Optimization of coagulation – ﬂ occulation process in the treatment of surface water for a maximum dissolved organic matter removal using RSM approach

The aim of this research work is the optimization of the coagulation/ ﬂ occulation process in the treatment of surface water for maximum dissolved organic matter (DOM) removal using response surface methodology (RSM). For this purpose, several jar test experiments were performed in order to identify the most in ﬂ uencing factors. Afterwards, RSM was done to investigate the effects and the interactions of three chosen variables (coagulant concentration, ﬂ occulant concentration, and initial pH), whereas the responses were the DOM removal in terms of chemical oxygen demand (COD), in terms of absorbance at the wavelength 254 nm (UV-254), and the ﬁ nal pH. The optimal conditions were as follows: 133 mg/L of coagulant, 60 mg/L of ﬂ occulant and an initial pH equal to 6.91. Under these conditions, the ef ﬁ ciency removals were 56% in terms of COD and 59% in terms of UV-254 with a ﬁ nal pH equal to 6.78. High variance coef ﬁ cient R 2 values, with 0.96 for the removal in terms of COD and 0.92 in terms of UV-254, con ﬁ rm the reliability and the validity of the obtained model.


INTRODUCTION
Drinking water production in a scenario where the contamination of aquatic environments is increasing has become a subject of global interest (Feihrmann et  Traditionally, to balance between water supplies and water demand for humans, natural water resources such as rainwater, surface freshwater run-off, catchment and groundwater reservoirs have been used in this case. However, the quality of natural waters has suffered serious deterioration, especially surface water (lakes, rivers, dams …), and the main cause of this degradation is the excessive increase of dissolved organic matter, the emer- Many studies have proven that the chlorination of water intended for public consumption, in the presence of DOM, leads to the formation of disinfection by-products such as trihalogenethanes, and carcinogenic and toxic organic compounds (Ghernaout & Elboughdiri ; Jiang et al. ). Furthermore, the presence of DOM can be the cause of many other problems such as taste, odour, colors, biofilms, metal complexes, membrane fouling, high consumption of chemicals …etc. (Vepsäläinen et al. ; Mostofa et al. ). For all these reasons, the elimination of DOM or at least its reduction to an acceptable level is an imperative process that must be done to protect the health of consumers by producing drinking water that meets water quality standards and to avoid any problems that may be caused by their presence.
The Timgad dam (NE Algeria) is a reservoir dam used mainly to produce drinking water for one million inhabitants of the city of Batna (Amrane & Bouhidel ). As all surface water, the main sources of the DOM in this dam are the effluents of domestic wastewater after treatment (biological wastewater treatment) and domestic sewage along with excreta by human beings and various animals, which are directly or indirectly discharged into the watershed of the dam.
The plant of this dam uses a conventional treatment based on the use of the coagulation/flocculation (CF) technique, since it has been widely used for the treatment of surface water due to its several advantages such as the removal of organic, suspended and colloidal matters. It can also be effective in removing many protozoa, bacteria and viruses, it has been found to be cost-effective and easy to operate, and it is an energy-saving treatment alternative (Mhamdi et al. ; Feihrmann et al. ).
Generally, CF treatment is influenced by several parameters, such as coagulant and flocculant nature and concentrations, pH, temperature, etc. (Hussain et al. ).
Therefore, finding the optimum value for these factors can improve the effectiveness of treatment.
Often, using traditional optimization, the optimal value for each variable is carried out by changing one factor while keeping all other factors constant. The problem with this method is that it takes a long time and that it is incapable of studying the effect of the interactions between the studied variables. This means this method is considered ineffective for determining the real optimal conditions. In order to find a solution to these deficiencies in the traditional method, the optimum values for the variables are determined using response surface methodology (RSM) (Mhamdi et al. ; Louhıchı et al. ).
In fact, RSM is a very useful tool, and a widely used technique for many treatment process optimizations. It is based on a group of mathematical and statistical methods in order to build a model of complex systems that makes it possible to evaluate the effects between the studied variables and to determine the optimal conditions which give the best possible processing performance and to predict the responses (Adlan et al. ; Guvenc et al. ).
The advantages of this method compared with the traditional method are that (i) it does not require a large number of experiences, (ii) it allows the studying of many variables at the same time and (iii) it makes it possible to study the interactions between the variables precisely and in a short time (Khouni et al. ).
Within this framework, this research work aims to enhance the performance of Timgad's treatment plant and to determinate the optimal operating conditions in order to obtain the best treatment performance in particular for the removal of dissolved organic matter from Timgad dam water by the CF treatment using RSM.
To the best of our knowledge, no study has so far been reported on the optimization of the CF process used on Timgad's treatment plant using RSM.

Study area
The studied water samples were taken from the Timgad dam

Analytical methods
The water samples used in this investigation were obtained from the Timgad dam in Algeria. Their physicochemical characteristics were investigated in order to measure their level of pollution before and after the treatment process.  In this study, the measurement of UV-254 was also used as a parameter to monitor the DOM amount before and after each treatment applying the CF process. UV-254 absorbance has been reported as a simple and reliable parameter used for real-time monitoring and control of DOM removal in many treatment processes, and it has also been reported that it can be used as a useful parameter to represent the aromatic character of the organic compounds present in water samples (Albrektienė et al. ). In this research work, UV-254 absorbance was measured using a spectrophotometer type Perkin Elmer with a quartz cell of 1 cm path length.

Optimization of CF process using RSM approach
The conventional 'one-factor-at-a-time-method' approach is laborious and time-consuming, especially for a large number of variables. Moreover, it seldom guarantees the determination of optimal conditions (Choudhari & Singhal ). These limitations of a single factor optimization process can be overcome by using statistical methods. For this purpose, response surface methodology (RSM) was used as optimization method. RSM, also known as Box-Wilson methodology, is an optimization method applying statistical techniques based upon the factorial designs of central composite design (CCD) and Box-Behnken design (BBD). An RSM approach will not only be able to determine the optimum conditions from a minimal number of experiments, but also give the information to estimate results in order to design a process (Humbird & Fei ).
In order to optimize the removal of the DOM from the Timgad dam water, a CCD model based on three factors was used as the experimental design model (Table 1). Coagulant concentration (50-150 mg/L), flocculant concentration (10-70 mg/L), and pH (5-9) were taken as input variables.
The experiments were performed in random order to avoid systematic error. Three-level Box-Behnken full factorial design was employed to optimize the CF treatment of the Timgad dam water. Coagulant (X 1 ), flocculant dose (X 2 ), and pH (X 3 ) were taken as input variables (  (1)): where α 0 is a constant, α i is the linear effect of the input factor X i , α ij is the linear-by-linear interaction effect between the input factors X i and X j , α ij is the quadratic effect of input factor X i , and ε is the error.
Responses included in this model are the removal efficiency (R%) in terms of COD and in terms of UV-254 nm of the treated samples.
Removal efficiencies were calculated according to Equation (2): where C 0 is the initial concentration, and C ec is the concentration in the treated simple.  overview of the water quality of Timgad's dam water and the temporal variations of different water quality indicators. The main result found by these authors is that the water samples were charged with DOM (average concentration around 14 mgO 2 /L) with a small variation as a function of environmental factors (COD ranged from 10 to 16 mgO 2 /L).
Therefore, in order to study the physicochemical characteristics of real untreated Timgad dam water and to carry out the coagulation/flocculation treatment, a water sample was collected from the Timgad dam according to Standard

Methods for the Examination of Water and Wastewater
(APHA ). The physicochemical characterization obtained is summarized in Table 3.
The analysis of the water sample collected from the Timgad dam (Table 3) showed that the dam water is charged with dissolved organic matter. Prior treatment should be considered to improve the quality of Timgad's dam water in order to satisfy the expectations of public authorities ( Décret exécutif nº - du  Rabie Ethani ) and most importantly to protect human health.

Screening of chemical reagents/flocculant
The results obtained during the first step are presented in the following section.

Effect of the nature and concentration of the coagulant
Adding coagulant chemicals having an opposite charge to those of pollutants present in the water contributes to the charge neutralization of non-settlable solids such as organic matter. Once neutralization of the charge is accrued, the small suspended particles are capable of sticking together and forming microflocs not visible to the naked eye. The rapid mixing is needed to disperse the coagulant effectively and promote particle collisions.
Colloidal dissolved organic matter in water has an electrical charge, which is generally negative. Since they all have the same charge, they repel each other and do not tend to form larger aggregates, and are therefore easier to remove.
The addition of coagulants makes it possible to destabilize the colloidal materials, and thus to allow them to agglomerate in order to easily eliminate them by decantation or by filtration. Among the five tested coagulants (Figures 2 and   3), aluminum sulphate Al 2 (SO 4 ) 3 and iron (III) chloride

).
It is important to mention that, at an initial pH (pH of raw water) equal to 6.56 (pH above 5), the higher the coagulant dosage used, the higher the dissolved organic matter removal obtained. In fact, in the case of aluminum sulfate, a higher dosage above 100 mg/L was required to enhance  The results obtained (Figure 4) showed that the addition of the flocculant at low concentrations, between 12.5 and     100 mg/L with Al 2 (SO 4 ) 3 as coagulant, improves the efficiency of the elimination of dissolved organic matter.
However, a significant improvement is observed from a concentration of flocculant of 50 mg/L. Beyond this concentration, the elimination yield remains stable, so this concentration of 50 mg/L is retained for the rest of the study.

Effect of the initial pH
The obtained results in Figure 5 show that the optimal pH for the highest elimination rates were obtained at pH in range of   Table 4. In general, the larger the magnitude of t and the smaller the value of p, the more significant is the corresponding coefficient term (Tripathi et al. ).
According to the sequential model sum of squares, the models were selected based on the highest-order polynomials where the additional terms were significant.
At the end of the coagulation process, the final empirical models in terms of coded factors after excluding the our obtained models were considered highly significant and indicated excellent correlations between the experimental results and the predicted values of COD and UV-254 removal taken from these models.  Overall, these findings on treated Timgad dam water quality and coagulation/flocculation treatment demonstrated that the optimizing method using RSM is a very effective technique, as it allows the saving of time, reduces chemical usage and allows the highest treatment efficiency to be reached. The obtained results can be useful for application in Timgad's treatment plant in order to provide a 'super treated water' with the lowest organic matter content.