 8.2 125.7 8 33.0 8.2 125.7 9 25.0 7.0 90.0 10 35.0 7.0 90.0 11 30.0 5.0 90.0 12 30.0 9.0 90.0 13 30.0 7.0 30.0 14 30.0 7.0 150.0 15 30.0 7.0 90.0 16 30.0 7.0 90.0 17 30.0 7.0 90.0 18 30.0 7.0 90.0 19 30.0 7.0 90.0 20 30.0 7.0 90.0 36 Table 3.4: Design table for range 20 mg/L to 120 mg/L of alum dosage Run No. Alum Dose (mg/L ) pH RSM1 34.6 5.00 RSM2 105.4 3.60 RSM3 34.6 17.00 RSM4 105.4 6.09 RSM5 20.0 6.00 RSM6 120.0 3.89 RSM7 70.0 4.24 RSM8 70.0 4.38 RSM9 70.0 4.64 RSM10 70.0 5.19 RSM11 70.0 4.17 RSM12 70.0 4.89 RSM13 70.0 4.72 Table 3.5: Design table for range 40 mg/L to 200 mg/L of alum dosage Run No. Alum Dose (mg/L ) pH 1 34.6 5.00 2 105.4 3.60 3 34.6 17.00 4 105.4 6.09 5 20.0 6.00 6 120.0 3.89 7 70.0 4.24 8 70.0 4.38 9 70.0 4.64 10 70.0 5.19 11 70.0 4.17 12 70.0 4.89 13 70.0 4.72 CHAPTER 4 RESULTS AND DISCUSSIONS 4.1 Preliminary Study At pH 7, the lowest turbidity value was 19 FTU at 20 mg/L to 60 mg/L of alum dose which represent 38.7% removal. When the alum dose was set at 40 mg/L, the lowest turbidity value achieved was 6 FTU at pH 7.5 (80.6%). Figures 4.1 and 4.2 illustrate the effect of pH and alum dosage on the removal of turbidity. The highest turbidity removal (i.e. 80.6%) was achieved at pH of 7.5 and 40 mg/L of alum dosage. However, since the OFAT approach did not consider all the space of experiment, these optimum pH and dosage are still arguable. 38 Table 4.1: Results of the OFAT experiment on the turbidity removal (Initial turbidity = 31 FTU) pH Alum Dose (mg/L) Final Turbidity (FTU) OFAT11 7.0 20 19 OFAT12 7.0 40 19 OFAT13 7.0 60 19 OFAT14 7.0 80 20 OFAT15 7.0 100 21 OFAT16 7.0 120 20 OFAT27 4.5 40 14 OFAT28 5.0 40 12 OFAT29 5.5 40 13 OFAT210 6.0 40 14 OFAT211 6.5 40 14 OFAT212 7.5 40 6 Run * OFAT11/OFAT27; refer to beaker number 1 or 7 in first or second run of jar test. ** 1 FTU = 1 NTU 40.0 38.7 39.0 38.7 38.7 % Removal 38.0 37.0 36.0 35.5 35.0 35.5 34.0 33.0 32.3 32.0 31.0 0 20 40 60 80 100 120 Alum dose (mg/L) Figure 4.1: Effect of alum dose on final turbidity (pH 7) 140 39 90.0 80.6 % Removal 80.0 70.0 61.3 60.0 58.1 54.8 54.8 54.8 50.0 40.0 30.0 4 4.5 5 5.5 6 6.5 7 7.5 8 pH Figure 4.2: Effect of pH on final turbidity (Alum dose = 40 mg/L) 4.2 Response Surface Method The results of the RSM are shown in Table 4.2. In this method the determination of optimum pH and its dose was directly determined by the design of experimental using MINITABTM statistical software. With the same ranges of alum dose (20 mg/L-120 mg/L), the turbidity value ranges achieved were 6 FTU to 16 FTU. This range is smaller than the ranges obtained in OFAT method possibly due to the different dosages and pH used in RSM approach. The lowest value for final turbidity in this experiment at RSM approach was 6 NTU, similar to the value achieved in the OFAT experiment. However, to achieve the 6 FTU, the alum dose applied was 35 mg/L at pH 7.5. Eventhough the pH value is exactly the same as in OFAT approached, the dose of alum in this method is lesser (as compared to 40 mg/L by OFAT), thus shows a better achievement towards the economic of coagulant dose usage. The results confirm the findings of Jimbat (2006) 40 that RSM approach is a better alternative as compared to OFAT in determining the optimum coagulant dosage and pH. Table 4.2: Turbidity results from the RSM experiment in preliminary works (Initial turbidity = 31 FTU) pH Alum Dose (mg/L) Final Turbidity (FTU) RSM11 5.0 35 15 RSM 12 5.0 105 13 RSM 13 7.5 35 6 RSM 14 7.5 105 16 RSM 15 6.25 20 12 RSM 16 6.25 120 13 RSM 27 4.5 70 9 RSM 28 8.0 70 12 RSM 29 6.25 70 12 RSM 210 6.25 70 12 RSM 211 6.25 70 12 RSM 212 6.25 70 12 Run * RSM11/RSM27; refer to beaker number 1 in first run of jar / beaker number 7 in second run of jar test ** 1 FTU = 1 NTU 4.3 Water Quality - Optimum Dose And pH Relationship As mentioned in Section 3.2.2, 20 sets of experiment using predetermined values of initial pH, SS and temperature were carried out to determine the relationship between the selected water quality parameters and the optimum pH and alum dosage relationship. For each set of water quality, 12 experimental runs were carried out to identify the optimum pH and dosage. Two types of approaches were 41 used to identify these optimum values namely Response Surface analysis and lowest FTU approach. 4.3.1 Response Surface Analysis The detail results from the response surface analysis for each set of experiments are given in Appendix A. The R-squared value and the p-value of lackof-fit from the response surface analysis are summarized in Table 4.3. Only sets of experiment with R-squared value more than 80% (i.e. Set 2, 8, 9, 10, 15, 17 and 18). The set which are more than 80% of its R-squared value considered to be significant and acceptable results for the set. However, in order for the results to be used in determining the optimum dosage and pH, the p-value of the terms need to be less than 0.1 for 90% confidence level. As can be seen in Table 4.3, not all experimental runs fulfill these requirements. Hence, from the overall analysis, the conclusion that can be made is that these RSM models are unacceptable and response surface analysis approach fails to identify the optimum pH alum dose from each set of experiment. So the other option of analysis based on the lowest FTU obtained in the experiment was made to determined optimum pH and alum dose. 42 Table 4.3: The summary of response surface regression for 20 sets of experiment 4.3.2 Results for Set R-square value (%) Lack-of-fit RSM1 59.6 0.000 RSM2 86.8 0.215 RSM3 64.6 0.004 RSM4 77.4 0.007 RSM5 57.7 0.071 RSM6 60.1 0.010 RSM7 62.1 0.000 RSM8 84.0 0.000 RSM9 86.1 0.001 RSM10 90.4 0.000 RSM11 29.5 0.000 RSM12 36.6 0.246 RSM13 4.0 0.000 RSM14 76.9 0.051 RSM15 90.7 0.002 RSM16 50.9 0.043 RSM17 88.0 0.010 RSM18 88.8 0.000 RSM19 59.7 0.045 RSM20 59.1 0.079 Lowest FTU Approach In this approach, the determination of the optimum pH and alum dosage was based on the lowest FTU obtained from the experiments. This approach is rather straight-forward. The results of this analysis are shown in Table 4.4. 43 From the table, it can be observed that the optimum pH is most monopolies by pH 5.0, followed by pH 6.3 and 4.5, and pH 7.5. The optimum dose of alum was monopolied by dose of 120 mg/L, followed by 176.6 mg/L, 35 mg/L, 63.4 mg/L and both 105 mg/L and 200 mg/L. Table 4.4: Optimum pH and dosage for the experiments based on Lowest FTU approach Water Quality Independent Variable Depend, Response Set Initial Temperature (°C) Initial pH Initial SS (mg/L) Optimum pH Optimum Dose (mg/L) RSM1 27.0 5.8 54.3 5.0 105.0 RSM2 33.0 5.8 54.3 5.0 35.0 RSM3 27.0 8.2 54.3 5.0 35.0 RSM4 33.0 8.2 54.3 5.0 35.0 RSM5 27.0 5.8 125.7 4.5 120.0 RSM6 33.0 5.8 125.7 7.5 176.6 RSM7 27.0 8.2 125.7 5.0 63.4 RSM8 33.0 8.2 125.7 5.0 63.4 RSM9 25.0 7.0 90.0 4.5 120.0 RSM10 35.0 7.0 90.0 4.5 120.0 RSM11 30.0 5.0 90.0 7.5 176.6 RSM12 30.0 9.0 90.0 5.0 176.6 RSM13 30.0 7.0 30.0 5.0 176.6 RSM14 30.0 7.0 150.0 4.5 120.0 RSM15 30.0 7.0 90.0 6.3 120.0 RSM16 30.0 7.0 90.0 5.0 176.6 RSM17 30.0 7.0 90.0 6.3 200.0 RSM18 30.0 7.0 90.0 5.0 176.6 RSM19 30.0 7.0 90.0 6.3 120.0 RSM20 30.0 7.0 90.0 6.3 120.0 44 4.3.2.1 Factorial Analysis The factorial analysis was conducted on the Hadamard matrix and centre points runs using MINITABTM statistical software. The objective in conducting factorial analysis on the results is to determine the significance of initial temperature, pH and SS on the optimum pH and alum dosage. The analysis was carried out at α = 0.1 (90% of significance level) and α = 0.2 (80% of significance level. The detail results of the analysis are given in Appendix B and summarized in Table 4.5. The significant and the effect of the factors or responses are illustrated in the Pareto chart (Figure 4.3). Table 4.5: ANOVA for turbidity removal at optimum dose and optimum pH Optimum Dose Optimum pH P-value Significancea P-value Significancea Significanceb Temperature 0.901 No 0.175 No Yes pH 0.067 Yes 0.340 No No SS 0.092 Yes 0.340 No No Temperature x pH 0.901 No 0.175 No Yes Temperature x SS 0.272 No 0.175 No Yes pH x SS 0.376 No 0.340 No No 0.272 No 0.175 No Yes Effect Main Two-way interaction Three-way interaction Temperature x pH x SS a significant at α = 0.1 significant at α = 0.2 b 45 (a) (b) Figure 4.3: Pareto chart for (a) Optimum Dose and (b) Optimum pH based on final turbidity (A: Temperature; B: pH; C: SS) With respect to the experimental conditions used in this study, at significance level of 90% of the p-value indicates that only initial pH and SS have significant effect on the optimum alum dose (< 0.1). However at this significant care level, none of the factors have significant role on the optimum pH. Temperature is considered a significant factor to optimum pH only at confidence level of 80%. Initial pH and SS were found not to have significant effect on optimum pH within the range of the study. From Figure 4.4, it seems that the shift of temperature from 27°C to 33°C decreased to optimum dose but increased the optimum pH. The optimum dose and pH decreased when the initial pH was shift from 5.8 to 8.2 but both of the responses were instantly increased when initial SS was shift from 54.3 mg/L to 125.5 mg/L. 46 Figure 4.4: Main effect plot for water quality parameters at (a) Optimum Dose (b) Optimum pH Interaction effects were found to be significant only for optimum pH at 80% confidence level. Significant interaction effects at p-value < 0.2 were for Temp x pH, Temp x SS and the three way interaction between Temp x pH and SS. Figure 4.5 and 4.6 illustrate the interaction effect between the factors. Mean Optimum Dose Figure 4.5: Interaction effect plot for water quality parameters at optimum dose 47 Mean Optimum pH Figure 4.6: Interaction effect plot for water quality parameters at optimum pH 4.3.2.2 Response Surface Analysis A response surface analysis was conducted to quantitatively characterize the behavior of the effect and to statistically model the relationship between both the selected water quality parameters and the optimum pH and alum dose. The analysis was initially carried out using the full quadratic terms. The details results are given in Appendices C and D. The p-values are summarized as in Table 4.6 for optimum dose and Table 4.7 for optimum pH. From the observation on optimum dose, it found that only Temperature x Temperature were significant. As the p-value is 0.118 in full quadratic terms, thus, this indirectly makes main effect on temperature also significant although the p-value was insignificant (> 0.2). Although switch to only Linear + Squared terms, the pvalue at an effect of temperature interaction still obtained the best significant value (< 0.1). 48 Table 4.6: The summary of the response surface analysis for the water quality effect on optimum dose Full quadratic terms Linear + Squared terms P-valuea Term Temperature 0.951 0.946 pH 0.285 0.241 SS 0.590 0.555 Temperature x Temperature 0.118 0.085 pH x pH 0.704 0.678 SS x SS 0.318 0.273 Temperature x pH 0.936 - Temperature x SS 0.454 - pH x SS 0.553 - a 0.01 – 0.04: Highly significant; 0.05 – 0.1: significant; 0.1 – 0.2: less significant; > 0.2: insignificant (Vecchio, 1997) Table 4.7: The summary of the response surface analysis for the water quality effect on optimum pH Full quadratic terms Linear + Squared terms P-valuea Term Temperature 0.288 0.319 pH 0.043 0.052 SS 0.674 0.695 Temperature x Temperature 0.037 0.045 pH x pH 0.415 0.446 SS x SS 0.082 0.098 Temperature x pH 0.173 - Temperature x SS 0.173 - pH x SS 0.351 - a 0.01 – 0.04: Highly significant; 0.05 – 0.1: significant; 0.1 – 0.2: less significant; > 0.2: insignificant (Vecchio, 1997) At full quadratic, it was found that pH (p-value = 0.043), Temperature x Temperature (p--value = 0.037) and SS x SS (p-value = 0.082) were significant to 49 optimum pH. Since two-way interaction of temperature and from SS were respectively significant, this has resulted the main effect of temperature (p-value = 0.288) and SS (p-value = 0.674) also significant. Although in e Temperature x pH and Temperature x SS are less significant but it still can be considered. When the terms switch to Linear + Squared terms only, clearly seen that p-value of all significant effect obtained the reducing value. To improve the quality of the model, another analysis was carried out with the insignificant terms omitted from the model. The detail results of this analysis are given in Appendix D. If the analysis eliminate three additional terms from the model, the R-square term for optimum dose were found < 50% while lack-of-fit and an adjustable R-square was increase up to 0.134 and 1.6%, respectively. However, these eliminating terms from the model found that R-square, adjustable R-square and lack-of-fit in optimum pH were decrease. It was better to optimum pH without any elimination terms which it brought to 69.1% in R-square, 41.2 in adjustable R-square and 0.384 lack-of-fit, respectively. From these analyses, the acceptable statistical model for this system based on the experimental condition of the study for water quality parameters is only at optimum pH and the equation for the model is: Optimum pH = -53.5591 + 3.59795 Temperat + 1.72582 pH - 0.0103599 SS -.0516633 Temperat2 + 0.114604 pH2 - .000289329 SS2 -0.106066 Temperat*pH + 0.00353553 Temperat*SS -0.00589256 pH*SS Where; Temperat = temperature in °C; pH = pH of the water; SS = suspended solid in mg/L. The response surface and contour plots of the modal are given in Figures 4.7 through 4.12. 50 Hold values: SS (mg/L: 90.0) 9 8 pH 5 7 6 6 5 7 25 26 27 (a) 28 29 30 31 32 33 34 35 T emperature Hold values: SS (mg/L: 90.0) 7 6 Optimum pH 5 9 4 8 7 25 (b) Temperature 6 30 pH 5 35 Figure 4.7: (a) Contour and (b) response surface plots representing relationship between pH and temperature at optimum pH 51 Hold values: pH: 7.0 150 3.7 4.2 4.7 5.2 100 SS (mg/L) 5.7 50 25 26 27 28 29 30 31 32 33 34 35 T emperature (a) Hold values: pH: 7.0 6 5 Optimum pH 4 150 3 100 25 (b) Temperature SS 50 30 35 Figure 4.8: (a) Contour and (b) response surface plots representing relationship between temperature and SS at optimum pH 52 Hold values: T emperat: 30.0 150 5.0 100 SS (mg/L) 6.0 6.5 7.0 5.5 50 5 6 7 8 9 pH (a) Hold values: Temperat: 30.0 7 6 Optimum pH 5 150 4 100 5 (b) 6 pH 7 SS S 50 8 9 Figure 4.9: (a) Contour and (b) response surface plots representing relationship between pH and SS at optimum pH 53 Hold values: SS (mg/L: 90.0) 9 65 90 8 115 pH 140 7 6 165 5 25 26 27 28 29 30 31 32 33 34 35 T emperature (a) Hold values: SS (mg/L: 90.0 150 100 Optimum Dose (mg/L) 50 9 8 7 25 (b) Temperature 6 30 pH 5 35 Figure 4.10: (a) Contour and (b) response surface plots representing relationship between pH and temperature at optimum dose 54 Hold values: pH: 7.0 150 100 SS (mg/L) 145 120 95 50 70 45 25 26 27 28 (a) 29 30 31 32 33 34 35 T emperature Hold values: pH: 7.0 150 100 Optimum Dose (mg/L) 50 150 0 100 25 (b) Temperature SS S 50 30 35 Figure 4.11: (a) Contour and (b) response surface plots representing relationship between temperature and SS at optimum dose 55 Hold values: Temperat: 30.0 150 150 SS (mg/L) 100 130 110 90 50 70 5 6 7 (a) 8 9 pH Hold values: Temperat: 30.0 150 Optimum Dose (mg/L) 100 150 50 100 5 6 pH 7 (b) SS S 50 8 9 Figure 4.12: (a) Contour and (b) response surface plots representing relationship between pH and SS at optimum dose CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions From the study, the conclusions that could be made are as followings: a) Response Surface Method approach provides a better results in jar test procedures compared to OFAT. While the initial turbidity of the two approaches is the same, the optimum dosage obtained in RSM is lower than the one obtained by OFAT and hence, contribute to the economy of the process. b) At 90% confidence level, only pH and SS have significant effect on the optimum alum dose. No interaction effect was observed. Effects of water quality parameters on optimum pH were observed only at 80% confidence level. At this level, temperature, temperature x pH, temperature x SS and three-way interactions were significant on optimum pH 57 c) Relationship between initial temperature, pH and SS could only be developing for optimum pH. The R-square and p-value for lack-of-fit for this relationship were 69.1% and 0.384, respectively. 5.2 Future Study Recommendations Based on the findings of the study, the followings are recommended for further study. a) To use others sources of suspended solid such as laterite, bentonite clay and peat. The usage of these suspended solid must in powder form to ease the experimental work. b) To include other water quality parameters in developing the relationship. This may include alkalinity, conductivity and dissolved oxygen content. c) To determine the relationship based on secondary data collected from jar test exercise at water treatment plant. REFERENCES Adhin, A., and Rebhun, M. (1974). High-rate contact flocculation-filtration with cationic polyelectrolytes. Journal AWWA. 66(2): 109-117. Adrion, R. F., G. R. Siebert, C. J. Weck, D. Yen and A. R. Manson (1984). Optimization of in vivo monoclonalantibody production using computer-as-sisted experimental design. Proceedings of the First Carolina Biomedical Engineering Conference. Research Triangle Park, North Carolina: North Carolina Biotechnology Center. 125-144. Al-Laya and Middlebrooks (1974). Algae removal by chemical coagulation. Water and Sewage Works. 121(9): 76-80. Amirtharajah, A, and Mills, K. M. (1982). Rapid-mix design for mechanisms of alum coagulation. Journal of American Water Works Association. 74(4): 210216. AWWA and ASCE (1990). Water Treatment Plant Design. Second Edition. New York, NY: McGraw-Hill, Inc. AWWA and ASCE. (1998). Water Treatment Plant Design. Third Edition. New York, NY: AWWA and ASCE, McGraw-Hill, Inc,. Aris, Azmi (2004). Fenton’s Reaction System for the Treatment of Textile Dyeing Wastewater. University of Manchester. Institute of Science and Technology: PhD Thesis. 59 Bagwell, T., Henry, H.B. and Kenneth, M.B. (2001). Handbook of public water systems. 2nd Edition. HDR Engineering Inc., New York. Bell-Ajy, K., Abbaszadegan, M., Ibraham, E., Verges, D., and LeChevallier, M. (2000). Conventional and Optimized coagulation for NOM Removal. Journal of the American Water Works Association. 92(10): 44-58. Box, G. E. P. and Wilson, K.B. (1951). On the Experimental Attainment of Optimum Conditions (with discussion). Journal of the Royal Statistical Society. Series B 13(1):1-45. Box, G. E. P. and Hunter, J. S. (1961a). The 2k−p Fractional Factorial Designs. Part I. Technometrics. 3(3): 311–351. Box, G. E. P. and Hunter, J. S. (1961b). The 2k−p Fractional Factorial Designs. Part II. Technometrics. 3(4), 449–458. Box, G. E. P. (1965). Experimental Strategy. Wisconsin Tech. Report 111. Department of Statistics,University of Wisconsin–Madison. Box, G. E. P., W. G. Hunter, and J. S. Hunter (1978). Statistics for Experimenters: An Introduction to Design,Data Analysis, and Model Building. New York: Wiley Interscience. Budd, G. C., Alan F. Hess, Holly Shorney-Darby, Jeff J Neemann et al. (2004) Coagulation Applications for New Treatment Goals. American Water Works Association Journal. 96 (2): 102. Czitrom, Veronica (1999). Teacher’s Corner: One-Factor-at-a-Time Versus Designed Experiments. The American Statistician. 53(2): 126-131. Dempsey, B. A., Sheu, H., Tanzeer Ahmed, T. M., and Mentink, J. (1985). Polualuminum Chloride and Alum Coagulation of Clay-Fulvic Acid Suspension. Journal of American Water Works Association. 77(3): 74-80. 60 Dempsey, B. A. (1994). Production and Utilization of Polyaluminum Sulphate. AWWA Research Foundation. 1: 71. Dennett, K. E. et. al. (1996). Coagulation: It’s Effect on Organic Matter. Journal American Water Works Association. 129-141. Diamond, W. J. (2001) Practical Experimental Designs. John Wiley and Sons, Inc. Edzwald, J. K., and Van Benschoten, J. E. (1990). Aluminum Coagulation of Natural Organic Matter. In: Hahn and Klute. Chemical Water and Wastewater Treatment. 341-359. Edzwald, J. k., Bunker Jr, D. Q. et. al. (1994). Dissolve Air Flotation: Pretreatment and Comparisons to Sedimentation. In: Hahn and Klute. Chemical Water and Wastewater Treatment III. Berlin: Springer-Verlag. 5-17. Edzwald, J. K., and Tobiason, J. E. (1999). Enhanced Coagulation: USA Requirements and a Broader View. Removal of Humic Substances from Water. Trondheim, Norway: IAWQ/IWSA Joint Specialist Group on Particle Separation. Eikebrook, B. (1990). Removal of Humic Substances by Coagulation. In: H. H. Hahn and R. Klute. Chemical Water and Wastewater Treatment. Berlin: Springer-Verlag. 173-187. EPA (1999). Cure Electrocoagulation Technology. Innovative Technology Evaluation Report. 504: 5. Haaland, Perry. D. (1989). Experimental Design in Biotechnology. New York: MARCEL DEKKER, Inc. Hammer, M. J., and Hammer Jr, M. J (2005). Water and Wastewater Technology. 5th Edition. Singapore: PEARSON-Prentice Hall. 61 Hendrix, C. D. (1979). What Every Technologist Should Know About Experimental Design. Chemtech. 167-174. Ho, L. (2005). Effect of NOM, turbidity and floc size on the PAC adsorption of MIB during alum coagulation. Water Research. 39: 3670. Hudson, H.E. Jr. (1981). Water Clarification Processes, Practical Design and Evaluation. Van Nostrand Reinhold Environmental Engineering Series, Litton Educational Publishing, Inc. Hooge, P.A. (2000). Evaluating adsorption clarification at cold water temperature for drinking water treatment. University of Saskatchewan, Saskatoon: M.Sc. Thesis. Jimbat, Albraham Enggong (2006). Jar Test: One-Factor-At-A-Time Versus Response Surface Design. Universiti Teknologi Malaysia: Degree Project. Kang, L.S., and Cleasby, J.L. (1995). Temperature effects on flocculation kinetics using Fe(III) coagulant. Journal of Environmental Engineering. 121(12). Kawumara, S. (1991). Integrated Design of Water Treatment Facilities. John Wiley & Sons, Inc. McGhee, T.J. (1991). Water Resources and Environmental Engineering. Sixth Edition. McGraw-Hill, New York. Odegard, H., Fetting J. et. al. (1990). Coagulation with Prepolymerized Metal Salts. In: Hahn and Klute. Chemical Water and Wastewater Treatment III. Berlin: Springer-Verlag. 189-220. Peavey, H. S., Donald, R. Rowe, and George Tchobanoglous (1985). Environmental engineering. International Edition. Singapore: McGraw-Hill. 62 Pernitsky, D. J. (2001). Drinking Water Coagulation with Polyaluminum coagulants – Mechanism and Selection Guidelines. University of Massachusetts Amherst: Ph.D. Dissertation. Pernitsky, D. J. (2003). Coagulation 101. Associated Engineering, Calvary, Alberta: Ph. D Thesis. Pushkin, D. O. (2004). Emergent Self-Similarity of Cluster Coagulation. University of Illinois, Urbana-Champaign: Ph. D. Thesis. Randtke, S. J. (1988). Organic Contaminant Removal by Coagulation and Related Process Combination. Journal of American Water Works Association. 80(5): 4056. Reed, B. E., Mark, P., Matsumoto, Roger Viadero Jr., Robert L. Segar Jr., Ronald Vaughan and David Masciola (1999). Physicochemical Process. Water Environment Research. 71(5): ProQuest Science Journals, 584. Reynolds, T.D., and Richards, P.A. (1996). Unit operations and processes in environmental engineering. 2nd Edition. PWS Publishing Company. Sawyer, Clair N., and Perry, L. McCarty (1978). Chemistry for Environmental Engineering. Third edition. New York: McGraw-Hill. Schultz, C.R., Singer, P.C., Gandley, R., and Nix, J.E. Evaluating buoyant coarse media flocculation. Journal AWWA. 1984. 76(8):51-63. Shea, T.G., Gates, W.E., and Argaman, Y.A. (1971). Experimental evaluation of operating variables in contact flocculation. Journal AWWA. 63(1): 41-48. Sinsaubaugh, R. L., R. C. Hoehn et. al. (1986). Precursor Size and Organic Halide Formation Rates in Raw and Coagulated Surface Waters. Journal of Environmental Engineering. 112(1): 139 – 153. 63 Snoeyink, V.L., and Jenkins, D. (1980). Water chemistry. New York: John Wiley and Sons, Inc. Tang, H. X., Luan, Z. K., Wang, D. S., and Gao, B. Y. (1998). Composite Inorganic Polymer Flocculants. In: H. H. Hahn, E. Hoffman and H. Odegaard. Chemical Water and Wastewater Treatment. New York: Springer. 25-34. Tebbuti, T.H.Y. (1973). Principles of Water Quality Control. Pergamon Press. 96104. Tiao and George et al. (2000). Box on Quality and Discovery with Design, Control, and Robustness. New York: John Wiley & Sons. Jin, Yan. (2005). Use of a High Resolution Photographic Technique for Studying Coagulation / Flocculation In Water Treatment. University of Saskatchewan, Saskatoon: Master Degree Thesis. APPENDIX A (DESIGN TABLE FOR 20 SET) Table A-1: Design table of RSM for Set 1 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM11 RSM12 RSM13 RSM14 RSM15 RSM16 27.0 RSM17 RSM18 RSM19 RSM110 RSM111 RSM112 RSM113 Best FTU = 3.60 5.8 54.3 Alum Dose (mg/L) 35 105 35 105 20 120 70 70 70 70 70 70 70 Variable Alum Amount (mL) 17.5 17.5 17.5 17.5 10.0 20.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 Alum Dose (mg/L) 35 105 35 105 20 120 70 70 70 70 70 70 70 Variable Alum Amount (mL) 17.5 17.5 17.5 17.5 10.0 20.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 Response pH Turbidity (FTU) 5.00 5.00 7.50 7.50 6.25 6.25 4.50 8.00 6.25 6.25 6.25 6.25 6.25 5.00 3.60 17.00 6.09 6.00 3.86 4.24 4.38 4.64 5.19 4.17 4.89 4.72 Table A-2: Design table of RSM for Set 2 Run Initial Water characteristic Suspended Temperature Solid, SS pH (°C) (mg/L) RSM21 RSM22 RSM23 RSM24 RSM25 RSM26 33.0 RSM27 RSM28 RSM29 RSM210 RSM211 RSM212 RSM213 Best FTU = 1.75 5.8 54.3 Response pH Turbidity (FTU) 5.00 5.00 7.50 7.50 6.25 6.25 4.50 8.00 6.25 6.25 6.25 6.25 6.25 1.75 2.67 4.93 4.26 2.32 3.96 2.31 4.48 3.18 3.57 3.96 4.12 3.68 65 Table A-3: Design table of RSM for Set 3 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM31 RSM32 RSM33 RSM34 RSM35 RSM36 27.0 RSM37 RSM38 RSM39 RSM310 RSM311 RSM312 RSM313 Best FTU =1.81 8.2 54.3 Alum Dose (mg/L) 35 105 35 105 20 120 70 70 70 70 70 70 70 Variable Alum Amount (mL) 17.5 17.5 17.5 17.5 10.0 20.0 35.0 35.0 35.0 35.0 35.0 35.0 35.0 Response pH Turbidity (FTU) 5.00 5.00 7.50 7.50 6.25 6.25 4.50 8.00 6.25 6.25 6.25 6.25 6.25 1.81 4.27 2.05 4.02 2.83 3.62 5.22 7.16 4.76 4.03 4.71 4.79 4.57 Table A-4: Design table of RSM for Set 4 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM41 RSM42 RSM43 RSM44 RSM45 RSM46 33.0 RSM47 RSM48 RSM49 RSM410 RSM411 RSM412 RSM413 Best FTU = 3.66 8.2 54.3 Variable Alum Alum Dose Amount (mg/L) (mL) 35 17.5 105 17.5 35 17.5 105 17.5 20 10.0 120 20.0 70 35.0 70 35.0 70 35.0 70 35.0 70 35.0 70 35.0 70 35.0 Response pH Turbidity (FTU) 5.00 5.00 7.50 7.50 6.25 6.25 4.50 8.00 6.25 6.25 6.25 6.25 6.25 3.66 3.86 5.45 4.84 6.47 4.28 4.36 6.45 6.05 6.36 6.04 5.76 6.05 66 Table A-5: Design table of RSM for Set 5 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM51 RSM52 RSM53 RSM54 RSM55 RSM56 27.0 RSM57 RSM58 RSM59 RSM510 RSM511 RSM512 RSM513 Best FTU = 7.59 5.8 125.7 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 7.75 8.78 8.41 10.70 8.88 10.30 7.59 18.03 8.70 8.17 11.40 10.90 9.79 Table A-6: Design table of RSM for Set 6 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM61 RSM62 RSM63 RSM64 RSM65 RSM66 33.0 RSM67 RSM68 RSM69 RSM610 RSM611 RSM612 RSM613 Best FTU = 1.89 5.8 125.7 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 8.80 6.72 9.26 1.89 9.03 6.55 8.44 11.10 9.28 9.39 7.75 8.58 8.75 67 Table A-7: Design table of RSM for Set 7 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM71 RSM72 RSM73 RSM74 RSM75 RSM76 27.0 RSM77 RSM78 RSM79 RSM710 RSM711 RSM712 RSM713 Best FTU = 4.39 8.2 125.7 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 4.39 8.89 9.80 8.85 10.63 8.92 8.95 16.80 9.00 9.08 9.45 9.89 9.35 Table A-8: Design table of RSM for Set 8 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM81 RSM82 RSM83 RSM84 RSM85 RSM86 33.0 RSM87 RSM88 RSM89 RSM810 RSM811 RSM812 RSM813 Best FTU = 1.81 8.2 125.7 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 1.81 8.28 8.66 14.60 7.56 12.20 5.76 7.97 7.75 7.19 7.53 7.31 7.44 68 Table A-9: Design table of RSM for Set 9 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM91 RSM92 RSM93 RSM94 RSM95 RSM96 25.0 RSM97 RSM98 RSM99 RSM910 RSM911 RSM912 RSM913 Best FTU = 2.38 7.0 90.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 For Jar Test Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 6.75 6.09 6.62 10.97 5.87 7.58 2.38 5.26 3.02 2.91 2.80 2.39 2.78 Table A-10: Design table of RSM for Set 10 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM101 RSM102 RSM103 RSM104 RSM105 RSM106 35 RSM107 RSM108 RSM109 RSM1010 RSM1011 RSM1012 RSM1013 Best FTU = 5.37 7.0 90.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 15.53 7.83 15.54 11.07 15.45 8.71 5.37 11.46 7.76 7.19 7.50 7.50 7.48 69 Table A-11: Design table of RSM for Set 11 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM111 RSM112 RSM113 RSM114 RSM115 RSM116 30.0 RSM117 RSM118 RSM119 RSM1110 RSM1111 RSM1112 RSM1113 Best FTU = 6.43 5.0 90.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 7.46 7.17 7.22 6.43 12.14 11.30 8.00 11.75 7.13 7.84 7.49 7.06 7.38 Table A-12: Design table of RSM for Set 12 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM121 RSM122 RSM123 RSM124 RSM125 RSM126 30.0 RSM127 RSM128 RSM129 RSM1210 RSM1211 RSM1212 RSM1213 Best FTU = 7.77 9.0 90.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 8.90 7.77 9.54 9.12 9.69 9.25 11.95 15.36 13.24 12.70 13.30 12.10 12.80 70 Table A-13: Design table of RSM for Set 13 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM131 RSM132 RSM133 RSM134 RSM135 RSM136 30.0 RSM137 RSM138 RSM139 RSM1310 RSM1311 RSM1312 RSM1313 Best FTU = 2.80 7.0 30.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 3.09 2.80 3.16 2.92 4.19 3.99 4.41 4.06 3.52 3.51 3.68 3.70 3.60 Table A-14: Design table of RSM for Set 14 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM141 RSM142 RSM143 RSM144 RSM145 RSM146 30.0 RSM147 RSM148 RSM149 RSM1410 RSM1411 RSM1412 RSM1413 Best FTU = 15.87 7.0 150.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 17.71 16.31 19.02 18.65 19.90 17.68 15.87 23.23 15.87 17.71 17.24 17.07 16.90 71 Table A-15: Design table of RSM for Set 15 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM151 RSM152 RSM153 RSM154 RSM155 RSM156 30.0 RSM157 RSM158 RSM159 RSM1510 RSM1511 RSM1512 RSM1513 Best FTU = 6.67 7.0 90.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 11.12 10.26 11.58 10.71 10.74 9.76 7.75 11.75 7.21 6.88 6.91 6.67 6.92 Table A-16: Design table of RSM for Set 16 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM161 RSM162 RSM163 RSM164 RSM165 RSM166 30.0 RSM167 RSM168 RSM169 RSM1610 RSM1611 RSM1612 RSM1613 Best FTU = 5.95 7.0 90.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 6.22 5.95 7.94 7.40 9.51 8,35 6.96 12.54 8.40 7.59 7.73 6.91 7.66 72 Table A-17: Design table of RSM for Set 17 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM171 RSM172 RSM173 RSM174 RSM175 RSM176 30.0 RSM177 RSM178 RSM179 RSM1710 RSM1711 RSM1712 RSM1713 Best FTU = 7.93 7.0 90.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 9.10 8.48 10.29 9.70 8.78 7.93 9.49 14.08 8.99 9.45 8.94 8.80 9.04 Table A-18: Design table of RSM for Set 18 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM181 RSM182 RSM183 RSM184 RSM185 RSM186 30.0 RSM187 RSM188 RSM189 RSM1810 RSM1811 RSM1812 RSM1813 Best FTU = 8.76 7.0 90.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 9.46 8.76 12.07 11.42 9.95 9.29 9.44 17.23 10.00 10.00 10.05 9.81 9.96 73 Table A-19: Design table of RSM for Set 19 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM191 RSM192 RSM193 RSM194 RSM195 RSM196 30.0 RSM197 RSM198 RSM199 RSM1910 RSM1911 RSM1912 RSM1913 Best FTU = 6.92 7.0 90.0 Alum Dose (mg/L) 63.4 176.6 63.4 176.6 40.0 200.0 120.0 120.0 120.0 120.0 120.0 120.0 120.0 Variable Alum Amount (mL) 31.7 29.4 31.7 29.4 20.0 33.3 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 8.11 8.21 12.81 11.87 7.15 7.08 10.51 10.25 8.94 6.92 7.81 8.80 8.12 Table A-20: Design table of RSM for Set 20 Run Initial Water characteristic Suspended Temperature pH Solid, SS (°C) (mg/L) RSM201 RSM202 RSM203 RSM204 RSM205 RSM206 30.0 RSM207 RSM208 RSM209 RSM2010 RSM2011 RSM2012 RSM2013 Best FTU = 6.59 7.0 90.0 Variable Alum Alum Dose Amount (mg/L) (mL) 63.4 31.7 176.6 29.4 63.4 31.7 176.6 29.4 40.0 20.0 200.0 33.3 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 120.0 20.0 Response pH Turbidity (FTU) 5.0 5.0 7.5 7.5 6.3 6.3 4.5 8.0 6.3 6.3 6.3 6.3 6.3 8.02 7.97 11.75 12.81 8.28 8.25 9.96 9.97 6.59 7.86 8.68 8.78 7.97 APPENDIX B (DETAILS FOR FACTORIAL ANALYSIS) Fractional Factorial Fit: Opt. Dose versus Temp, pH, SS Estimated Effects and Coefficients for Opt. (coded units) Term Constant Temp pH SS Temp*pH Temp*SS pH*SS Temp*pH*SS Ct Pt Effect Coef 79.17 -1.67 -29.97 26.67 1.67 15.82 -12.47 -15.82 73.03 -3.35 -59.95 53.35 3.35 31.65 -24.95 -31.65 SE Coef 12.83 12.83 12.83 12.83 12.83 12.83 12.83 12.83 19.60 T 6.17 -0.13 -2.34 2.08 0.13 1.23 -0.97 -1.23 3.73 P 0.002 0.901 0.067 0.092 0.901 0.272 0.376 0.272 0.014 Analysis of Variance for Opt. (coded units) Source Main Effects 2-Way Interactions 3-Way Interactions Curvature Residual Error Pure Error Total DF 3 3 1 1 5 5 13 Seq SS 12902.9 3270.9 2003.4 18283.4 6586.1 6586.1 43046.7 Adj SS 12902.9 3270.9 2003.4 18283.4 6586.1 6586.1 Adj MS 4301 1090 2003 18283 1317 1317 F 3.27 0.83 1.52 13.88 P 0.118 0.533 0.272 0.014 Fractional Factorial Fit: Opt. pH versus Temp, pH, SS (Alpha = 0.1) Estimated Effects and Coefficients for Opt. (coded units) Term Constant Temp pH SS Temp*pH Temp*SS pH*SS Temp*pH*SS Ct Pt Effect 0.7500 -0.5000 0.5000 -0.7500 0.7500 -0.5000 -0.7500 Coef 5.2500 0.3750 -0.2500 0.2500 -0.3750 0.3750 -0.2500 -0.3750 0.6167 SE Coef 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.3626 T 22.12 1.58 -1.05 1.05 -1.58 1.58 -1.05 -1.58 1.70 P 0.000 0.175 0.340 0.340 0.175 0.175 0.340 0.175 0.150 Analysis of Variance for Opt. (coded units) Source Main Effects 2-Way Interactions 3-Way Interactions Curvature Residual Error DF 3 3 1 1 5 Seq SS 2.12500 2.75000 1.12500 1.30381 2.25333 Adj SS 2.12500 2.75000 1.12500 1.30381 2.25333 Adj MS 0.7083 0.9167 1.1250 1.3038 0.4507 F 1.57 2.03 2.50 2.89 P 0.307 0.228 0.175 0.150 75 Pure Error Total 5 13 2.25333 9.55714 2.25333 0.4507 Fractional Factorial Fit: Opt. pH versus Temp, pH, SS (Alpha = 0.2) Estimated Effects and Coefficients for Opt. (coded units) Term Constant Temp pH SS Temp*pH Temp*SS pH*SS Temp*pH*SS Ct Pt Effect 0.7500 -0.5000 0.5000 -0.7500 0.7500 -0.5000 -0.7500 Coef 5.2500 0.3750 -0.2500 0.2500 -0.3750 0.3750 -0.2500 -0.3750 0.6167 SE Coef 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.2373 0.3626 T 22.12 1.58 -1.05 1.05 -1.58 1.58 -1.05 -1.58 1.70 P 0.000 0.175 0.340 0.340 0.175 0.175 0.340 0.175 0.150 Analysis of Variance for Opt. (coded units) Source Main Effects 2-Way Interactions 3-Way Interactions Curvature Residual Error Pure Error Total DF 3 3 1 1 5 5 13 Seq SS 2.12500 2.75000 1.12500 1.30381 2.25333 2.25333 9.55714 Adj SS 2.12500 2.75000 1.12500 1.30381 2.25333 2.25333 Adj MS 0.7083 0.9167 1.1250 1.3038 0.4507 0.4507 F 1.57 2.03 2.50 2.89 P 0.307 0.228 0.175 0.150 APPENDIX C (RESPONSE SURFACE ANALYSIS FOR FACTORS THAT AFFECT COAGULATION - FULL QUADRATIC TERMS) Response Surface Regression: Optimum Dose versus Temperature, pH, SS The analysis was done using coded units. Estimated Regression Coefficients for Optimum dose Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Temperat*pH Temperat*SS (mg/L pH*SS (mg/L S = 57.45 Coef 154.42 -0.98 -17.56 8.66 -25.92 -5.91 -15.92 1.67 15.82 -12.47 R-Sq = 38.7% SE Coef 23.43 15.55 15.55 15.55 15.13 15.13 15.13 20.31 20.31 20.31 T 6.591 -0.063 -1.130 0.557 -1.713 -0.391 -1.052 0.082 0.779 -0.614 P 0.000 0.951 0.285 0.590 0.118 0.704 0.318 0.936 0.454 0.553 R-Sq(adj) = 0.0% Analysis of Variance for Optimum dose Source Regression Linear Square Interaction Residual Error Lack-of-Fit Pure Error Total Observation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 DF 9 3 3 3 10 5 5 19 Optimum 105.000 35.000 35.000 35.000 120.000 176.600 63.400 63.400 120.000 120.000 176.600 176.600 176.600 120.000 120.000 176.600 200.000 176.600 120.000 120.000 Seq SS 20859 5247 12341 3271 33003 26417 6586 53862 Fit 121.586 84.623 108.068 77.805 132.197 158.535 68.779 101.817 82.757 79.457 167.238 108.177 94.850 123.964 154.424 154.424 154.424 154.424 154.424 154.424 Adj SS 20859 5247 12341 3271 33003 26417 6586 SE Fit 47.016 47.016 47.016 47.016 47.016 47.016 47.016 47.016 44.769 44.769 44.769 44.769 44.769 44.769 23.430 23.430 23.430 23.430 23.430 23.430 Adj MS 2318 1749 4114 1090 3300 5283 1317 Residual -16.586 -49.623 -73.068 -42.805 -12.197 18.065 -5.379 -38.417 37.243 40.543 9.362 68.423 81.750 -3.964 -34.424 22.176 45.576 22.176 -34.424 -34.424 F 0.70 0.53 1.25 0.33 P 0.697 0.672 0.344 0.804 4.01 0.077 St Resid -0.50 -1.50 -2.21R -1.30 -0.37 0.55 -0.16 -1.16 1.03 1.13 0.26 1.90 2.27R -0.11 -0.66 0.42 0.87 0.42 -0.66 -0.66 77 R denotes an observation with a large standardized residual. Estimated Regression Coefficients for Optimum dose using data in uncoded units Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Temperat*pH Temperat*SS (mg/L pH*SS (mg/L Coef -2382.56 158.887 55.9955 0.0757627 -2.93269 -4.17929 -0.0125048 0.473762 0.149200 -0.294039 Response Surface Regression: Optimum pH versus Temperature, pH, SS The analysis was done using coded units. Estimated Regression Coefficients for Optimum pH Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Temperat*pH Temperat*SS (mg/L pH*SS (mg/L S = 0.7230 Coef 5.8626 0.2197 -0.4543 0.0849 -0.4566 0.1621 -0.3683 -0.3750 0.3750 -0.2500 R-Sq = 69.1% SE Coef 0.2949 0.1957 0.1957 0.1957 0.1905 0.1905 0.1905 0.2556 0.2556 0.2556 T 19.881 1.123 -2.322 0.434 -2.398 0.851 -1.934 -1.467 1.467 -0.978 P 0.000 0.288 0.043 0.674 0.037 0.415 0.082 0.173 0.173 0.351 R-Sq(adj) = 41.2% Analysis of Variance for Optimum pH Source Regression Linear Square Interaction Residual Error Lack-of-Fit Pure Error Total Observation 1 2 3 4 5 6 7 8 9 10 11 DF 9 3 3 3 10 5 5 19 Optimum 5.000 5.000 5.000 5.000 4.500 7.500 5.000 5.000 4.500 4.500 7.500 Seq SS 11.670 3.576 5.344 2.750 5.228 2.974 2.253 16.898 Fit 5.100 5.539 5.441 4.380 5.019 6.959 4.361 4.800 4.202 4.940 7.085 Adj SS 11.670 3.576 5.344 2.750 5.228 2.974 2.253 SE Fit 0.592 0.592 0.592 0.592 0.592 0.592 0.592 0.592 0.563 0.563 0.563 Adj MS 1.2967 1.1921 1.7814 0.9167 0.5228 0.5949 0.4507 Residual -0.100 -0.539 -0.441 0.620 -0.519 0.541 0.639 0.200 0.298 -0.440 0.415 F 2.48 2.28 3.41 1.75 P 0.087 0.142 0.061 0.219 1.32 0.384 St Resid -0.24 -1.30 -1.06 1.49 -1.25 1.30 1.54 0.48 0.66 -0.97 0.92 78 12 13 14 15 16 17 18 19 20 5.000 5.000 4.500 6.300 5.000 6.300 5.000 6.300 6.300 5.557 4.678 4.964 5.863 5.863 5.863 5.863 5.863 5.863 0.563 0.563 0.563 0.295 0.295 0.295 0.295 0.295 0.295 -0.557 0.322 -0.464 0.437 -0.863 0.437 -0.863 0.437 0.437 -1.23 0.71 -1.02 0.66 -1.31 0.66 -1.31 0.66 0.66 Estimated Regression Coefficients for Optimum pH using data in uncoded units Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Temperat*pH Temperat*SS (mg/L pH*SS (mg/L Coef -53.5591 3.59795 1.72582 -0.0103599 -0.0516633 0.114604 -0.000289329 -0.106066 0.00353553 -0.00589256 APPENDIX D (RESPONSE SURFACE ANALYSIS FOR FACTORS THAT AFFECT COAGULATION - LINEAR + SQUARED TERMS) Response Surface Regression: Optimum Dose versus Temperature, pH, SS The analysis was done using coded units. Estimated Regression Coefficients for Optimum Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L S = 52.82 Coef 154.42 -0.98 -17.56 8.66 -25.92 -5.91 -15.92 R-Sq = 32.7% SE Coef 21.54 14.29 14.29 14.29 13.91 13.91 13.91 T 7.168 -0.069 -1.228 0.606 -1.863 -0.425 -1.144 P 0.000 0.946 0.241 0.555 0.085 0.678 0.273 R-Sq(adj) = 1.6% Analysis of Variance for Optimum Source Regression Linear Square Residual Error Lack-of-Fit Pure Error Total Observation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 DF 6 3 3 13 8 5 19 Optimum 105.000 35.000 35.000 35.000 120.000 176.600 63.400 63.400 120.000 120.000 176.600 176.600 176.600 120.000 120.000 176.600 200.000 176.600 120.000 120.000 Seq SS 17587.9 5247.0 12340.9 36274.3 29688.3 6586.1 53862.2 Fit 116.561 114.598 81.443 79.480 133.872 131.910 98.754 96.792 82.757 79.457 167.238 108.177 94.850 123.964 154.424 154.424 154.424 154.424 154.424 154.424 Adj SS 17587.9 5247.0 12340.9 36274.3 29688.3 6586.1 SE Fit 28.679 28.679 28.679 28.679 28.679 28.679 28.679 28.679 41.165 41.165 41.165 41.165 41.165 41.165 21.544 21.544 21.544 21.544 21.544 21.544 Adj MS 2931.32 1749.00 4113.64 2790.33 3711.03 1317.22 Residual -11.561 -79.598 -46.443 -44.480 -13.872 44.690 -35.354 -33.392 37.243 40.543 9.362 68.423 81.750 -3.964 -34.424 22.176 45.576 22.176 -34.424 -34.424 F 1.05 0.63 1.47 P 0.438 0.610 0.267 2.82 0.134 St Resid -0.26 -1.79 -1.05 -1.00 -0.31 1.01 -0.80 -0.75 1.13 1.22 0.28 2.07R 2.47R -0.12 -0.71 0.46 0.94 0.46 -0.71 -0.71 R denotes an observation with a large standardized residual. Estimated Regression Coefficients for Optimum using data in uncoded units 80 Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Coef -2699.65 175.631 43.7448 2.49348 -2.93269 -4.17929 -0.0125048 Response Surface Regression: Optimum pH versus Temperature, pH, ... The analysis was done using coded units. Estimated Regression Coefficients for Optimum Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L S = 0.7834 Coef 5.8626 0.2197 -0.4543 0.0849 -0.4566 0.1621 -0.3683 R-Sq = 52.8% SE Coef 0.3195 0.2120 0.2120 0.2120 0.2064 0.2064 0.2064 T 18.350 1.036 -2.143 0.400 -2.213 0.785 -1.785 P 0.000 0.319 0.052 0.695 0.045 0.446 0.098 R-Sq(adj) = 31.0% Analysis of Variance for Optimum Source Regression Linear Square Residual Error Lack-of-Fit Pure Error Total Observation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 DF 6 3 3 13 8 5 19 Optimum 5.000 5.000 5.000 5.000 4.500 7.500 5.000 5.000 4.500 4.500 7.500 5.000 5.000 4.500 6.300 5.000 6.300 5.000 6.300 6.300 Seq SS 8.9203 3.5762 5.3441 7.9777 5.7244 2.2533 16.8980 Fit 5.350 5.789 4.441 4.880 5.519 5.959 4.611 5.050 4.202 4.940 7.085 5.557 4.678 4.964 5.863 5.863 5.863 5.863 5.863 5.863 Adj SS 8.92029 3.57616 5.34413 7.97771 5.72438 2.25333 SE Fit 0.425 0.425 0.425 0.425 0.425 0.425 0.425 0.425 0.610 0.610 0.610 0.610 0.610 0.610 0.319 0.319 0.319 0.319 0.319 0.319 Adj MS 1.48671 1.19205 1.78138 0.61367 0.71555 0.45067 Residual -0.350 -0.789 0.559 0.120 -1.019 1.541 0.389 -0.050 0.298 -0.440 0.415 -0.557 0.322 -0.464 0.437 -0.863 0.437 -0.863 0.437 0.437 F 2.42 1.94 2.90 P 0.085 0.173 0.075 1.59 0.317 St Resid -0.53 -1.20 0.85 0.18 -1.55 2.34R 0.59 -0.08 0.61 -0.90 0.85 -1.13 0.66 -0.94 0.61 -1.21 0.61 -1.21 0.61 0.61 R denotes an observation with a large standardized residual. Estimated Regression Coefficients for Optimum using data in uncoded units 81 Term Constant Temperat pH SS (mg/L Temperat*Temperat pH*pH SS (mg/L*SS (mg/L Coef -37.1189 3.17369 -1.98649 0.0544582 -0.0516633 0.114604 -0.000289329