Performance analysis of a half-batch multi-cell three-dimensional electrode reactor for drilling wastewater: COD removal, energy consumption and hydrodynamic characteristic

It is reported that China's oil production can reach 0.85 billion tons in 2022, which means that a large amount of drilling wastewater will be produced. These refractory organic pollutants in drilling wastewater are hardly to be effectively degraded by physical, physical-chemical, biological methods (Bin et al. 2013; Maslennikova & Ermakov 2020). Moreover, the application of these methods is constrained by the high cost, complex operation, and long processing time, which have led to scholars to search for new, easy and environment-friendly mechanisms for degrading these refractory pollutants (Shestopalov et al. 2019; Li et al. 2020).

Within the past decade, an electrochemical oxidation process based on advanced oxidation processes (AOPs) has been receiving greater attention due to its typical advantages such as environmental flexibility, compatibility, easy to realize automation and especially for the degradation of low concentration pollutants (Chen et al. 2021). The key to electrochemical oxidation process is the generation of active radicals with strong oxidation potential (e.g., hydroxyl radicals·OH, active chlorine species ACl and persulfate S₂O₈²⁻) during the electrolysis process (Ferrández-Gómez et al. 2021). In particular, electrochemical oxidation technology has been successfully applied to domestic wastewater, industrial wastewater, and medical wastewater (Pikaar et al. 2011; Zhenyu et al. 2020; Rahmani et al. 2021). Up to now, most of the work has focused on the development of new particle electrode materials, improving the property of electrode and the control of by-products in the electrochemical oxidation process (Hong et al. 2018; Meng et al. 2020; Salohiddinov et al. 2021; Wang et al. 2021). It is worth noting that electrochemical reactor is a device that converts electrical energy into chemical energy, which structure also affects the performance of electrochemical oxidation obviously. Furthermore, requirements for designing electrochemical reactors depend on the composition of the wastewater to be treated and of the target substances for elimination (Rivera et al. 2021). Körbahti used a continuous tubular reactor to treat textile wastewater and obtained good COD removal efficiency (Körbahti & Tanyolac 2009). The decolorization efficiency of over 90% in industrial wastewater was achieved by Sakalis using electrochemical reactor with parallel plate electrodes (Sakalis et al. 2005). Other types of electrochemical reactors have also been reported (Oduoza & Wragg et al. 2002; Wang et al. 2015). Among the various reactors, the parallel plate reactor is considered to be one of the simplest and easy-to-scale-up reactors for industrial application, which rising greater interest (Cano et al. 2016; Vidales et al. 2016). Unfortunately, the research on the reactor for electrochemical treatment of drilling wastewater has not been reported yet. And it is also worth noting that due to the complex composition of drilling wastewater, the disadvantages of a traditional single-cell parallel plate reactor, such as long reaction time and low current efficiency, impede its application. Recently, the performance of a multi-cell reactor has been confirmed to be better than the single-cell reactor by Ling (Ling et al. 2015).

Like all other chemical reactors, it is an alternative for electrochemical reactor to operate in different modes. The choice of batch, half-batch, recycle, or continuous operation for industrial application depends on the simplicity of the process required and economical efficiency. Until now, some modes of operation have been tested for the treatment of industrial effluent to understand hydrodynamic mixing differences and energy consumption, which indicates that compared with a continuous reactor, the batch reactor is considered as having the most potential for industrial application (Susree et al. 2013).

In this study, a multi-cell half-batch electrochemical reactor consisting of seven cells was proposed by us to treat drilling wastewater. Adjacent graphite plates are arranged in a zigzag pattern and also act as a separate baffle to realize a serpentine flow pattern in the reactor. Five operational parameters were examined with RSM to optimize the performance of the reactor via COD removal and energy consumption. Furthermore, the reactor's hydrodynamic characteristic was analyzed via the residence time distribution (RTD) method, and also gained to achieve equivalent function of a plug flow reactor.

An HH-WO 1 L-100 L constant temperature water bath magnetic stirrer, DDS-11A conductivity meter, ZD-X regulated power supply of direct current (Shanghai Lei Magnetic Co. Ltd); HH-6 chemical oxygen consumption tester (Jiangsu Instrument Co. Ltd); PHS-3E pH meter; KO-1 injector (Jiangsu Kangyou Co.Ltd); WQ-1A ammeter (Chengdu Suji Co.Ltd); LZB-6 flowmeter, LabV1 peristaltic pump, LP-20 air pump (Fujian Qixu Co.Ltd).

Potassium dichromate (K₂Cr₂O₇), sodium chloride (NaCl), sodium hydroxide (NaCl), sodium sulfate (Na₂SO₄), ammonia ferrous sulfate (Fe(NH₄)₂ (SO₄)₂), silver sulfate (Ag₂SO₄), sulfuric acid (H₂SO₄) were purchased from Chengdu Kelon Company. Granular activated carbon (GAC), glass beads, and graphite plates were purchased from Zhejiang Leqing Company, Sulfonated phenolic resin (SMP) was regarded as the characteristic pollutant in drilling wastewater, which was obtained from Daqing Oilfield.

The schematic diagram of the experiment setup is shown in Figure 1. The electrochemical reactor was made of plexiglass and its size was 25 cm × 15 cm × 8 cm, with an effective volume of 2 L. Graphite plates were used as the anode and cathode and their dimensions were 15 cm × 5 cm × 1 mm. They were positioned vertically and parallel to each other with a gap of 4.8 cm, and these graphite plates were connected to DC power. The synthetic drilling wastewater (initial COD: 580 ± 20 mg/L) in the reservoir was brought into the electrochemical reactor by the peristaltic pump, with a number of aeration vents arranged at the bottom of the reactor. An average set of experiments consumed 250 g GAC average size 3–5 mm, specific area ≈ 1,000 m² g⁻¹, porosity: 68.19%). Samples were collected at the outlet of the reactor every 20 minutes and the analysis of COD was implemented in duplicate, with results obtained as the average of the two (Ji et al. 1992) .All the solutions were prepared in double distilled water. All experiments were conducted at ambient temperature (25 ± 3 °C).

The degradation effect of wastewater in the early stage is obvious, and the maximum COD removal rate (97.49%) can be obtained at 100 min, and the COD removal rate does not increase after 100 min, which can be attributed to the weakening of the performance of mass transfer (Figure 3(a)). It is to be noted that the energy consumption was increased from 72.12 to 108.93 kWh/(kg COD⁻¹) with correspondingly the current efficiency decreasing from 27.61% to 6.12% for the increasing the time from 100 to 140 min. This indicates energy was wasted in hydrogen side reactions and oxygen side reactions.

The influence of current density was investigated in the range of 1.2 mA/cm²–11.2 mA/cm². The COD removal substantial increased with the increase of current density from 1.2 to 9.2 mA/cm². Further increase in current density up to 11.2 mA/cm² did not bring about any remarkable improvements (Figure 3(b)). This is because at higher current density, production of extra electrons may contribute to undesirable side-reactions such as oxygen evolution reaction. Moreover, the large amount of hydroxyl radical generated at high current density may not be sufficient utilized by organic matter, which owed to the limitation of mass transfer (Panizza & Cerisola 2005). We also observed that the temperature of the liquid phase in the reactor increased from 25 °C to 43 °C with increasing current density from 1.2 to 11.2 mA/cm², which can be interpreted as the result of an increase in power consumption from 28.16 to 98.14 kWh/kg COD. In addition, high current density will lead to shorten life of electrodes, which is undesirable.

The influence of various configurations of GAC on the performance of the reactor was ascertained (Figure 3(c)). Compared with the disorder configurations of GAC and GAC mixed with glass beads, the novel configurations of GAC (Figure 4(a) and 4(b)) significantly improved the performance of the reactor. In particular, the GAC vertical configuration (Figure 4(a)) can achieve more than 90% COD removal rate (Figure 3(c)). This is mainly because the short-circuit current can be completely eliminated, which results in current efficiency increased by more than 20% while reduced energy consumption by more than 50%. In addition, it was also observed that compared with the parallel configuration (Figure 4(a)), the energy consumption of GAC vertical configuration (Figure 4(b)) was reduced by 3.88 kWh(kg COD)⁻¹ correspondingly the current efficiency was increased by 3.27%. We concluded that the GAC vertical configuration can increase the average length of GAC, which is more likely to be motivated by electric field than the parallel configuration. As a result, the effective utilization efficiency of the electric field of the GAC vertical configuration is obviously better than that of the GAC parallel configuration (Figure 5).Compared with the hanging style three-dimensional electrodes, the novel configuration of GAC in this study is success (Zhuangzhuang et al. 2015). Besides, the purpose of mixing glass beads with GAC is to reduce short-circuit current, which is slightly useful. But it did not meet the economic requirements of industrial applications.

We observed that the reactor's capacity to remove COD increased substantially when the aeration rate from 0 (no aeration) to 2.67 L/min. However, when the aeration rate was further increased from 2.67 to 6.67 L/min, the COD removal decreased obviously (Figure 3(d)). Additionally, Figure 3(d) also showed that at the highest aeration rate of 6.67 L/min, the COD removal was only 74.08% after 100 min, compared to 97.39% at the aeration rate of 2.67 L/min. A possible reason was that amounts of organics exit the reactor without being completely degraded due to the fact that excessive aeration will lead to critical fluid short-circuiting in the reactor, which can lead to a decreased COD removal efficiency. It was also noted that the highest current efficiency of 34.16% and the minimum of energy consumption of 77.89 kWh(kg COD)⁻¹ were both obtained at an aeration rate of 2.67 L/min. These result indicates that excessive aeration may destroy the migration process of chloride ions, which may destroy the the concentration polarization interface layer of ACl near the anode. This is harmful to the efficiency of oxidation (Scialdone et al. 2009).

We noticed that the COD removal efficiency increased obviously up with the volumetric recirculation flow increased from 50 mL/min to 100 mL/min, above which a reduction in efficiency was observed with increased in volumetric recirculation flow (Figure 3(e)). We assert that certain volumetric recirculation flow will be conductive to take the degraded pollutants away from the electrode plates, which can avoid destroying the concentration polarization interface layer of the anode and lead to a higher organic degradation rate. (Scialdone et al. 2009) However, excessive volumetric recirculation flow will destroy the migration process of electrons in the wastewater to the cathode, being adverse to the efficiency of mass transfer. Hence, the energy consumption is increased; correspondingly, current efficiency is decreased with the volumetric recirculation flow increasing from 100 mL/min to 400 mL/min.

In Equations (10) and (11), coefficients with a positive sign represent synergistic effects, which means they strengthen COD removal and increase energy consumption respectively, while those with negative coefficients have the opposite effect on the response. So, electrolysis time and aeration rate in Equation (10) and current density, electrolysis time and volumetric recirculation flow in Equation (11) have a significant effect on the COD removal and energy consumption, respectively. The analysis of ANOVA and R² are necessary to determine the appropriacy of the quadratic model, which can be conducive to state the performance of the electrochemical reactor. (Tian et al. 2021).

Table S3 and Table S4 show the detailed ANOVA of results of COD removal rate and energy consumption. The R² of the COD removal model and energy consumption model are 0.9466 and 0.6770, respectively, which demonstrates that the experimental date fit into the two models well. Moreover, it is generally accepted that if the difference between the predicted R² value and the adjusted R² value is less than 0.2, then the fitted model can properly explain the relationship between factor and response. In our study, the differences between R² and adjusted R², which exist in the COD removal model and energy consumption model, respectively, are both less than 0.2. This supports the fact that the two models could be used for predicting the results with adequate precision. Both models have adequate precision of 13.587 and 13.541, respectively, indicating the probability of error due to noise is very small. The ratio of standard deviation to mean value is expressed by coefficient of variation (CV), which indicates the degree of data dispersion. CVs of 1.17% and 5.58% were obtained in the two models, respectively. Based on the above analysis, we conclude that the model fitting is successful (Rodrigues 2021).

According to the above optimized process, the optimal conditions for maximum COD removal efficiency and the lowest energy consumption were obtained with current density of 9.001 mA/cm², aeration rate of 2.67 L/min, circulation flow rate of 100 mL/min and electrolysis time of 99.675 min. The experimental values shown in Table 1 were obtained in optimization conditions and also compared with the values predicted by the model. The results show that the relative error of COD removal and energy consumption were less than 2 and 3%, respectively.

The RTD experiment was carried out under different operating conditions in the continuous flow reactor. Before the RTD experiment, a calibration curve between the concentration of NaCl and it's conductivity was obtained, which showed a linear relationship (R² = 0.998) (Figure S1).

RTD function curves under various operating conditions were scattered to separate graphs, which in order to investigate the influence of parameters on the hydraulic performance of the reactor in detail. As shown in Figures 6–8, it can be observed that the shapes of the RTD function curve with a long extended tail indicate a middle flow pattern between plug flow ideal and completely mixed regimes. Similar reports have also been reported by Li (Li et al. 2013). In addition, a pronounced tail in each condition indicated the existence of an obvious stagnant dead volume (Figures 6–8), for all operating conditions tested.

The E(t) curve for the flow rate of 50 min/L and 100 min/L in series peaked after approximately 25 min and 15 min, respectively, while the E(t) curve for the flow rate of 200 min/L peaked at approximately 8.5 min (Figure 6). Alejandro Regalado-Méndez found that as the flow rate increased from 0.5 mL/min to 1.8 mL/min, the average residence time of the liquid decreased by about 20 minutes (Regalado-Mendez et al. 2018). We conclude that the above difference is caused by the electrochemical reactor geometry, such as internal baffles, inlet/outlet location, and operation variables (flow rates, temperature, pressure). In addition, we found that for all the tested flow rates, the RTD curves with a long tail, which clearly indicates the existence of a dead volume. Hence, the shape of the curves depends on the flow rates. Regalado-Méndez et al. believe that high flow rates can cause a low axial dispersion coefficient and high Peclet number in the reactor (Regalado-Mendez et al. 2018). The conclusions have also been confirmed by Rivero et al.. Furthermore, it also can be seen that the curves of the peak were smaller and correspondingly wider, which obtained at the lower flow rate (50 mL/min). This result demonstrates the higher back-mixing levels at lower flow rate (Vacca et al. 2011).

Aeration has a positive effect on residence time, as can be observed in Figure 7. Compared with no aeration, the rapid increase in the E(t) curve for an aeration rate of 2.67 L/min curve followed by a slow but steady decline with a long tail, which indicates certain degree of fluid short-circuiting exists in the reactor. This is because of the agitation of the bubbles on the flow behavior of the reactor, forcing some of the tracer to exit the reactor in advance. Gresch et al. also demonstrated that the pattern of flow is sensitive to the location of aeration, the rising bubbles from the bottom of the reactor exert a greater drag on liquid, which leads the momentum and turbulence on liquid flow (Gresch et al. 2011). Figure 8 shows the relationship between the E(t) curve and time with and without current, and we noticed that the applied current will cause the residence time to advance. It is also observed that some curtains were formed near the graphite plate due to the rising bubbles, whose thickness slightly increases along the upward vertical direction. Part of the effective volume of reactor is occupied by the bubbles released during the electrolysis process, which in turn causes part of the liquid to reach the water outlet early. Based on the above analysis, we find that liquid flow rate has the most significant impact on residence time distribution, so it is necessary to conduct a systematic analysis of flow rate.

The dispersion number and variance are important indicators that reflect fluid flow patterns in hydraulic characteristics testing to reflect the degree of back mixing inside the reactor. Table 2 presents the calculated results of the dispersion number and variance for various liquid flow rate. For the flow rates of 50 and 100 mL/min, the calculated dispersion numbers were 0.197 and 0.214, respectively, indicating that the flow pattern of the reactor was of medium dispersion type, which approached the low dispersion type (Mansouri et al. 2012). Moreover, the dispersion number for the liquid flow rate of 200 mL/min was 0.267, which was a high dispersion type. Similarly, the decrease of dispersion number from 0.164 to 0.081 is the result of increasing flow rate from 30 LPH to 60 LPH, which was confirmed by R. Mythilishri (Mythilishri et al. 2020). We find that the configurations of the flow channel are important to investigate the flow patterns in the reactor. In this study, increasing flow rate can increase the dispersion number, which is connected with the reduction of void faction inside the serpentine channel. Because graphite plates can also act as a separate baffle to enhance turbulence (Rivera et al. 2015).

Stagnant zones/dead zones are usually found in reactors with corners, baffles, or obstructions (Tanimura et al. 2021). Another important indicator of electrochemical reactor performance is the percentage of dead volume. We also determined that the liquid flow rate had a substantial influence on the percentage of dead volume. We found that the liquid flow rate has a positive influence on the percentage of dead volume. As the liquid flow rate from 50 to 200 mL/min, the percentage of dead volume decreased from 5.36% to 3.41% (Table 3), which indicates that there is a good hydraulic performance that can be attributed to the increasing of flow rate. Nadji Bouakaz found that increasing the flow rate from 5 mL/s to 20 mL/s can slightly reduce the dead volume (Bouakaz et al. 2018). So we concluded the difference in hydrodynamic characteristics described above depends on the internal geometry. The results also confirmed slight fluid short-circuiting in the reactor, indicating that the tested electrochemical reactor had been successfully designed. It is worth noting that increasing flow rate is not a practical way for suppressing the percentage of dead volume, and a previous study indicated using turbulence promoters to eliminate the percentage of dead volume (Regalado-Mendez et al. 2018). Based on our previous study, we probably say that the continuous flow operation is better suited for field scale application due to its excellent advantages, such as its stability in effluent quality and larger throughout. However, the performance of plug-flow ideal has been confirmed to be more popular than continuous flow reactor owing to the low-percentage of dead volume and low dispersion number.

This work presented a novel half-batch multi-cell electrochemical reactor for drilling wastewater treatment. The operation ranges of the five parameters (electrolysis time, current density, the novel configuration of GAC, aeration rate, and recirculation flow rate) were investigated via uniparameter optimization, and further were systematically analyzed by RTD method, which to acquire the great performance that the maximum of COD removal (97.39%) and the minimum of energy consumption (77.89 kWh(kg COD)⁻¹). Hydraulic characteristic studies demonstrated that the electrochemical reactor is close to a plug-flow ideal reactor, and the rate of flow has a dominant effect on the type of flow. After comprehensive consideration of the COD removal and energy consumption, we concluded the flow rate of 100 mL/min, current density 9.2 mA/cm² and aeration rate of 2.67 L/min are accord with the desired effect. In addition, we believe that apart from looking for the design of electrochemical reactor, future research should pay attention to the performance of mass transfer in the electrochemical reactor.

This work was supported by the major national R&D projects of China (Grant No. 2016ZX05040003-004-002).

Manuscript is approved by all authors for publication

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

The authors declare that they have no competing interests

All relevant data are included in the paper or its Supplementary Information.