Studies on Adsorption Characteristics of Corn Cobs Activated Carbon for the Removal of Oil and Grease from Oil Refinery Desalter Effluent in a Downflow Fixed Bed Adsorption Equipment

The discharge of oil and grease (O&G) containing effluent without treatment may contaminate the aquatic environs and freshwater. The removal of O&G from simulated refinery desalter effluent (SRDE) by activated carbon (AC) originated from chemical activation/carbonization of corn cobs (CCs) was investigated through fixed-bed column studies. The corn cobs activated carbon (CCAC) was characterized to determine its physicochemical properties, and the functional groups presently active on it partaking in the column adsorption process. The CCAC size (150, 300 and 600 µm), initial adsorbate concentration (200, 300 and 400 mg/L), and bed height (100, 200 and 300 mm) were varied to observe their influence on the adsorption of O&G and breakthrough time ( τ ) at a constant flow rate of 10.5 mL/min in a 10 mm diameter column of length: 60 mm. The removal of O&G from SRDE was inspected using the Bohart-Adams (B-A) and Yoon-and-Nelson (Y-N) kinetic models. Breakthrough time and %O&G removal decreased with increasing CCAC particle size and feed concentration and improved with rising bed height (BH). The void fractions ( ε ) at BHs of 100, 200 and 300 mm were 0.0247, 0.0124 and 0.0082, respectively. The ideal residence time ( 𝑡 𝑅 ) was 4.49 min. The B-A model yielded the highest degree of fit to the data than the Y-N model with R 2 within 0.8217 and 0.9771. This means that the B-A model can be used to predict the breakthrough curve of any desired values for the present study. This work also revealed that CCs could be packed in a fixed-bed column for O&G reduction from refinery desalter effluent.


INTRODUCTION
The permissible level for oil and grease (O&G) by WHO in refinery effluent is 10 mg/L (Umembamalu et al., 2020). These contaminants may appear in forms such as emulsified oil, free oil, or like a coating or suspended matter (Al-Malack and Siddique, 2013). Oil and grease is used to sum up the total hydrocarbon (petroleum and its derivatives) pollutants such as naphthenic, aromatic, and paraffinic hydrocarbons pollutants present in water (ABNT, 2003;Boni et al., 2016). Usually, crude refining is the preliminary step in the pretreatment process which aims at subjecting the crude oil to a series of treatments to reduce the contaminants to acceptable levels. In other to remove all these contaminants, the following petroleum pretreatment processes are adopted: crude oil desalting, crude heating, desulphurization, and pre-flashing. The process of crude oil desalting in the refinery will produce liquid effluent that contains oil, grease, salts, mud, and other impurities (Afshin and Toraj, 2008). This oily waste discharge is responsible for harmful effects such as objectionable odours, undesirable appearance, and flammability on the surface of the receiving water, thus leading to a potential safety hazard. Most importantly, the presence of O&G in water bodies constitutes a major threat to aquatic life as it consumes dissolved oxygen necessary for aquatic life survival (Islam et al., 2013), and in greater quantities, it limits oxygen transfer (Alade et al., 2011;Facchin et al., 2013). Water pollution has also resulted in diseases and deaths globally (Yalcinkaya et al., 2020;. This work seeks to solve this challenge through the aid of an adsorption column. Nowadays, there are lots of technologies or methods used successfully for the removal of O&G, they include conventional coagulation, electrocoagulation, membrane distillation, adsorption, filtration, etc. (Adams et al., 2017;Ariana et al., 2016;Diraki et al., 2019;Kulkarni, 2016;Kulowiec, 1979;Manilal et al., 2020;Mazumder and Mukherjee, 2011;Pintor et al., 2016;Rahmat et al., 2018;Umembamalu et al., 2020;Yalcinkaya et al., 2020). The choice of adsorption for this research is due to its effective and economical nature regarding the usage of agricultural derived biomass as adsorbents with neither little nor economic value (Bharathi and Ramesh, 2013;Choi, 2019;Goel et al., 2005;Onyechi, 2014;Umembamalu et al., 2020). Agricultural biomass are renewable and also possess low inorganic content (Younis et al., 2020). Adsorption method is the most efficient preferred and established method for the removal of organic and inorganics Igwegbe et al., 2018Igwegbe et al., , 2020Rashed, 2013). Adsorption is also advantageous due to the involvement of simple design and low investment cost .
The adsorbent selected for this research is very accessible and affordable since it is obtained from corn cobs which is an agricultural biomass; it does not compete with human and animal survival since it is not a source of food. Numerous agricultural wastes' ACs such as rice husks (Umembamalu et al., 2020), banana peels (Borhan et al., 2014), carbonised grass (Rahmat et al., 2017), and sugarcanne bagasse (Nadzirah et al., 2015) have been harnessed for the reduction of O&G in effluents even commerically obtained granular and powdered ACs (Al-Kaabi et al., 2019;Fulazzaky and Omar, 2012;Grieves et al., 1980). Also, raw agricultural wastes such as sugarcane bagasse (Boni et al., 2016;Hamid et al., 2016) and banana pith (Hamid et al., 2016;Sasirekha et al., 2018), and plants such as neem, Posidonia oceanica (Jmaa and Kallel, 2019). and curry leaves (Sasirekha et al., 2018) have been used for O&G removal from effluents. In the present work, corn cobs (CCs) were transformed to AC. CCs have been used successfully for adsorption of pollutants (Adams et al., 2017;Choi and Yu, 2019;Janani et al., 2019;Malode and Mamilwar, 2017;Muthusamy and Murugan, 2016;Norozi and Haghdoost, 2016;Sharma et al., 2019;Vu et al., 2018). As biomass, CCs have desirable properties such as porous structure, and chemical reactive groups (e.g. carboxy, hydroxyl) essential in sequestration of contaminants from effluents (Dai et al., 2018). But no research has been performed for the removal of O&G from desalter effluent via CCs in column studies. This study adopts the use of continuous column adsorption for O&G removal over the more common batch mode adsorption as it provides the most realistic application of adsorption activites (Rao, 2011), especially in water management. The impact of particle size, bed height (BH) and feed concentration at 10.5 mL/min on breakthrough time (τ) were examined. It is also necessary to determine the best CCAC size since its has a great effect on column adsorption studies. Smaller sizes create bubble effects and will shift a fixed bed to a fluidized bed. The data generated from the variation of the various process variables in the column adsorption process will give valuable perceptions into the adsorption mechanism and pathways of the reaction (Albadarin et al., 2012;Nwabanne and Igbokwe, 2012). Furthermore, the experimental data were modeled with the Bohart-Adams (B-A) alongside the Yoon-and-Nelson (Y-N) kinetic models.

Corn Cobs (CCs) and Reagents Collection
The CCs (Figure 1) were obtained from corn farmers in Awka, Nigeria. They were meticulously cleaned with distilled water and subsequently dried for 8 h at 105 o C with the use of oven. Thereafter, their sizes were reduced with a mortar and pestle, sieved to CCs sizes of between 1-2 mm with a molecular sieve and stored in a container.
Analytical graded chemicals were used as reagents for the experiment. A simulated refinery desalter effluent (SRDE) was prepared using deionized water (25 L), calcium chloride (CaCl), sodium chloride (NaCl), and crude oil. Freshly prepared solution of sodium hydroxide (NaOH) was also used for the activation of the CCs.

Chemical Activation and Carbonization of the CCs
Firstly, the stored CCs were redried. A solution of NaOH was poured into 100 g of the CCs (ratio = 1:1). The sample was then heated in a water bath at 80 o C with the shaker set at the speed of 150 rpm. Later, it was dehydrated at 120 o C for 24 h. The impregnated CCs was carbonized under N2 flow of gas for 3.5 h at 500°C in a muffle furnace to produce charcoal. The AC, having been conditioned to normal temperature, was rewashed several times with warm distilled water until pH of 6 -7 was attained to remove any remains of NaOH, passed through a filter paper (Whatman No. 1) and dried for 8 h at 115°C. The produced corn cobs activated carbon (CCAC) size was reduced and separated in varying CCAC particle sizes (150, 300, and 600 µm) with the use of sieves and airtighted.

Properties of the CCAC
Physicochemical characteristics of the CCAC were verified. The moisture content (MC) was obtained by drying 10 g of the AC placed in a crucible in the oven at 105 o C for 4 h. (Rengaraj et al., 2002). The percentage of moisture content (%MC) was determined using Eq. (1): The porosity (ƞ) of the CCAC was determined using Eq. (2):

Figure 1. Image of CCs
Where Vv = volume of void (cm 3 ) = Vt, total volume used for the experiment ( 2 ℎ) -Vs, volume of CCAC used ( ); r is the cylinder's radius in cm; Gs is the specific gravity of the CCAC = 0.365; Ms is the weight of CCAC in g, is the density of water in g/mL, and h is the height of cylinder used in cm.
The bulk density (BD) was determined by drying the AC in an oven at 105 o C for 1 h. The weight of the AC that filled the 25 cm 3 empty density was taken. The bulk density was computed using Eq. (3) (Devi et al., 2012): (3) ASTM D2866-94 was used to evaluate the ash content (AHC%) of the CCAC by first heating the CCAC placed in a crucible in a muffler furnace at 500 o C. The heated CCAC was cooled and the weight taken. 1 g CCAC was put into the crucible and reweighed. Then, it was placed in the muffler furnace and the temperature let to increase to 500 o C and the CCAC was removed, allowed to cool to room temperature and reweighed again. The AHC% was estimated via Eq. (4): For the pH evaluation, 2 g CCAC was poured into a beaker, 20 mL of distilled water was introduced and the mixture heated under reflux for 15 min. Then, the pH of the sample was taken after the sample was let to stabilize (Egwaikhide et al., 2007).

Preparation of Desalter Effluent
1000 mL de-ionized water was measured into 1000 mL beaker. Thereafter, 50 g of NaCl and 5 g of CaCl were measured with the aid of an electric weighing balance and was then poured into a volumetric flask of 1000 mL. After this, the deionized water was gradually poured into the flask and mixed properly until the 1000 mL mark was reached. 400 mL brine solution was measured out and introduced back into the 1000 mL beaker. Thereafter, 300 mg/L of crude oil (with properties: API gravity = 17.8, density = 0.963g/cm 3 at 15 o C, vapour pressure = 7 kPa, flash point = 97 o C, pour point = -10 o C, kinematic viscosity = 81.32 cSt and specific gravity = 0.959 at 60 o F) (Figure 2a) was injected into the 400 mL brine solution with the aid of a syringe. The mixture was then stirred using the magnetic stirrer at 15000 rpm for 15 min. The remaining 600 mL brine solution was then added and stirred at a reduced speed of 11000 rpm for additional 5 min. The simulated desalter effluent (Figure 2b) was then poured into a collection bottle (effluent bottle). This process was repeated for other concentrations of the feed (200 and 400 mg/L). The total time for the preparation of effluent is within the range of 45 min to 1 h.

Column Adsorption Experiment
Adsorption of O&G on CCAC was studied using packed adsorption column of 10 mm inside diameter and 600 mm length which was loaded with CCAC of varying bed heights (100, 200, and 300 mm) having a mesh at the bottom of the column. The containing vessel having the effluent feed was kept at a high elevation and a peristaltic pump (pump model: The effluent samples collected were tested for absorbance using a UV-visible spectrophotometer at 980 nm. The breakthrough curve was plotted as per Seader et al., (2011). The schematic representation of the fixed bed adsorption experiment is shown in Figure 4 (which was drawn using the Microsoft Visio).

Modelling of Column Study Results
The Bohart-Adams (B-A) model, extensively used in fixedbed columns design (Song et al., 2015;Dutta and Basu, 2013) was employed to predict the character of the column. The mathematically relationship can be given as (Swarup and Mishra, 2015;Song et al., 2015): The saturation concentration (No) and kinetic constant (KBA) were evaluated from the intercept and slope of Ln (Ct/Co) versus t which also gives R 2 (correlation coefficient) too.

Charactristics of CCAC
The characteristics obtained for CCAC are stated below ( Table 1). Bulk density (BD) specifies the fiber content of the precursor (Baseri et al., 2012). The BD value of 0.362 g/mL was measured. The pH value of the CCAC was observed to be near neutral which are useful for purification of water (Baseri et al., 2012;Igwegbe et al., 2020). The AC with high percentage of fixed carbon will have high capacity of adsorption (Dada et al., 2012); the CCAC has a good percentage of carbon.
The functional groups on the CCAC responsible for O&G adsorption were identified through the FTIR analysis using the Shimadzu 8400S spectrophotometer. The FTIR spectra of the CCAC is displayed in Table 2 and Figure 5 protraying the existence of alcohols, alkenes, alkanes, nitro compounds, aromatics, carboxylic acids and phenols. O-H stretching in phenols and alcohols, a broadband (Igwegbe et al., 2016) was seen; this strong band is necessary in adsorption processes because of the existence of hydrogen bonding (Batool et al., 2018;Zhang et al., 2017).

Influence of particle size:
The impact of the various CCAC particle sizes (150, 300, 600 µm) on the breakthrough curve was investigated at constant BH: 300 mm, flow rate (FR): 10.5 mL/min and feed strength: 300 mg/L (Figure 6). The breakpoint time, τ decreased from 475, 350, and 240 min as the particle size enlarged from 150-600 µm,. This was carried out at a breakthrough concentration of 90 %, that is Ct/C0=0.9. Also, it was observed that 150 µm particle size was best since it took longer time to attain saturation time than it took 300 and 600 µm particle sizes. Hence, it can be recommended for scale-up purposes, whereas, for laboratory-scale, 300 µm was selected because 150 µm shifted the adsorption column from being a fixed bed to a fluidized bed. Therefore, there was a need for more control to maintain the fixed bed adsorption column at 150 µm. Increasing the CCAC size decreased the %removal at different times but 5 min gave the maximum removal. Similar observation was reported by Umembamalu et al. (2020) Figure 7; 300 mm BH required a longer time to attain saturation when related to BHs of 100 and 200 mm. The %removal was improved as the BH rose from 100 to 300 mm. Maximum removals of 18.04, 27.97 and 33.66 % were observed at BHs of 100, 200 and 300 mm, respectively at 5 min. In other words, the higher BH corresponds to a higher amount of active sites and adsorbed oil and grease, and vice versa (Nwabanne and Igbokwe, 2012;Figure 6. A graph of the effect of particle size on breakthrough curve for O&G onto CCAC (BH = 300 mm, FR = 10.5 mL/min and feed strength = 300 mg/L) Figure 7. Graph of the impact of BH on the curve of breakthrough for O&G onto CCAC (CCAC particle size = 300 µm, FR = 10.5 mL/min and feed strength = 300 mg/L) Sivakumar et al., 2010). Similar observations were reported by Hernandez-Eudave et al. (2015), Umembamalu et al. (2020). As BH rose from 100, 200, and 300 mm, the breakthrough time increased from 390, 480 to 700 min, respectively. This was done using a breakthrough concentration of 90 %.

Feed concentration on breakthrough curves:
Varying feed strengths (concentrations) (200, 300, and 400 mg/L) on the curve of breakthrough were studied at FR: 10.5 mL/min, particle size: 300 µm, and BH: 300 mm (Figure 8). This was done using a breakthrough strength of 90 %. The breakthrough time decreased from 620, 300, and 180 min when the inlet feed strength increased from 200, 300, and 400 mg/L. Similar trend was observed by De Franco et al. (2018). Maximum removals of 60.38, 28.76 and 15.46 to % were obtained at feed concentrations of 200, 300 and 400 mg/L, respectively. This shows that increased feed concentration decreased the percentage removal. Lower feed strengths favoured the O&G percentage reduction. In other words, the higher the feed strength, the shorter the time of saturation of the bed (Hernandez-Eudave et al., 2015;Nwabanne and Igbokwe, 2012;Sivakumar et al., 2010). Also higher adsorption time declined the O&G removal. Similar observation was made by Umembamalu et al. (2020).

B-A kinetic results
Linear graphs of B-A model at varying CCAC sizes (Figure  9), bed heights (Figure 10) and feed strengths (Figure 11) were plotted. The values of the mass transfer coefficient, that  Table 3. The KBA, improved with BH. As the particle size decreased, KAB also decreased due to the reduction in the available adsorption sites, which is higher for small particles. Therefore at 150, 300, and 600 um, the KBA values were 500, 367, and 200 mL/min, respectively. As particle size increases, the adsorptive capacity decreases, as clearly outlined in Table 3; at 150 um, the No is 76780.048 mg/L and at 600 um, the No is 64518.053 mg/L. The B-A model was used on the experimental data to describe the initial part of the breakthrough curves. The high regression (R 2 ) values shows a high degree of fit of the linear equations obtained, therefore the equations can be used to predict the breakthrough curve of any desired values. For the bed height, the adsorptive capacity or saturation capacity (No) and the KBA increased with increasing BH. The decrease in the KBA with increasing feed concentration suggests that the kinetics of the entire system is manipulated by the external mass transfer in the preliminary section of the adsorption column (De Franco et al., 2018;Gong et al., 2015). Similar pattern was observed by Yunnen et al. (2017).

Y-N kinetic results
The linear Y-N graphs at different particle sizes, BHs and feed concentrations are shown in Figures 12-14; τ and KYN values are given in Table 4. The KYN improved with rising feed strength (Sivakumar and Palanisamy, 2009). The KNY also improved with increasing particle size and decreasing BH. τ (period of breakthrough) declined with rising feed strength, particle size, and bed height. Similar behavior was observed by Kapur and Mondal (2015) and Bhaumik et al. (2013).  The high values of correlation coefficients (R 2 ) for the B-A model ( Table 3) in comparison to that of Y-N model ( Table 4) points that the B-A model fitted best to the O&G elimination on CCAC experimental data. Therefore, it means that the B-A model can be used to predict the breakthrough curve of any desired values for this study.

The voidage of the packed bed
Voidage or void fraction of the packed bed is such an important parameter of great concern during industrial adaptation that must be evaluated in column adsorption studies. This delivers key information on bed stability and data for the better modeling of this system (Willoughby et al., 2000).
Where is the void volume and is the total volume; BH is the packed bed height, is the cross-sectional area of column and is the volume of particles; is the mass of the CCAC particles at different BHs, is the specific gravity of the CCAC (0.365) and is the density of water.
The ideal residence time ( ) of the effluent in the adsorbent bed was evaluated using Eq. (11) (Michel et al., 2018): Where Q is the volumetric feed flow rate.
The corresponding void fractions, mass of the CCAC measured, volume of particles and the residence time obtained at the different bed heights are shown in Table 5. The void fraction (ε) at BHs of 100, 200 and 300 mm were 0.0247, 0.0124 and 0.0082, respectively. It was observed that the void fraction increased with increasing BH. Similar observation was made by Willoughby et al. (2000). Also, the masss and volume of CCAC in the packed bed was increased with BH. Very low void fractions were obtained; this may be as a result of 150 -600 micron size particles been used for the study, so they could pack to form beds of very low voidage and such beds would offer enhanced resistance to liquid flow. The ideal residence time, was 4.49 min.

CONCLUSION
The adsorption O&G from simulated refinery desalter effluent by activated carbon originated from chemical carbonization of corn cobs (CACC) has been investigated. The breakthrough time (τ) was observed to decrease with increasing particle size and feed O&G concentration (strength) while it increased with increasing BH. Maximum removals of 54.93, 37.28 and 25.63 % were obtained for CCAC sizes of 150, 300 and 600 µm, respectively at constant BHs of 300 mm and feed concentration of 300 mg/L. Maximum removals of 18.04, 27.97 and 33.66 % were observed at BH of 100, 200 and 300 mm, respectively at 5 min, CCAC size of 300 µm and feed concentration of 300 mg/L. Also, maximum removals of 60.38, 28.76 and 15.46 to % were obtained at feed concentrations of 200, 300 and 400 mg/L, respectively at CCAC size of 300 µm and BH of 300 mm. The void fraction (ε) at BHs of 100, 200 and 300 mm were 0.0247, 0.0124 and 0.0082, respectively. The residence time, was 4.49 min. Bohart-Adams (B-A) and Yoon-and-Nelson (Y-N) equations were used to describe the fixed bed column kinetics/relationship between the operating factors. The saturation concentration (N0), and time required for O&G breakthrough (τ) were dependent on the feed concentration, particle size, and bed height (BH). The B-A best fitted the adsorptive elimination of O&G data than the Y-N model due to its high R 2 . This means that the B-A model can be used to predict the breakthrough curve of any desired values for this study. The decrease in the B-A constant of kinetics (KBA) with increasing feed concentration suggests that the kinetics of the entire system is manipulated by the external mass transfer in the preliminary section of the adsorption column. Scale-up of the controlled parameters can satisfactorily be applied for industrial columns design.