At the beginning, the electro-coagulation treatment was designed to treat directly raw printing wastewater. This succeeded for wastewater S1 and S3 but did not work for wastewater S2. For raw S2 samples, applying EC treatment resulted in unstable electric currents with no visible change in color after 30 min of EC treatment. Also, very less sludge was generated and floated. Hence, different dilution factors were tested for these samples. Table 3 shows the changes in COD and color of the S2-1, S2-2, and S2-3 wastewater before and after treatment at different dilution factors.

For S2-1 and S2-3, applying EC treatment for samples diluted at the factors (f) of 2 and 3 still resulted in no visible change of color after 30 min of treatment. This is because of too high COD and color in raw wastewater (Table 2) as well as the diluted wastewater. When f was equal to or greater than 4, change in color of the treated wastewater was observed.

The results from Table 3 also suggest that EC treatment seemed to work obviously for the wastewater with COD about 10000 mg/L. No more dilution was required for the wastewater with COD lower than this recommended value. It is noted that in the case of S2-2, we selected the dilution factor f = 3 instead of f = 2 since we desired COD in EC treated wastewater lower than 1000 mg/L to facilitate the further treatment step of the effluents. Our approach in selecting dilution differed from previous studies using EC treatment without dilution for paint manufacturing wastewater (COD = 19700 mg/L [6]) and lithographic wastewater (TOC = 18400 mg/L [7]). The treated wastewater from these studies still contained very high in COD (about 1182 mg/L) and in TOC (3825 mg/L), which could be still hard for selecting the next step of treatment, especially for biological treatment. For the real industrial scale, we suggested designing the system with the recycled line of treated wastewater for the dilution instead of using tap water as in this study.

Variation in pH affects solution conductivity, electrodes’ dissolution, forms of aluminum species, and zeta potential of colloidal particles [5]. In this study, the effect of pH from 5 – 9 was examined and the results are shown in Fig. 2. For S1, changing pH from 5 – 9 varied slightly the removal efficiency of COD from 84.7 to 87.8%, and color from 99.4 to 99.9%, suggesting the optimum pH of 6. For S2-1 and S3, increasing pH affected strongly on COD and color removal efficiency, which firstly increased and then reduced. Maximum COD and color removal efficiency was 89.7 and 96.3%, respectively, for S2-1 at pH 6, and 89.3 and 99.9%, respectively, for S3 at pH 7. For all three types of wastewater, COD removal efficiencies in this study at optimum pH were clearly higher compared to those provided by electrocoagulation treatment in Papadopoulos et al. [10] (72.0–85.8%). Also, the results suggested that pH ~ 6 – 7 was the optimum range for the treatment of COD and color in printing wastewater, possibly because of the dependence of the forms of aluminum species on pH. It is well known that Al³⁺ ions generated from anode during electrocoagulation combine with water and hydroxyl ions to form monomeric, polymeric, amorphous and less soluble species of Al [14]. Since the solubility of aluminum hydroxide Al(OH)₃ precipitate is minimum at pH of 6.0 [15], large amounts of these precipitates were formed leading to the rapid adsorption of soluble organic compounds and trapping of colloidal particles [16]. However, due to aluminum amphoteric character, Al(OH)₃ solubility increases as the wastewater becomes either more acidic or alkaline leading to the increase of Al soluble forms (e.g. Al³⁺, Al(OH)²⁺ or Al(OH)₄⁻), resulted in the decreases of COD and color removal efficiency. Our approach in the adjustment of pH to optimum value differed from Adamovic et al. [8] where the EC treatment was directly performed at initial pH (11 – 12) for printing wastewater without the evaluation of pH effect. Optimum pH found in this study is similar to the values finding from Bensadok et al. [14] for COD and turbidity removal in cutting oil emulsions and close to the optimum pH (5.7) found by Valero et al. [12] for TOC removal in almond industry wastewater. However, these pH values differed from Suárez-Escobar et al. [7] who found optimum pH 8.23 for lithographic wastewater, and from Akyol [6] with optimum pH from 4 – 8 for paint manufacturing wastewater. Accordingly, no pH adjustment was required for S1 and S3 whose initial pH values were in the range of 6 – 7, but slight acidification was needed for S2 with initial pH about 8 – 9.

The distance between the two electrodes affects the electrocoagulation process as the electrostatic field depends on the distance of the electrodes [17]. Optimum electrode distance in the EC treatment varied from different studies, e.g. 0.1 cm [18], 0.5 – 1.0 cm [8], or 2.8 cm [19]. In this study, the effect of electrode distance (1 – 5 cm) on COD and color removal was tested. As can be seen from Fig. 3, electrode distance of 1 and 2 cm provided similar COD in effluents and the best color removal was found for the distance of 2 cm. Further broadening electrodes gap (>2 cm) resulted in lower removal efficiencies of both COD and color which would be due to: (i) the decrease of current density at a fixed voltage, e.g. current densities were 263.74, 219.78, 142.86, and 87.91 corresponding to the distance of 1, 2, 3, and 4 cm for S2-2. Similarly, current densities were 43.96, 32.97, 21.98, 16.48, and 10.99, corresponding to the distance of 1, 2, 3, 4, and 5 cm for S1. Current density reduction with the increase of electrode distance was also observed by Phan et al. [20], and further mentioned by Thella et al. [21] as a consequence of resistance increase; (ii) the decrease in the electrostatic attraction that caused a slower movement of the generated ions [17] leading to the lower removal efficiency of pollutants since Fekete et al. [22] found that the dissolved Al from anode being active as a coagulant for seconds only after its formation; and (iii) the diminution of mass transfer in the reactor [5].

On the other hand, a too narrow distance between two electrodes is also not recommended because metal hydroxides generated get degraded by collision with each other [23]. Thella et al. [21] also preferred not to apply too short distances to avoid short-circuiting and Sahu et al. [5] mentioned the frequent floc deposits induced by the low level of turbulence between the electrodes. Hence, we selected a distance of 2 cm instead of 1 cm for the next experiments to facilitate the floating of sludge, its harvesting and the convenience in cleaning electrodes.

Adding salt into the EC system can increase the conductivity of wastewater, reduce adverse effect of other anions (HCO₃⁻, SO₄²⁻), decrease in power consumption, and provide additional disinfection [24]. In this study, electrolyte concentration in S1 and S3 was varied by adding different NaCl dosages, i.e. 0, 0.05, 0.1, 0.2, 0.3, and 0.4 g/L. For S1, the effect of adding salt had an unclear impact on COD removal efficiency, ranged from 84 – 88%. In a previous study, Valero et al. [12] also found the removal efficiency of TOC was unchanged (i.e. 80%) between adding or non-adding of NaCl. We further observed a slight decrease in COD removal efficiencies for the treatment of S3 with the increase of salt concentration added, within 91 – 95%. Hence, the addition of salt may not be necessary for printing wastewater, possibly due to the sufficient ion existing in the raw wastewater.

Many studies apply mixing at different speeds, e.g. 200 rpm [25], 250 rpm [14], 450 rpm [8], and 900 rpm [7] to make the reaction solution homogeneous. In this study, mixing at different speeds (i.e., 0, 100, 200, 300, and 400 rpm) was applied for EC treatment of S3. Interestingly, mixing led to a slight reduction in COD removal compared to no mixing, and provided no effect on color removal. The result from this study is in agreement with Phan et al. [20] who achieved the highest COD and color removal efficiency in coffee wastewater without the addition of agitation. However, it differed with the results from several studies where the optimum speed was 300 rpm for cadmium removal [26], 150 rpm for arsenic removal [27], COD removal in poultry slaughterhouse wastewater [28], and boron removal [29]. In our study, the increase of COD in the effluent when the EC treatment applied stirring would be due to the breaking of flocs, and some floating sludge returning back to the solution under stirring. It is further noted that bubbles generated at two electrodes already assisted the mixing in the solution. Since the effect of stirring speeds has not been tested for wastewater S1 and S2, further investigation is needed for this factor with a broader type of printing wastewater. For the next experiments, we decided not to apply stirring in the electrocoagulation reactor.

According to the result from laboratory scale, ratio between electrodes area/wastewater volume (S/V) was 2×91 (cm²)/1 (L) = 18.2 (m²/m³). For the pilot scale, we tended to keep the value close to this ratio, by applying 6 electrodes with monopolar arrangement (Fig. D in the Supplementary data), resulted in the ratio of [(6×720) (cm)²/240 (L) = 18.0 (m²/m³). This ratio was a few higher than the value selected by Khaled et al. [26] (i.e. 13.6 m²/m³). In addition, the effect of the number of electrodes from 2 to 8 on COD and color removal was investigated for S1. The current intensity was also kept at optimum condition found from laboratory scale, i.e. 33 A/m² for individual unit.

As can be seen from Fig. 4, the increase in the number of electrode improved the removal of both COD and color because of the enlarging active surface [26]. This trend was similar as the findings from Akbal and Camcı [30] for electrode number increasing from 2 to 6 in the removal of Cu, Cr, and Ni, and from Ahmadian et al. [31] with electrode number from 4 to 8 to treat slaughterhouse wastewater. In term of S/V ratio, our results reinforced that increase in S/V ratio by increase of electrode number led to the increase in removal efficiency, which is consistent with the trend found by Khaled et al. [26] where much better removal efficiency of cadmium was achieved with larger S/V ratio. It is noted that this study changed the total electrical current applied for the system to keep the current density of each unit at 33 A/m² as the optimum value found from the laboratory experiment. Hence, when the number of electrode increased, the total current required increased, resulted in the increase in power consumption. Since the improvement of removal efficiency was more obvious when the number of electrodes increased from 4 to 6 (i.e. 7.2% for COD and 1.0% for color) compared from 6 to 8 (i.e. 4.0% for COD and 0.1% for color), 6 electrodes were selected as appropriate number.

It is obvious that performing the treatment in the laboratory scale where most of the conditions are well controlled will result in better removal efficiency as compared to the treatment on the pilot scale. As demonstrated in Fig. 5, removal efficiencies at the pilot scale were 81.9 % COD and 98.6% color for S1 and 88.9% COD and 95.8% color for S2-2. These efficiencies were lower than those achieved from laboratory scale but the difference in efficiency was small, i.e. 2.6% for COD and 0.9% for color of S1, and 1.4% for COD and 1.5% for color of S2-2. The images of feed wastewater and effluents after lab and pilot-scale treatment can be found in Fig. E in the Supplementary data.

Selection of batch or continuous operation modes depends on the discharging regime of the source as well as the volume of the printing wastewater. While the batch mode is suitable for the wastewater discharged intermittently in low volume, the continuous mode is preferred for the case of wastewater discharged continuously with a larger volume. In this study, we examined the performance of both operation modes using S1 and S2-2 samples, and the results are given in Fig. 6. Obviously, compared to continue mode operation, batch mode provided a slightly better COD and color removal efficiency. The difference in efficiency between batch and continue modes were 1.5% COD and 0.62% color for S1, and 2.07% COD and 1.52% color for S2.

To test the stability of the wastewater treatment system in the pilot scale, experiments were conducted using continuous mode with 8 h/day for 13 days using in-situ S1 wastewater. As can be seen from Fig. 7, though COD in the raw wastewater varied strongly during 13 days, from 1600 to 3360 mg/L, COD in the effluent was quite stable in the range of 302 – 544 mg/L. Similarly, the color in raw wastewater was from 6060 to 9619 Pt-Co, resulted in color of 51 – 123 Pt-Co after EC treatment. This relatively stable outlet concentration of COD and color despite the high variation in the inlet suggests EC as a very potential technology for the treatment of printing wastewater in practical application.

Table 5 shows the characteristics of treated wastewater (S1, S2-3) at the pilot scale and the standard for industrial wastewater discharging. Obviously, the values of all parameters were significantly reduced. Though pH, TSS, total nitrogen met the standard, COD and color (for S2-3) were still higher than the allowable values. The remaining COD, color, and BOD in treated wastewater possibly existed in the dissolved forms which could not be effectively removed by electro-coagulation. Similarly, previous studies also using electrocoagulation for paint manufacturing wastewater [6] and lithographic wastewater [7], printing ink wastewater [10], resulted in the treated wastewater still high in COD (1182 mg/L), TOC (3825 mg/L), and COD (≥2000 mg/L), respectively. Hence, the electrocoagulation treatment is suitable as an effective pre-treatment for printing wastewater, and an advanced treatment process is needed to remove the remaining contaminants.