Abstract
A synthetic wastewater based on Algiers refinery real effluent was prepared and treated using anodic oxidation. Full factorial plan design was used to conduct the statistical analysis of the results. The aim of the study was to assess the interaction between current density (CD) and stirring degree (SD), and quantify their effects on chemical oxygen demand (COD) removal and electric energy specific consumption (EESC). With an initial COD of 487 mg/L, pH of 5.5 and 0.05 M of Na2SO4 as supporting electrolyte, it was found that a 55 rpm stirring degree variation led to a substantial gain in COD removal and energy consumption: 6% and 8.5 KWh/kg, respectively. Current density was found to have a different effect on removal efficiency within the applied stirring domain, and mass transport coefficient (km) is inversely correlated to energy consumption. A theoretical model describing the process was reviewed and the relation between concentration, hydrodynamics and applied current was emphasized.
HIGHLIGHTS
A simple empirical modeling to assess charge rate and mass transport rate balance (i.e. operating regime).
Lead anode was first applied for electro-oxidation of toluene.
Perspectives on the up-scaling approach taken and comparative data were discussed.
Theoretical modeling review.
INTRODUCTION
Fresh water quality and availability is a major concern worldwide; the increasing demand of this valuable resource is significant in the last years, due to human demographic and industrial growth. It is estimated that more than 40% of the global population will face water stress or scarcity within the next decades (Bartolomeu et al. 2018). In response to the increasing demand for energy and petrochemicals, where oil will account for the major world needs in the next decades, a huge amount of water is consumed by the oil processing industry. Refineries' liquid effluents are complex matrices of organic and inorganic pollutants; the quantitative and qualitative contamination depends on the unit configuration and the crude oil nature. The major pollutants found in the effluent are oil and greases, which are the source for other hazardous compounds such as phenols, BTEX, naphthenic acids (NA) and poly-aromatic hydrocarbons (PAH) (Fakhru'l-Razi et al. 2009; Aljuboury et al. 2017; Ali et al. 2018; Al-Khalid & El-Naas 2018; Singh & Shikha 2019).
Electrochemical treatment pathways have gained attention due to the possibility of treating a wide range of organic substances (Yan et al. 2014; García-Espinoza & Nacheva 2019). The main advantage to electrochemical treatment is the capability to convert pollutants into CO2 and water, rather than creating concentrated waste streams. However, this technique still suffers from high-energy consumption and production of critical intermediates. Compared to conventional methods that consume a substantial amount of chemicals, electrochemical techniques are regarded as green technology (Martínez-Huitle & Panizza 2018).
Anodic oxidation technology in particular presents mass transport polarization as the main cell potential inefficiency as it deals with low-concentration pollutants present in wastewater (Nava & Ponce de León 2018). As the relevant reactions require the direct contact between the electrode surface and the pollutant, maximum efficiency is therefore obtained when the electric charge transfer rate and the pollutant's mass transport rate are balanced (Martínez-Huitle et al. 2015). The amount of oxidative species generated is regulated by current density, whether the OH radicals are electro-generated at the electrode surface or the oxidizing agents in the bulk (Moreira et al. 2017). On the other hand, the targeted substances mass transport is correlated to the hydrodynamic regime (Scialdone 2009). However, the use of deflectors or turbulent promoters at the laboratory scale is not frequently encountered in the literature.
Theoretical modeling provides a reliable representation of the electrochemical oxidation process; it offers a better description of the competition between targeted reactive paths and unwanted reactions. Furthermore, the experimental validation of the mathematical models confirms the modeling assumptions, providing precious insights for the up-scaling phases. Several models therefore have been proposed, particularly in the field of electrochemical treatment of organic pollutants. Comninelis et al. proposed a model based on estimation of limiting current, developing a theoretical relationship between instantaneous current efficiency and operative parameters. The model assume mass transfer dependence; thus two behaviors are reported, oxidation control and mass transfer control. They considered the oxidation of water as the only competitive reaction and no homogeneous oxidation processes (Scialdone Galia & Randazzo 2012). Canizares et al. proposed an easy-to-use model with no adjustable parameters by representing the reactor as three interconnected stirred-tank reactors. The direct electrochemical process is modeled based on the applied current and the oxidation/reduction potentials of the organic substances involved. The model assumed that the concentration profiles are time dependent and not a function of position. Moreover, at low concentrations, the mass transfer coefficient depends only on flow rate (Cañizares et al. 2004). Polcaro et al. reported a model describing the evolution of organic concentration; the kinetic model assumes both the direct oxidation of organic substrates adsorbed at the anode surface and the indirect process in the bulk. The model was found to accurately reveal the effect of current density, linear velocity of electrolyte and anode surface to effluent volume ratio on the removal efficiency (Polcaro et al. 2002).
In this study, and from an energetic efficiency perspective, CD and SD were of particular interest as their interaction is expected to be significant in anodic oxidation. The limiting step and optimization of CD and SD, in respect to the anodic oxidation of refinery effluent, will be discussed and outlined.
MATERIALS AND METHODS
Materials
The synthetic effluent was prepared the same day the experiments were conducted, and its characteristics were based on Algiers refinery real effluent (Table 1). The characteristics of real wastewater were not the same in various times; a range of variation was reported. To prepare the synthetic effluent within the desired chemical oxygen demand (COD) range, 0.5 mL of toluene was added to each 1,000 mL of distilled water. The effluent was preserved in a water bath at 25 °C (toluene solubility in water at 25 °C is 0.535 g/L). Due to its environmental compatibility, sulfate sodium was added as a supporting electrolyte at 0.05 M. Although adding such chemicals is not of interest at the industrial scale, the possibility of enhancing the real effluent conductivity by a pre-concentration of the desalting stream makes this option worth the investigation (Fıl et al. 2014; Belkacem et al. 2017). The anode material was a commercial lead plate (75 × 25 × 3 mm).
Characteristics of Algiers refinery real effluent and the synthetic wastewater
. | Real effluent . | Synthetic effluent . |
---|---|---|
Temperature (°C) | 25–28 | 25 |
pH | 6.5 | 5.5 |
Conductivity (mS) | 3 | 11.8 |
COD (mg/L) | 500–600 | 487 |
. | Real effluent . | Synthetic effluent . |
---|---|---|
Temperature (°C) | 25–28 | 25 |
pH | 6.5 | 5.5 |
Conductivity (mS) | 3 | 11.8 |
COD (mg/L) | 500–600 | 487 |
Electrochemical reactor and experiments
The experiments were performed in a closed and undivided electrochemical reactor of 750 mL (Figure 1), in a galvanostatic mode and at room temperature (i.e. at 20 °C ± 2 °C). The reactor was filled with 400 mL of effluent, which is magnetically stirred to ensure mass transfer from and towards the electrode surface. The cell was equipped with lead anode and stainless steel cathode with active area of 15 cm² (partially immersed). The electrodes were fixed with an inter-electrode distance of 1 cm and were connected to a direct current power supply (AD-305D, AIDA, China). To determine COD removal, the samples were transferred to a spectrophotometer (Jenway 6305 UV/Vis) at 600 nm. Using standard methods (MA315-DCO-1.1), the COD was measured.
Schematic diagram of the experimental setup: (1) reactor, (2) anode, (3) cathode, (4) power supply, (5) magnetic stirrer.
Schematic diagram of the experimental setup: (1) reactor, (2) anode, (3) cathode, (4) power supply, (5) magnetic stirrer.
Each experiment used 400 mL of synthetic wastewater and the reaction time was 90 min in all experiments. Current density and stirring degree were the desired parameters to study, and the goal was to quantify their effect and interaction on COD removal and electric energy specific consumption (EESC) with insights on mass transfer coefficient (km).
Experimental design and statistical analysis
The experimental design was used to model the COD removal, mass transfer and energy consumption. To assess the relation between current and hydrodynamics, the effect of CD and SD and their interactions were investigated. Table 2 presents the input factors (CD and SD), their dimensions and their values.
Experimental domain and levels of factors
Factors . | Unit . | Symbol . | Study domain . | ||
---|---|---|---|---|---|
Low (−1) . | Middle (0) . | High (+1) . | |||
Current density | mA/cm² | CD | 26.7 | 30 | 33.3 |
Stirring degree | rpm | SD | 310 | 365 | 420 |
Factors . | Unit . | Symbol . | Study domain . | ||
---|---|---|---|---|---|
Low (−1) . | Middle (0) . | High (+1) . | |||
Current density | mA/cm² | CD | 26.7 | 30 | 33.3 |
Stirring degree | rpm | SD | 310 | 365 | 420 |
The design of experiments included six runs (four runs and two replicates at the central point). Table 3 presents the response values. Experiments were carried out randomly to minimize experimental errors.
Experimental matrix and responses
. | Factors . | Responses . | |||
---|---|---|---|---|---|
Run No. . | CD (mA/cm²) . | SD (rpm) . | COD removal (%) . | km (cm/s) . | EESC (KWh/kg) . |
01 | 26.7 | 310 | 36.35 | 0.000037 | 61.01 |
02 | 33.3 | 310 | 42.51 | 0.000045 | 69.75 |
03 | 30 | 365 | 46.61 | 0.000052 | 53.52 |
04 | 30 | 365 | 50.72 | 0.000058 | 49.19 |
05 | 26.7 | 420 | 52.77 | 0.000062 | 39.12 |
06 | 33.3 | 420 | 50.10 | 0.000057 | 57.63 |
. | Factors . | Responses . | |||
---|---|---|---|---|---|
Run No. . | CD (mA/cm²) . | SD (rpm) . | COD removal (%) . | km (cm/s) . | EESC (KWh/kg) . |
01 | 26.7 | 310 | 36.35 | 0.000037 | 61.01 |
02 | 33.3 | 310 | 42.51 | 0.000045 | 69.75 |
03 | 30 | 365 | 46.61 | 0.000052 | 53.52 |
04 | 30 | 365 | 50.72 | 0.000058 | 49.19 |
05 | 26.7 | 420 | 52.77 | 0.000062 | 39.12 |
06 | 33.3 | 420 | 50.10 | 0.000057 | 57.63 |
RESULTS AND DISCUSSION
Designing experiments and analyzing the variance of responses
Figure 2 shows the predicted values of the models in comparison with the observed values. These results display a good agreement between predicted and experimental values.
Table 4 presents the results of ANOVA test for coefficients assessment. With a level of confidence of 95% and degrees of freedom 6−4 = 2, Ti was compared to the tabulated Tcri = 4.303 to assess coefficient significance.
Analysis of variance (ANOVA) and Student's t test for presented models
Model . | COD removal . | km . | ESSC . | |
---|---|---|---|---|
Variance of residuals (S²) | 11.1890625 | 2.25 × 10−11 | 25.0192292 | |
Variance of common effects (Si²) | 1.86484375 | 3.75 × 10−12 | 4.16987153 | |
Ti | a0 | 34.0584685 | 26.8526845 | 26.9519868 |
a1 | 0.63891666 | 0.51639778 | 3.33614699 | |
a2 | 4.39552692 | 4.64758002 | 4.1637563 | |
a12 | 1.61651406 | 1.80739223 | 1.19611582 |
Model . | COD removal . | km . | ESSC . | |
---|---|---|---|---|
Variance of residuals (S²) | 11.1890625 | 2.25 × 10−11 | 25.0192292 | |
Variance of common effects (Si²) | 1.86484375 | 3.75 × 10−12 | 4.16987153 | |
Ti | a0 | 34.0584685 | 26.8526845 | 26.9519868 |
a1 | 0.63891666 | 0.51639778 | 3.33614699 | |
a2 | 4.39552692 | 4.64758002 | 4.1637563 | |
a12 | 1.61651406 | 1.80739223 | 1.19611582 |
As shown, the calculated Ti for the SD coefficient is significant for the modeled responses (Ti > Tcri). However, the binary interaction between the variables was found to be non-significant. Similarly, the test reveals a higher significance effect of CD coefficient on energy consumption compared to the other responses.
Charge rate and mass transfer rate interaction

On the other hand, Figure 4 indicates a significant decrease in EESC over the SD domain when operating at 26.7 mA/cm², with 10.5 KWh/kg gain for one order of magnitude (55 rpm). Simultaneously, SD has a remarkable effect on COD removal, around 8.2% removal gain, as shown in Figure 3. Although, these responses reflect higher current efficiency when increasing SD and lowering CD, operating at these conditions is not favorable as it slows the reaction kinetics, thus affecting the overall process efficiency. Therefore, balancing the reaction kinetics and the removal efficiency, for a given electrochemical cell, by manipulating the cell hydrodynamics and the applied current is crucial for process optimization (Kapałka et al. 2008).
Optimization of operating parameters


From an energy efficiency perspective, and for comparison's sake, reporting the characteristics of the electrochemical cell in details is crucial, since the mass transport coefficient is correlated to both flow rate and cell design. Alongside the conventional parameters reported in the major studies, concrete insights about the limitations of an adapted treatment are then accessible. Reporting these data in future studies is expected to significantly contribute to the electrochemical cell design adaptation towards large-scale applications.
Mass transport and process performance
As poor hydrodynamic conditions can affect substantially the action of the current density, monitoring species mass transfer coefficient is of high interest. In addition to mass transfer enhancement, proper hydrodynamics contribute to the overall process efficiency, through the applied shear stresses, by lowering the film layer thickness at the catalytic surface (Fıl et al. 2014), thus reducing the effect of electrode poisoning. Figure 5 shows the plot of COD removal, EESC and km in term of CD and SD. Mass transfer coefficient was found to be inversely correlated to EESC. Beyond the studied domain, it was found that around 62% COD removal can be achieved at 26.7 mA/cm² (−1) and 475 rpm (2), with only 26 KWh/kg. It is explained by enhancing the rate at which targeted contaminants collide with the electrode surface, which leads to favoring the desired reactions. However, this effect occurs up to a threshold above which no significant change is observed. At this point, the process is under current limitation and further removals require higher current values. Species concentration is another factor that is found to affect this balance. Trends where degradation efficiency is higher at early stages were reported (Ibrahim 2013; Jawad & Abbar 2019).

It can be seen that current efficiency is a function of applied current, hydrodynamic regime (as it influences the mass transfer coefficient) and the substrate concentration (with a linear dependence). Interestingly, the current efficiency is independent of the reaction mechanisms when operating under mass transport control.
As electrochemical technologies deal with low-concentration effluents, it is clear that mass transport enhancement in continuous reactors is vital to the up-scaling of this technology.
CONCLUSION
In this study, the relation between current density and stirring degree in anodic oxidation of synthetic wastewater was investigated. A full factorial plan was used to design experiments. The effects of the factors and their interactions on COD removal, EESC and mass transport were quantified and the results were analyzed using ANOVA. It was found that the stirring degree was significant and effective on all responses within the studied domain. A 55 rpm variation in stirring degree was found to affect COD removal and EESC by a gain of 6% and 8.5 KWh/kg, respectively. The highest COD removal and minimum energy consumption obtained by the presented model, computed through the desirability function, were 53.85% and 37.28 KWh/kg at 420 rpm and 26.7 mA/cm².
Beyond reporting trends, the balance between charge rate and mass transport rate was emphasized, as it offers precious insights on how to align cell design and regulation with proper feeds. Reporting similar data in future studies will conveniently reflect the system limitations and, eventually, allow concrete adjustments towards large applications. Finally, the necessity of considering new comparative concepts was discussed.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.