Anaerobic treatment of complex chemical ... - Wiley Online Library

Anaerobic Treatment of Complex Chemical Wastewater in a Sequencing Batch Biofilm Reactor: Process Optimization and Evaluation of Factor Interactions Using the Taguchi Dynamic DOE Methodology S. Venkata Mohan,1 N. Chandrasekhara Rao,1 K. Krishna Prasad,1 P. Murali Krishna,1 R. Sreenivas Rao,2 P.N. Sarma1 1 Biochemical and Environmental Engineering Center, Indian Institute of Chemical Technology, Hyderabad-500007, India; telephone: þ 91 40 27015744; fax: þ 91 40 27160123; e-mail: [email protected] 2 Department of Microbiology, Osmania University, Hyderabad-500007, India

Received 7 June 2004; accepted 18 January 2005 Published online 5 April 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20477

Abstract: The Taguchi robust experimental design (DOE) methodology has been applied on a dynamic anaerobic process treating complex wastewater by an anaerobic sequencing batch biofilm reactor (AnSBBR). For optimizing the process as well as to evaluate the influence of different factors on the process, the uncontrollable (noise) factors have been considered. The Taguchi methodology adopting dynamic approach is the first of its kind for studying anaerobic process evaluation and process optimization. The designed experimental methodology consisted of four phases—planning, conducting, analysis, and validation connected sequence-wise to achieve the overall optimization. In the experimental design, five controllable factors, i.e., organic loading rate (OLR), inlet pH, biodegradability (BOD/COD ratio), temperature, and sulfate concentration, along with the two uncontrollable (noise) factors, volatile fatty acids (VFA) and alkalinity at two levels were considered for optimization of the anaerobic system. Thirty-two anaerobic experiments were conducted with a different combination of factors and the results obtained in terms of substrate degradation rates were processed in Qualitek-4 software to study the main effect of individual factors, interaction between the individual factors, and signal-to-noise (S/N) ratio analysis. Attempts were also made to achieve optimum conditions. Studies on the influence of individual factors on process performance revealed the intensive effect of OLR. In multiple factor interaction studies, biodegradability with other factors, such as temperature, pH, and sulfate have shown maximum influence over the process performance. The optimum conditions for the efficient performance of the anaerobic system in treating complex wastewater by considering dynamic (noise) factors obtained are higher organic loading rate of 3.5 Kg COD/m3 day, neutral pH with high biodegradability (BOD/COD ratio of 0.5), along with mesophilic temperature range (408C), and low sulfate concentration (700 mg/L). The optimization resulted in enhanced anaerobic performance (56.7%) from a substrate degradation rate (SDR) of 1.99 to 3.13 Kg COD/m3 day. Considering the obtained optiCorrespondence to: S. Venkata Mohan

ß 2005 Wiley Periodicals, Inc.

mum factors, further validation experiments were carried out, which showed enhanced process performance (3.04 Kg COD/m3-day from 1.99 Kg COD/m3 day) accounting for 52.13% improvement with the optimized process conditions. The proposed method facilitated a systematic mathematical approach to understand the complex multispecies manifested anaerobic process treating complex chemical wastewater by considering the uncontrollable factors. ß 2005 Wiley Periodicals, Inc. Keywords: anaerobic process; anaerobic sequencing batch biofilm reactor (AnSBBR); chemical wastewater; Taguchi robust methodology; signal-to-noise (S/N) ratio; Qualitek-4; sulfate; biodegradability

INTRODUCTION The wastewater generated from the chemical-based industries are highly variable in composition and are considered complex due to the consumption of large quantities of raw materials falling into general and specific classification of chemicals, ranging from organic and inorganic, salts, solvents, animal extracts, etc. (Venkata Mohan and Sarma, 2002). Due to the complex nature and presence of recalcitrant compounds in the wastewater, treatment of these is considered difficult and challenging. The anaerobic process is a sequential series of biological processes manifested by complex multispecies reactions in which hydrocarbons are converted in the absence of free oxygen, from complex to simple molecules and ultimately to carbon dioxide and methane (a renewable energy source) by a series of fermentation reactions mediated by a divergent group of microorganisms, i.e., hydrolytic microorganisms (hydrolysis) degrade polymeric materials such as polysaccharides and proteins to monomers without COD reduction producing acetate, hydrogen, and varying amounts of volatile fatty acids (VFA); acidogenic bacteria (acetogenesis) converts monomers into acids and H2; and methanogenic bacteria (methanogenesis) converts the organic acids to CH4 and CO2. The

stepwise destruction of the molecules is conducted by a specific group of microorganisms that thrive during the catabolic process, and capture the energy of the hydrocarbons. The metabolic interaction between the various groups of organisms is essential for the successful and complete mineralization of the organic molecules. Anaerobic processes have a wide application in the treatment of sewage and high strength industrial wastewater treatment (Ahring, 2003; Angelidaki et al., 2003; Athanasopoulos, 1992; Lettinga et al., 2001; Rebac et al., 1995; Rittmann and McCarty, 2001; Venkata Mohan and Sarma, 2002; Venkata Mohan et al., 2001; Vossoughi et al., 2003). The success of new highrate anaerobic technology has encouraged environmental researchers to extend its application to treat wastewater of a more complex nature. Process efficiency is mainly dependent on the reactor operating conditions and wastewater characteristics and various parameters are monitored to maintain efficient operating conditions within the reactor. Most advances in anaerobic technology have been achieved during the last three decades with the development of high-rate reactor systems and a better knowledge of the microbiology of methanogenic ecosystems (Lier et al., 2001). Anaerobic biological processes have received great attention in wastewater treatment, owing to their attractive advantages of a high capacity to treat slowly degradable substrates at high concentrations, very low sludge production, low energy requirements, and the possibility for energy recovery through methane combustion (Ahring, 2003; Rittmann and McCarty, 2001). The distinct disadvantages of anaerobic process are that methanogenic organisms grow slowly, the stability of anaerobic processes can be upset either by toxic substrates or by overloading, and the process is not completely understood. The presence of recalcitrant compounds, heavy metals, sulfate, TDS, etc. in wastewater affects the anaerobic process efficiency. In this context, research aims not only to extend the potential of anaerobic processes (Verstraete and Vandevivere, 1999), but also to optimize anaerobic processes and increase their robustness (Lier et al., 2001). The operation of an anaerobic treatment system requires a solid understanding of the role of the important factors involved in the operation and control (Rittmann and McCarty, 2001). Problems encountered in anaerobic process operation are difficult to identify before the process is affected. Even though the anaerobic process is a mature technology, there is plenty of scope for optimization and improvement. Optimization of the process parameters for improving the overall microbial process is one of the important ways to enhance the process efficiency along with the biogas yield (Ahring, 2003; Wasserman, 1998). Recently, considerable attention has been given to optimizing the system performance for anaerobic wastewater treatment processes, through different approaches like advanced control of anaerobic process through disturbance monitoring, black box and neural network models to understand the anaerobic digestion process (Bryers, 1985; Denac et al., 1990; Premier et al., 1999; Steyer et al., 1999). A common problem for bioreactor control

originates from the nonlinear nature of such systems, that renders the conventional feedback control strategies unsatisfactory (Ryhiner et al., 1992). Moreover, the anaerobic treatment process is extremely complex and dynamic compared to the aerobic operation. It involves multiple sequencing steps of fermentation and each step is controlled by a specific condition. Compared to a static system, a dynamic system differs by signal factors to achieve the target performance and the response varies with the level of the signal factor (Phadke and Dehnad, 1988). The design of experiments (DOE) methodology by Taguchi orthogonal array (OA), a factorial-based approach, has gained exceeding importance recently for its application in optimizing biochemical processes. The approach involves the study of a given system by a set of independent variables (factors), both controllable and uncontrollable (dynamic/ noise), over a specific region of interest (levels) (Mitra, 1998). Unlike traditional experimental design, which focuses on the average process performance characteristics, it concentrates on the effect of variation on the process characteristics (Phadke, 1989; Ross, 1996) and makes the product/ process performance insensitive (robust) to variation with respect to uncontrolled or noise factors by the proper design of parameters. In this methodology, the desired design is sought by selecting the best performance under conditions that produce consistent performance exposed to the influence of the uncontrollable (noise) factors by a dynamic approach (Roy, 2001). This method makes the process robust and minimizes the effect of noise by tuning the important controllable factors (Dehnad, 1989; Mitra, 1998; Tong et al., 2004). Moreover, conclusions drawn from the small-scale experiments are valid over the entire experimental region spanned by control factors and their setting levels (Phadke, 1989). Also, this approach facilitates the identification of the influence of individual factors, establishing the relationship between factors and operational conditions, and finally establishing performance at the optimum levels obtained. Analysis of the experimental data using ANOVA (analysis of variance) and factor effects give the factors, which are statistically significant and result in finding the optimum levels of factors for design parameters through confirmation experiments. The Taguchi method not only helps in saving considerable time and cost, but also leads to a more fully developed process (Raghu, 1985; Phadke, 1989; Taguchi, 1993). Its application created significant changes in several industrial organizations in Japan and the USA in manufacturing processes and total quality control (Taguchi, 1993). Recently, it has been used for a few biochemical (Cobb and Clarkson, 1994; Jeney et al., 1999) and bioprocess applications (Han et al., 1998; Venkata Dasu et al., 2003; Sreenivas Rao et al., 2004; Krishna Prasad et al., 2005). Its use in environmental engineering, especially for anaerobic process optimization has not been reported. However, Barrado et al. (1996, 1998) used the method to optimize a physical process for purification of metal containing wastewater. In this article, the applicability of the Taguchi robust experimental design methodology for dynamic anaerobic

VENKATA MOHAN ET AL.: ANAEROBIC TREATMENT OF COMPLEX CHEMICAL WASTEWATER IN AN AnSBBR

733

system optimization with consideration of uncontrollable/ noise factors is investigated. Our main objectives are to: Optimize the complex anaerobic process by considering its dynamic behavior Understand the influence of factors individually and in combination with anaerobic process performance Establish the optimum conditions with the selected factors by fine-tuning controllable factors for an effective performance of the anaerobic system In the robust experimental design for optimizing the anaerobic system, five controllable factors, i.e., organic loading rate (OLR), inlet pH, biodegradability (BOD/COD ratio), operating temperature and sulfate concentration, which significantly influence the anaerobic treatment process, along with two uncontrollable (noise) factors, alkalinity and VFA at the two level, have been considered.

impractical and Taguchi’s method seeks to minimize the effect of noise in determining the optimal levels of the important controllable factors, based on the concept of robustness (Dehnad, 1989). Taguchi’s main innovation was design of a robust system with discrete event simulation making provision for reducing noise variation, by way of controlling the noise interactions (Madu and Madu, 1999). Noise interaction control has been carried out by choosing control factor settings, to reduce the sensitivity of a response to noise factors (Hou, 2003; Taguchi, 1993). Taguchi used the signal-to-noise (S/N) ratio as a performance measure in a dynamic system to assess the robustness of a process (Tong et al., 2004) and showed the magnitude of the interactions, between control factors and noise factors (Madu and Madu, 1999). The term signal represents the square of the mean value of the performance/quality characteristics, whereas noise is a measure of the variabilities (as measured by the variance) of the characteristics and S/N ratio and system sensitivity are represented as follows S=N ¼ 10 log10 ðb2 =s2 Þ S ¼ 10 log10 b2

EXPERIMENTAL METHODOLOGY Robust Experimental Design (Phase I) The Taguchi method involves the establishment of a large number of experimental situations described as orthogonal arrays (OA) to reduce experimental errors and to enhance the efficiency and reproducibility of laboratory experiments. Robust design has been considered in this study to minimize the noise/uncontrollable factor on the overall process, which helped to eliminate the effect of uncontrollable parameters in the process of optimization, leading to a dynamic/robust experimental design. In Taguchi’s method, performance is measured by the deviation of a characteristic from its target value and a loss function [L(y)] is developed for the deviation (Mitra, 1998), as represented by LðyÞ ¼ kðy mÞ2

ð1Þ

where k denotes the proportionality constant, m represents the target value, and y is the experimental value obtained for each trial. In case of ‘‘bigger is better’’ quality characteristics, the loss function can be written as LðyÞ ¼ kð1=y2 Þ

ð2Þ

and the expected loss function can be represented by: E½LðyÞ ¼ k Eð1=y2 Þ

ð3Þ

where E(1/y2) can be estimated from a sample of n as n X

½1=y2i =n

ð4Þ

i¼1

Uncontrollable factors, known as noise factors cause deviations and thereby lead to loss. Elimination of noise factors is

734

ð5Þ ð6Þ

where b reflects the system’s sensitivity under a certain control condition, s represents the variance and S represents the system sensitivity in a dynamic system. The mathematical expression for the S/N ratio for the ‘‘bigger is better’’ case for the performance statistics that measure deviation from the target, called as mean square deviation (MSD) was given by n P 2 Z ¼ 10 logðMSDÞ ¼ 10 log 1=y =h ð7Þ i¼1

In this approach, the desired design was sought not by selecting the best performance under ideal conditions but by looking for the design that produces consistent performance exposed to the influence of the uncontrollable factors. The process flow diagram indicating the inputs, outputs, controllable and uncontrollable/noise factors considered in the anaerobic dynamic system of process optimization is depicted in Figure 1. The total performance model for the optimization of the experimental methodology adopted in this study has been broadly divided into four phases (with various steps)—planning, conducting, analysis, and validation. The methodology adopted is depicted in Figure 2. Each phase had a separate objective, interconnected in sequence to achieve the overall optimization process. The first step in phase one was to determine various important factors to be optimized in the anaerobic process having a critical effect on the performance. The normal practice has been to experiment with the feasible range so that the variation inherent in the process does not mask the factor effect. Factors were selected and the ranges were further assigned, based on the consensus obtained by the group consisting of design engineers, scientists, and technicians with relevant experience. Operational factors affect the habitat of microorganisms and consequently the anaerobic treatment process. The most important operational factors

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 90, NO. 6, JUNE 20, 2005

Figure 1. Process flow diagram of proposed Taguchi dynamic/robust experimental design. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

Figure 2. Schematic representation of designed experimental methodology.

that have been considered in this study were the characteristics and composition of selected chemical wastewater (sulfate concentration and biodegradability), OLR, inlet pH, and operating temperature, as controllable parameters to understand the influence on the performance of anaerobic treatment. Due to the complex and dynamic nature of the anaerobic process, the uncontrollable/noise parameters, alkalinity and VFA concentration have also been selected in the design to reduce signal-to-noise ratio (S/N), to enhance the process understanding in a dynamic environment. The variation of VFA and alkalinity (buffering capacity) is dependent on a biochemical reaction, manifested by operating conditions and influent feed characteristics. VFA, as an indicator of imbalance in the anaerobic process have been one of the significant factors that could not be controlled. VFA production and accumulation are dependent on the concentration and composition of the wastewater, the anaerobic reactions involved, and operational conditions. VFA accumulation results in unbalanced microbial consortia, which is detrimental in the anaerobic process operation and has led to total system failure (Ahring et al., 1995; Komisar et al., 1998). It was recommended that the desirable operating range of VFA is 0–8.3 meq/L (500 mg/L) (Barford, 1988). Therefore, VFA concentration had to be maintained below the inhibitory level to enhance the reactor performance. Alkalinity was one of the important factors, which was governed by the VFA production and accumulation along with the operation conditions of


735

Table I. Selected factors and assigned levels. Sample No. 1 2 3 4 5 S N

Factor

Level 1

Level 2

1.7 7.0 0.3 700 20

3.5 8.5 0.5 1500 40

10 500

33 2000

Controllable factors OLR (kg COD/m3-day) Inlet pH Biodegradability (BOD/COD) Sulfate concentration (mg/L) Temperature (8C) Uncontrollable (noise) factors VFA (meq/L) Alkalinity (mg/L)

the reactor. Alkalinity concentration in a system acts as a buffer to sustain anaerobic performance in the presence of VFA production. These two factors were uncontrollable and were entirely dependent on the wastewater characteristics, retention time of the reactor operation, reactor configuration, existing buffering capacity of the influent and operating conditions. The second step was to design the matrix experiment and define the data analysis procedure with an inner and outer array. Taguchi’s methodology for robust parameter design involves the use of orthogonal designs, where an orthogonal array (OA) involving control variables was crossed with an OA for the uncontrollable/noise variables. The cross-array format consisting of a product of two OAs, one for control factors, called an outer array and one for noise factors, called an uncontrollable array referred to as an inner–outer array was adopted in this study. In the design, the OA of each column consisted of a number of conditions depending on the levels assigned to each factor. In the present case, two levels of factor variation were considered and the size of experimentation was represented by symbolic arrays, e.g., L8 L4 [which indicates 32 (8 4) experimental trials]. Anaerobic experiments for the treatment of complex chemical wastewater were carried out employing selected eight experimental trials in combination with five control-

lable factors and two uncontrollable factors at two levels (Table I). Table II represents a designed experimental matrix composed of an inner and outer array along with the selected levels of experiments. The diversity of factors was studied by crossing the orthogonal array of factors as shown in Table II. The matrix, that designated the settings of the controllable factors (design parameters) for the experiment was called the inner array and the matrix, that designates the settings of the uncontrollable/noise factors was called the outer array. Anaerobic Experiments With Selected Factors and Levels (Phase 2) The composite wastewater aggregated from various chemical industries producing a variety of intermediates (chemical), dyes, drugs, pharmaceuticals, pesticides, etc., was used as feed for the anaerobic reactors. The characteristics (pH 7.8, total dissolved inorganic solids (TDIS)—11 g/L, SS—0.9 g/ L, nitrogen—35 mg/L, COD—6 g/L, BOD—1.9 g/L, chlorides—5 g/L, sulfates—1.6 g/L) of the wastewater was complex and could be assessed by the presence of a low BOD/COD ratio (0.31), high sulfate content (1.75 g/L), and high TDS concentration (11 g/L). The wastewater was stored at 48C, to maintain more or less uniform characteristics of the wastewater throughout the study. All the experiments were carried out in an anaerobic sequencing batch biofilm reactor (AnSBBR) fabricated, using perspex acrylic material with a total working volume of 1.6-L capacity (internal diameter of 0.045 m; L/D ratio of &4). The reactor was operated in upflow mode, with biofilm configuration employing recirculation to a feed ratio of 2. Inert stone chips (average size of 0.025 cm 0.015 cm; void ratio—0.49) were used as fixed bed for supporting the biofilm growth with a bed height of 0.36 m (liquid height 0.4 m). A total cycle period of 24 h (HRT) consisting of 15 min of fill phase, 23 h of react (anaerobic) phase with recycling, 30 min of settle phase, and 15 min of decant phase was used for all the experiments. At the beginning of each cycle, immediately after withdrawal (earlier sequence), a

Table II. Orthogonal array (OA) of designed experiments. Outer array

Factors Experiment No.

1

2

3

4

5

1 2 3 4 5 6 7 8

1 1 1 1 2 2 2 2

1 1 2 2 1 1 2 2

1 2 1 2 1 2 1 2

1 2 1 2 2 1 2 1

1 2 2 1 1 2 2 1

N

1

2

1

2

S

1

1

2

2

Substrate degradation rate (SDR) (kg COD/m3 day) 0.80 0.86 0.98 0.89 2.41 2.27 2.34 1.94

0.84 0.91 1.02 0.91 2.62 2.17 2.28 2.34

} Inner Array

736


0.81 0.72 1.00 0.79 2.44 2.07 2.04 2.01

0.61 0.62 0.82 0.54 1.69 1.18 1.34 1.19

}

Results

predefined feed volume (1.3 L) was pumped into the system and the reactor volume was recirculated during the reaction phase. At the end of the cycle, during settle phase, the suspended biomass settled, and the wastewater was withdrawn from the reactor. Feeding, withdrawal, and recirculation operations were carried out using peristaltic pumps (Watson Marlow 101U/R). As per the designed methodology, the reactor was operated with two OLRs of 1.7 and 3.5 kg COD/ m3 day. The influent pH was adjusted, before feeding wastewater using 3N NaOH. The influent wastewater OLR and sulfate concentrations as required were adjusted by diluting with tap water prior to feeding. Biodegradability of the wastewater was also adjusted by diluting with synthetic feed (glucose—1.0 g/L, yeast extract—0.2 g/L, Na2HPO4—0.3 g/ L, NH4Cl—0.5 g/L, KH2PO4—0.25 g/L, MgCl2 6H2O— 0.3 g/L, CuCl2—0.15 g/L, MnCl2—0.15 g/L, ZnCl2— 0.015 g/L; CaCl2—0.065 g/L) prior to feeding. Reactor aqueous-phase temperature was maintained by adjusting the temperature of recirculated wastewater as and when required. Biogas was collected by a water displacement method through an outlet provided at the top of the reactor. To start-up the reactor, an anaerobic biomass (VSS—2 g/L) acquired from a UASB reactor treating chemical effluents in the laboratory was used as inoculum and operated with synthetic feed (glucose—1.0 g/L, yeast extract—0.2 g/L, Na2HPO4— 0.3 g/L, NH4Cl—0.5 g/L, KH2PO4—0.25 g/L, MgCl2 6H2O—0.3 g/L, CuCl2—0.15 g/L, MnCl2—0.15 g/L, ZnCl2—0.015 g/L, CaCl2—0.065 g/L) to facilitate the biofilm formation on the supporting medium. Constant COD removal and gas production were considered as indictors for the satisfactory formation of biofilm and subsequently the reactor was operated with the chemical wastewater until constant COD removal was observed. During the reactor operation, the biogas yield varied between 0.12 m3 CH4/kg of COD removed and 0.36 m3-CH4/kg of COD removed. All the combinations of experiments as per the assigned levels were conducted in a random manner. The experimental results were presented as substrate degradation rate (SDR), to represent the anaerobic process performance and the SDR values obtained from the designed 32 experiments are given in Table II. Data Analysis and Performance Evaluation (Phase 3) The experimental data obtained has been processed using Qualitek-4 software to evaluate the influence of individual factors, multiple interaction of the selected factors on the anaerobic process performance, determination of optimum conditions for the anaerobic process treatment of the complex chemical wastewater, and the anaerobic process performance at the obtained optimum conditions was estimated. Qualitek-4 software (Nutek Inc., Bloomfield Hills, MI) was used for the design and analysis of the Taguchi experiments in the present study. The software was equipped to use L-4 to L-64 arrays along with a selection of 2 to 63 factors with two, three, and four levels to each factor. The automatic design

Table III. Main effects of selected controllable factors. Sample No. 1 2 3 4 5

Factor

Level 1

Level 2

L2–L1

OLR Inlet pH Biodegradability Temperature Sulfates concentration

1.46 2.16 1.89 1.82 2.15

2.53 1.83 2.10 2.17 1.84

1.07 0.33 0.21 0.35 0.31

option allowed Qualitek-4 to select the array used and assign factors to the appropriate columns. Data analysis options offer the choice of standard or S/N analysis by utilizing bigger, smaller, or nominal-is-best categories. In the present study S/N analysis was employed with bigger-is-better performance characteristics for all the experimental cases. The results obtained from the data processing are shown in Tables III to IX. Experimental Validation (Phase 4) To validate the method, anaerobic experiments were further performed with the chemical wastewater under study, by employing the established optimized process conditions from the proposed methodology (as shown Table IX). Analytical Protocols The performance of the reactor treating chemical wastewater was assessed by monitoring COD removal and represented as SDR. In addition, sulfates, BOD, pH, alkalinity, soluble sulfide, and VFA as acetic acid have also been determined during the operation to examine the performance of the reactor. Standard analytical procedures were followed for monitoring the above parameters [American Public Health Association (APHA), 1998]. RESULTS Influence of Individual Factors Results obtained with the designed experimental sets showed significant variation in the performance of the anaerobic operation. The process efficiency has been found to be very much dependent on the selected process conditions. The effect of the controllable factors and uncontrollable (noise) factors on the process efficiency of anaerobic operation is presented in Tables III and IV, respectively. The difference between values at levels 2 and level 1 (L2 L1) of each factor indicates the relative influence of the effect. The larger the difference, the stronger is the influence. The negative value has been ignored in assessing the main effect as the Table IV. Main effects of selected uncontrollable (noise) factors. Sample No. 1 2

Factor

Level 1

Level 2

L2–L1

VFA Alkalinity

1.59 1.47

1.18 1.31

0.42 0.16


737

Figure 3. Influence of controllable factors on the performance at selected levels. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

placement order of levels assigns either positive or negative values. Among the selected factors, OLR showed a stronger influence on the process, while the biodegradability of wastewater showed the least influence. Figure 3 represents the influence of the individual controllable factors. Individually at the level stage, the OLR along with biodegradability and temperature showed the highest effect at level 2; whereas pH and sulfate exhibited higher effects at level 1 on the anaerobic process performance. From the data, the relative influence of the controllable factors on the anaerobic performance was as follows OLR > operating temperature, pH > sulfate concentration > biodegradability In the case of the uncontrollable factors studied, VFA concentration showed relatively higher effect compared to the alkalinity (buffering capacity) (Table IV). It was evident that the concentration of VFA affects the system alkalinity concentration.

Influence of Factors Interaction Understanding the interaction between two factors gave a better insight into the overall process analysis. Any one factor may interact with any of the other factors creating the possibility of the presence of a large number of interactions. The estimated interaction (severity index—SI) of the different factors under study, helped to know the influence of two individual factors at various levels of the interactions (Table V). In the table, the ‘‘columns’’ represent the locations to which the interacting factors were assigned. The SI interaction presented 100% of SI for a 908 angle between the lines while it was 0% SI for the parallel lines. The ‘‘Reserved’’ column shows the column that should be reserved if this interaction effect is to be studied. ‘‘Levels’’ indicate the factor levels desirable for the optimum conditions (based on the first two levels). The relative interactions of the factors on the process performance are depicted in Figure 4. From the data, biodegradability with temperature (at level 1,

Table V. Estimated interaction of severity index for two factors. Sample No. 1 2 3 4 5 6 7 8 9 10

738

Factors Biodegradability Temperature Biodegradability Sulfates Biodegradability pH Temperature Sulfates pH Temperature pH Sulfates OLR Biodegradability OLR pH OLR Sulfates OLR Temperature

Columns

Severity index interaction (%)

Reserved column

Levels

45 46 24 56 25 26 14 12 16 15

75.4 51.3 48.5 37.6 35.6 32.4 24.6 17.7 17.5 14.5

1 2 6 3 7 4 5 3 7 4

[1,2] [1,1] [1,1] [2,1] [1,2] [1,1] [2,1] [2,1] [2,1] [2,2]


Figure 4. Interaction of selected factors on the performance. [Color figure can be seen in the online version of this article, available at www. interscience.wiley.com.]

2; column 1) interaction showed highest SI (75.38%) followed by biodegradability and sulfate (with 51.33% of SI) and biodegradability and pH (with 48.46% of SI). Biodegradability (least impact factor) with sulfate and pH combination also showed higher SI, respectively. It is interesting to note that biodegradability and temperature with relatively low effect, showed higher SI in combination. Contrary to this, the combination of biodegradability (the least effect factor) with OLR (high effect factor) showed relatively low SI (24.61%). Subsequently, temperature (the second higher effect factor) with sulfate and temperature showed SI of 37.57% and 35.65%, respectively. pH with sulfate combination resulted in SI of 32.40%. The individual factor (OLR) that showed the maximum effect on the

performance resulted in relatively low SI in combination with other factors. OLR interaction with temperature (second highest effect factor) gave a low SI (14.48%). ANOVA The contribution of individual factors is the key for the control to be enforced on the anaerobic processes. In the Taguchi approach, analysis of variance (ANOVA) was used to analyze the results of the OA experiment and to determine the variation due to each factor. ANOVA with the percentage of contribution of each factor with interactions are shown in Table VI. It is evident from F ratios, that all the factors and interactions considered in the experimental design had


739

Table VI. Analysis of Variance (ANOVA). Sample No. 1 2 3 4 5 Other/Error Total

Controllable factors

Degree of freedom

Sum of squares df

Variance

F Ratio

Pure sum

Percent contribution (%)

1 1 1 1 1 2 7

230.34 21.49 8.50 24.56 19.02 23.96 327.88

19.23 21.49 8.50 24.56 19.02 11.98

19.23 1.79 0.71 2.05 1.59

218.37 9.52 0 12.58 7.04

66.59 2.90 0 3.84 2.15 24.51 100.00

OLR Inlet pH Biodegradability Temperature Sulfates

statistically signiﬁcant effects at 95% confidence limit. The variability of the experimental data has been explained in terms of significant effects. By studying the main effects of each of the factors, the general trends of the influence of the factors towards the process could be characterized. The characteristics could be controlled such that a lower or a higher value of a particular influencing factor produced the preferred result. Thus, the level of factors to produce the best results could be predicted. It is evident from Table VI that the experimental-df was 7, while the factor-df was 1. The percentage contribution was calculated for each individual factor by the ratio of pure sum to the total sum of the squares. The most inﬂuential factor was the OLR, accounting for 66.59% of the overall variance of the experimental data followed by temperature (3.84%), pH (2.9%), and sulfate (1.15%). However, biodegradability showed negligible % contribution. Signal-to-Noise (S/N) Analysis and Performance Measure Taguchi had combined two components into one measure (average value of the characteristics called desirable components and undesirable components) known as the S/N ratio for performance measure and attempts have been made to select the parameter levels that maximize this ratio (Mitra, 1998). In this study, ‘‘bigger is better’’: performance characteristics for S/N analysis were opted. Eight trials with four samples were performed and corresponding S/N ratios are presented in Table VII. It can be seen from the table that S/N ratios varied between 0.62 and 2.68. The estimated result from the S/N ratio was 3.13 with MSD of 0.00074 [by

Equation (7)]. Expected SDR was found to be 3.67 kg COD/m3 day from the optimum S/N ratio. A pooled ANOVA of the S/N ratio values revealed that the pH was the only significant factor responsible for the variability to the S/N ratio (72%) contributing the total variance. An attempt has also been made to pool (above 95% of confidence level) until the degree of freedom (df) for the error terms was about half the experimental df. Other factors showed very little variance (2.7), the MB predominated, while


743

at lower COD/SO4 ratios (