Wash water from the secondary centrifuge generated during the purification of virgin olive oil (two-phase manufacturing system) was collected from the Experimental Olive Oil Factory located in the “Instituto de la Grasa (CSIC)” of Sevilla, Spain. Prior to its characterization and use, olive oil washing water was preserved under freeze conditions (-4 ºC) to avoid undesirable fermentation processes. The main characteristics of the used olive oil washing water are shown in Table 1.

The BES consisted of two Plexiglas plates with a single-flow channel, two electrodes, and two Plexiglas support plates (Figure 1). The two Plexiglas plates with a flow channel were separated by a cation exchange membrane (Fumasep FKB, Fumatech, St. Ingbert, Germany). The other side of the flow channel faced the electrode. The anode and cathode electrodes were made of flat graphite (MR200, gas tight impregnated, from Müller & Rössner GmbH & Co., Troisdorf, Germany). The surface area of the flow channel and thus, the projected surface area of the electrodes in contact with the solution was 22 cm². The volume of the flow channel was 33 mL (11.2 cm long x 2.0 cm width x 1.5 cm high) in each the cathode and anode compartment. The voltage of the system was acquired every 10 minutes via a data acquisition card (NI USB-6009) connected to a computer. The BES was kept at a temperature of 35 ºC. The anode compartment of the BES was connected in a closed system to a 1-liter bottle, where inoculum and used substrate were placed. The cathode compartment of the BES was connected in a closed system to a 1-liter bottle, where the catholyte was placed.

The inoculum placed in the anode was obtained from an industrial anaerobic reactor (UASB) treating brewery wastewater and operating at mesophilic conditions (35 ºC). The main characteristics of this anaerobic inoculum were: pH: 7.5; Total Solids (TS): 69 g/L; and Volatile Solids (VS): 45 g/L. The BES was run in batch mode and substrate was batch-fed to the anode every time the current and the power densities were almost zero. During the first 36 days of operation, The BES was fed with acetate in the anode compartment. The catholyte was composed of a phosphate buffer solution and flushed with air for 36 days. On day 36 of operation, the catholyte was replaced by a solution of Fe(III)[CN]₆ ³⁻ (0.050M) and phosphate buffer (reduction of Fe(III)[CN]₆ ³⁻ to Fe(II)[CN]₆ ⁴⁻). From day 36 to day 54, acetate was still fed to the anode compartment. On day 54, acetate was no longer fed to the anode compartment but olive oil washing water till the end of the experiment at day 63.

The current density I (A/m²) was calculated from the measured voltage potential (V) of the BES, the used resistance (R), whose value was 500 Ω (ohms) throughout the experiment and the projected surface area of the electrode (A_elec ) by using equation (1). Power density E (W/m ² ) was calculated by using equation (2):

The Coulombic Efficiency (CE) was described as the percentage of electrons recovered from organic matter versus the theoretical maximum whereby all electrons are used for electricity production (Tee et al., 2017). Therefore, CE was calculated by using equation 3:

where Ca is the total coulombs calculated by integrating the current over time, and Ct is the theoretical amount of coulombs available from the oxidation of olive oil washing water.

All chemical analyses were performed according to the Standard Methods of APHA (APHA, 1998). The following parameters were analyzed: Chemical Oxygen Demand (COD), Soluble Chemical Oxygen Demand (sCOD) (method 5220D), Total Solids (TS), Fixed Solids (FS), and Volatile Solids (VS). pH and alkalinity were determined by using a pH-meter model Crison 20 Basic. Total and soluble phenols were quantified by spectrophotometry through the Folin-Ciocalteu method with a calibration curve of gallic acid (García et al., 2016).

During the start-up, BES was fed with 1 g/L of acetate each 4-day period with the aim of promoting the formation of the microbial community in the anode. Figure 2 shows the voltage potential throughout the experimental time and the different operational conditions. During the first 35-day period, the voltage potential remained constant at a value of +41.4 ± 6.3 mV. At day 35, voltage potential increased up to +210.3 ± 13.7 mV (Figure 2). This potential was reached by using an air fed cathode, with oxygen as the electron acceptor. The increase in the voltage potential indicated the development of the microbial community in the anode.

At day 36, the air-fed cathode was substituted by a ferric/ferrous iron cathode. The substitution was carried out because a ferric/ferrous iron cathode allows for obtaining a higher standard potential than the air cathode, i.e. +770 mV vs. a normal hydrogen electrode (Uría et al., 2012). This kind of cathode presents other advantages compared to the air cathode such as fast reaction at carbon electrodes (Uría et al., 2012, Ter Heijne et al., 2006).

As can be seen in Figure 2, the substitution of the air cathode by a ferric/ferrous iron cathode resulted in a rapid increase in the voltage potential, which reached a maximum potential of +680 mV. This value was very similar to the standard potential for the ferric/ferrous iron cathode, indicating that the microbial film in the BES was properly developed. From day 36 to day 54, each feed batch of acetate resulted in an instantaneous increase in the voltage potential until achieving a value close to +680 mV. After the initial increase, the voltage potential slightly decreased during the 2-day period. A drastic drop in the voltage potential was observed around 4 days after the feed batch, which decreased to +40 mV (Figure 2). This decrease was a consequence of the substrate limitation in the BES, given that the acetate is a readily digestible substrate which requires a very short time for its degradation. Voltage potential was recovered to the reference value after each feed to the BES. After a 54-day period, the behavior of the system was kept under stable and reproducible conditions, and the reference value of the voltage potential was kept virtually constant (Figure 2).

The electrochemical active anode was fed with olive oil washing water on days 54 and 61. This substrate allowed for reaching a maximum voltage potential of +520 mV, i.e. 76.4% of the maximum voltage potential determined for acetate degradation (Figure 2). However, voltage potential reached lower values after the subsequent load of olive oil washing water on day 61, i.e. around +390 mV (Figure 2). After the addition of olive oil washing water at day 54, COD decreased during the time from 2600 ± 50 mg O₂/L on day 54 to 1535 ± 50 mg O₂/L on day 61 (41% COD removal). During these 7 days, two different periods can be observed. One period from day 54 to day 57, where current and power density were stable (Figure 3A) and COD was effectively degraded (Figure 3B). In the second period, from day 57 to day 61, COD gradually decayed (Figure 3B), while current and power density decreased greatly (Figure 3A). At days 54-57, most likely only the most readily digestible organic matter was degraded by the microorganisms present in the anode supporting electricity production. While at the second period, the most recalcitrant organic matter did not support electricity production.

A maximum COD removal of 43%, very similar to that achieved in the present work (41%), was obtained by Sonowane et al. (2013) in a multi-electrode MFC (multiple anodes acting as a baffle) using diluted distillery wastewater (pH: 7.8; COD: 2303 mg/L; total solids: 35.2 mg/L) in three batches at ambient conditions. Maximum COD and Biological Oxygen Demand (BOD₅) removal efficiencies of 65% and 50% respectively were reported in a single-chamber air cathode BES with platinum anode treating classical olive mill wastewater diluted with domestic wastewater at the ratio of 1:14 (COD: 4300 mg/L). DNA-fingerprinting showed the high bacterial diversity and the presence of exoelectrogenic bacteria on anodes, such as Geobacter spp (Sciarria et al., 2013).

In the present study, a feed batch cycle of 4 days was enough for acetate degradation, however, only 41% COD degradation of olive oil washing water was achieved after a 7-day cycle. Maximum dissolved organic carbon (DOC) removal of 90% in feed batch cycles of 15 days was achieved in a dual chambered BES operated in batch mode with platinum and mixed metal oxide titanium (Ti-TiO₂) with initial DOC values of 3600 mg/L. A decrease in feed batch cycles from 5 to 15 days adversely affected BES performance (Cirik, 2014). In the same way, a COD removal efficiency of 79% was reported in a dual chambered BES operating with wastewater with sulfate at a COD/sulfate ratio of 0.8 and feed batch cycles of 2 days (Ghangrekar et al., 2010).

Low power densities are usually reported when real wastewaters or severe environmental conditions are used. Lower power density values (65.82 mW/m²) than that achieved in the present research (Figure 3A) were reported by Zhang et al. (2013) using diluted molasses in a two-chamber BES with an external resistance of 1000 Ω. Maximum power densities as low as 17.6 mW/m² were reported in two-chamber BES treating mining process wastewaters containing tetrathionate at pH below 2.5 with ferric iron as the terminal cathodic electron acceptor (Sulonen et al., 2014). On the contrary, maximum power density values as high as 1180 mW/m² were achieved in MFC treating wastewater produced in a bio-refinery process (COD: 5.3 g/L) containing residual sugars, 5-furfural, phenolics, and other pre-treatment and fermentation by-products (Borole et al., 2013). In this case, power density increased with loading reaching the above-mentioned maximum value at COD of 5.3 g/L (8% dilution) but decreased thereafter. It was demonstrated that excessive loading led to poor electrogenic performance (Borole et al., 2013).

Figure 3C shows the variation in the CE percentage during the experimental time, ranging from 3.0% to 0.5%. These results are consistent with those obtained by Koók et al. (2016) and Capodaglio et al. (2015) treating the liquid fraction of pressed municipal solid waste and urban wastewaters, respectively. However, other authors, using synthetic waters, obtained higher percentages (Aelterman et al., 2008; Sleutels et al., 2009). The differences in the CE values in the reported studies can be attributed to the reactor configuration, source of inoculum, external resistance used, etc. The low CE is a general issue in BES using real wastewater. The presence of inhibitory compounds, such as phenols, limits the COD removal as observed in the residual COD of the present study (Figure 3B). These phenolic compounds are known to show antimicrobial properties (Acar et al., 1992; Khoufi et al., 2004; Chen et al., 2010; Mohamed et al., 2010).