Potential Efficient Separation of Oil from Bilgewater and Kitchen Wastewater by Fractional Freezing Process

Oily wastewater discharge to water bodies can have many negative consequences, especially on the marine ecological environment. Although there are numerous techniques for treating oily wastewater, this paper aims to introduce and evaluate the potential of the fractional freezing (FF) process as a new oil–water separation technique to overcome the several weaknesses found in the conventional oil–water separation methods. FF separates two liquid compounds based on their freezing point difference. In this study, two oily wastewater samples were used: oily bilgewater and oily kitchen wastewater. The effects of coolant temperature, freezing time, and stirring rate on the FF process efficiency were studied, and the significance of the data was supported by statistical analysis. The results show that a low coolant temperature is essential for allowing crystal nucleation formation and inducing crystal growth for an efficient separation process. However, the higher crystal growth rate that occurs at an even lower temperature might entrap the impurities inside the growing crystal. Consequently, continuing the crystallization for a longer time may yield a less efficient separation process. Furthermore, a too high stirring rate will rupture the solid formation, hence reducing the process efficiency. The final values of oil/grease and free fatty acids (FFA) obtained after the FF process of both samples were found to comply with the standard permitted by the International Maritime Organization (IMO) and Palm Oil Refiners Association of Malaysia (PORAM). Moreover, the p-values obtained for both of the above-mentioned samples were below 0.05 for all experiments. It can be concluded that this method has the potential to separate oil from the oily bilgewater and kitchen wastewater.


Introduction
Water pollution is one of the most significant issues worldwide [1] caused primarily by human activities [2]. Generally, sources of pollutions can be categorized in point-source and non-point source pollution. A pollutant that originates from an identifiable spot and single source is known as point-source pollution. On the other hand, non-point source pollution cannot be identified from a specific time or location and typically comes from diffuse sources [3,4]. Wastewater effluent, which includes municipal and industrial, and storm sewer discharge are examples of point-source pollution [5], whereas the water drainage from urban and agricultural areas are the examples of non-point-source pollution [4,5]. Oily wastewater released by marine and food services industries is considered as point-source pollution, since the origins of these pollutants are identifiable.
Bilgewater is water collected at the lowest corner of a ship where two sides of the ship meet, which is known as the bilge. Water that does not drain off the side of the deck will drain down into the ship and into the bilge. This collected water may be from manufacturing and repairing problems, along with high energy consumption [11]. On the other hand, coagulation is a process of destabilization of colloidal particles in the water to form bigger particles, which subsequently aids the separation of suspended solids and oils [11]. Although the efficiency of these methods is high, immoderate operator attention and maintenance are required to ensure proper process works.
Furthermore, the bilge water discharge from this technique produces high oil and chemical residues, which require proper handling for disposal. As a consequence, high operating and maintenance cost is needed to apply these methods on board [21]. Additionally, evaporation has been widely used to recover the used vegetable oil, in which water is removed from the solution by increasing the temperature to 100 • C, leaving concentrated vegetable oil. In some way, some nutrients are also vaporized, resulting in the degradation of taste and food value [22].
Fractional freezing (FF) is a physical separation process that works based on the difference in freezing point of the liquid components. The concept of FF has been applied in liquid concentration, which is known as freeze concentration (FC) or cryoconcentration. In this process, one liquid component is separated through crystallization at its freezing point and subsequently leaves behind a highly concentrated unfrozen solution [23]. Generally, there are three steps in fractional freezing: crystal formation (nucleation), crystal growth, and separation of the crystals and the solution [24,25]. The control of nucleation during crystal formation and ice crystal growth are vital for producing high-purity crystals.
A few manipulated and determinant parameters are used to evaluate the efficiency and performance of the separation via crystallization. The manipulated parameters are factors that can affect the efficiency of the freezing system and the quality of the ice and solid formed [26,27]. The relationship between those parameters is studied to determine the optimum condition to produce the highest separation efficiency [28]. Examples of the manipulated parameters are the coolant temperature, solution flowrate, initial solution concentration, freezing time, and stirring rate. Meanwhile, the standard determinant parameters used in studying FC's performance are partition constant (K), solute yield (Y), impurity ratio (IR%), and percentage of reduction or removal. The selection of the parameters is made based on the objective of the study.
Currently, there are two types of FF methods available, which are suspension fractional freezing (SFF) and progressive fractional freezing (PFF). The main difference between these two methods is the ice formation at the end of the process. In SFF, the ice crystals formed in the mother liquor are in small sizes as a suspension, whereas a large single ice crystal is formed in PFF [25,29]. Typically, the ice crystal layer will form on the inner wall of the crystallization where coolant is applied. Hence, the separation of the ice from the bulk solution is much easier for PFF compared to that for SFF.
This method is deemed to be practical in the removal of waste from the wastewater [30]. Hence, this method has been widely applied in liquid food and dairy industries and desalination [30][31][32][33][34] for years. In these reports, FF has proven to be effective in removing impurities or solutes from water, in which they were separated during the formation of ice crystals. FF's greatest attraction is maintaining the thermally sensitive components in the liquid due to the low process temperatures involved and low energy requirement, allowing the end products to have a better quality [35].
Therefore, this study introduced the new FF-precisely, the PPF technique-as an alternative method for oil-water separation and tested it on two types of oily wastewater: oily bilge water and used vegetable oil. Since there is a significant difference between oil and water freezing points, this study expects the FF method to yield an efficient oil-water separation that obeys both the national and international standards. The two samples used have a different freezing point compared with that of water. The oil in the bilge water has a lower freezing point than pure water; hence, it is expected that the water will freeze first. On the other hand, the vegetable oil from the kitchen wastewater samples with a higher freezing point than water will freeze first. Moreover, the effects of different parameters such as coolant temperature, freezing time, and stirring rate toward the efficiency of the FF process were also assessed by determining oil and grease values measured before and after the FF process, at which point the differences between these two readings were used to represent the efficiency of the FF process in oil-water separation. Based on the theoretical aspects elaborated in the above, it is hypothesized that the FF process can efficiently separate oil from the bilge water and used cooking oil as well as obey the limit mandated by IMO and PORAM.

Materials
For this study, the bilgewater sample, used vegetable oil, and ethylene glycol are the primary raw materials. The actual bilge water samples were collected from PETRONAS ULG95 Arwen tanker located at Lumut Maritime Terminal, Lumut, Perak, Malaysia, and used vegetable oil samples were collected from a drainage nearby Sri Nava Johthi Restaurant at Ipoh, Perak, Malaysia. Both samples were stored in a low-temperature room to preserve the composition and purity of the samples. Meanwhile, ethylene glycol and n-hexane were purchased from Avantis Laboratory Supply, Ipoh, Malaysia. Ethylene glycol was selected as the coolant due to its low-temperature range in the heat transfer process [36]. In the preparation of the coolant, pure ethylene glycol was mixed with distilled water in a 50% volume ratio to obtain a suitable freezing temperature range for the process. N-hexane was used as a solvent in extracting the oil and grease before measuring its value using a TOG analyzer. For the titration experiment, absolute ethanol alcohol, phenolphthalein, and potassium hydroxide (KOH) were used to test the FFA content after oil recovery.

Fractional Freezing Setup
The experimental setup of the FF process comprised of a cylindrical freeze crystallizer (RnR Tool & Machining, Ipoh, Malaysia), a digital stirrer (IKA, Rawang, Malaysia), two retort stands (Benua Sains Sdn. Bhd., Puchong, Malaysia), and a refrigerated water bath (Protech, Shah Alam, Malaysia). The critical apparatus of the system is the cylindrical freeze crystallizer, as it provides the place for crystallization to occur. Ethylene glycol-water solution was stored in the refrigerated water bath where the desired coolant temperature readings were controlled. On the other hand, two retort stands with a clamp were used to hold the crystallizer and digital stirrer, while the crystallizer was immersed into the water bath containing the coolant. The digital stirrer was used to stir the bilge water in the crystallizer at the desired stirring rate to have a well-mixed liquid mixture. Figure 1 shows the experimental setup for the fractional freezing process.

Experimental Procedure for Fractional Freezing
The experimental procedures were initiated by filling the refrigerated water bath with ethylene glycol and distilled water at a ratio of 1:1. Once the water bath was turned on, the coolant's temperature was reduced to the desired coolant temperature. Next, the cylindrical crystallizer was immersed into the refrigerated water bath and stabilized by

Experimental Procedure for Fractional Freezing
The experimental procedures were initiated by filling the refrigerated water bath with ethylene glycol and distilled water at a ratio of 1:1. Once the water bath was turned on, the coolant's temperature was reduced to the desired coolant temperature. Next, the cylindrical crystallizer was immersed into the refrigerated water bath and stabilized by the retort stand. On the other hand, 500 mL samples (i.e., oily bilgewater and used vegetable oil) were prepared, and the initial reading of the oil/grease value of the sample was recorded. As the coolant temperature reached the desired reading, the sample was poured into the crystallizer. At this time, the digital stirrer was operated to mix the sample at the designated stirring speed. The sample was left in the crystallizer to allow the crystallization process to occur. The stirring process was stopped at the particular designated freezing time, and the crystallizer was taken out from the refrigerated water bath. The unfrozen sample was first collected before thawing and collecting the frozen fraction. For kitchen wastewater recovery, the volume of both portions was measured and recorded to be used to calculate oil recovery.
This work had three varying operating parameters, which are coolant temperature, freezing time, and stirring rate, as shown in Tables 1 and 2. Since the one factor at a time (OFAT) method was implemented in this study, only one factor was varied, while the other two parameters were kept constant in each run. A summary of values of the parameters is shown in Tables 1 and 2. After each run of the experiment, the frozen fractions of the oily bilge water sample were analyzed for the oil/grease value, while for the kitchen wastewater sample, the unfrozen fractions were analyzed for oil/grease values. Additional analysis on free fatty acids (FFAs) content was conducted for the frozen fractions.

Determination of Oil/Grease Value
The collected sample from the FF process was poured into a separatory funnel and rinsed. The pH of the sample was adjusted to reach pH 2 by using hydrochloric acid. One-tenth of the n-hexane, which acted as a solvent used for extraction, was added into the separatory funnel containing the sample. Then, the funnel was shaken for 2 min, and the pressure was released subsequently. The mixture was left for the separation to take place. The filter paper was placed in a filter funnel, and 1 g of sodium sulfate was added. The mixture was drained into a clean container. A 5 mL syringe was used to withdraw 4 to 5 mL of the solvent layer. Approximately 50 µL of the solvent were extracted using a syringe and were placed onto the plate. The TOG analyzer was run, and the total oil and grease values in the sample were recorded. Then, the percentage of oil and grease removal, as well as the oil recovery, were calculated by using the equations below: Collectively, a high percentage of oil and grease removal (for oily bilgewater) and recovery (for oily kitchen wastewater) was aimed, indicating a high amount of oil and grease separation from the oily wastewater samples.

Determination of Free Fatty Acid Content (FFA%)
An amount of 5.0 g of oil samples was added into a dried conical flask. Then, the sample was mixed with 25 mL of absolute ethanol glycol and 2 to 3 drops of phenolphthalein. The mixture was heated in a water bath for 10 min while being stirred slowly. Then, the mixture was titrated with 0.1 N KOH solution until its color changed to pink. The volume of KOH solution was recorded, and FFA (%) was calculated by using the equations below: AV = (mL of KOH × N × 56) / mass of sample = mg of KOH (5) Free fatty acid content (FFA%) = AV × 0.503 (6) where N = Normality of KOH = 0.1 AV = Acid value.

p-Value Calculation
Statistical analysis was carried out via Microsoft Excel for Office 365 to identify the probability value (p-value) to determine the significance of the experiment. A t-test is used in the quantitative study to evaluate the null hypothesis by calculating the p-value. The p-value can be either in percentage or decimal form. However, it is commonly expressed in decimal form. If the value is equal to or lower than 0.05, then the hypothesis is considered significant; whereas if the value is equal to or lower than 0.1, the hypothesis is marginally significant; and it is considered insignificant if the p-value is larger than 0.1.
The statistical tests were performed in triplicate, and the average values were recorded. In general, there are two ways to find the p-value in Excel: using the t-test tool in the Analysis ToolPak and the 'T.TEST' function [37]. For the first method, ToolPak was loaded into Microsoft Excel before the analysis. Then, from the 'Data Analysis' icon, 't-test: Paired Two Sample for Means' was chosen before selecting all the required data. In this study, two variable ranges were involved, in which the variable 1 range was the initial values, and the variable 2 range was the final values after the experiment had been conducted. The alpha value selected was 0.05. The second method used to obtain the p-value was by utilizing the 'T.TEST' function. In this method, the syntax used was as follows: =T.TEST(array1,array2,tails,type) where 'array1' = the data range for the first data set (before) 'array2' = the data range for the second data set (after) 'tails' = number of distribution tails 'type' = type of t-test to perform.
For 'tails', there were two options that can be chosen, which were (1) for one-tailed and (2) for two-tailed analysis. Whereas, for the 'type', 3 different types of t-test needed to key-in, which are (1) for paired, (2) for two-sample equal variance, and (3) for two-sample unequal variance. In this study, the number of tails used was (2) two-tailed, whereas the t-test type performed was (1) paired.

Effect of Coolant Temperature
Coolant temperature is an important parameter that controls the growth rate of the solid during crystallization process [38]. The solid front's growth rate increases with a decrease in coolant temperature [26]. Hence, low coolant temperature is preferable for higher heat transfer efficiency from the crystallizer's wall to the coolant, providing a low surface temperature to allow ice nucleation formation [23]. However, impurities may trap in the solid layer formed when the temperature is too low, since the solid freezes faster. The phenomenon was explained in an experiment conducted by Jusoh et al. (2008) for assessing the effect of coolant temperature on a new progressive concentration system [39]. From the study, decrement of the temperature resulted in an increasing partition constant value (K), indicating the lower efficiency of the system [39], since the purity of the crystal formed decreased. The phenomenon explained the trend obtained in the study of coolant temperature toward the final value of oil and grease and the percentage of oil and grease removal/recovery.
The above statements were positively related to the results obtained in this study. As shown in Figure 2, the plots for both samples are more likely the same. As the coolant temperature decreased from the highest value to the lowest value, the percentage of oil and grease removal/recovery increased, indicating that both samples' final values of oil and grease decreased. However, the percentage started to decline back when the coolant temperature was further reduced, indicating that there were impurities trapped in the solid phase, hence decreasing the purity of the solid formed. The phenomenon is due to the high crystallization rate, since the temperature was too low. The same trend was also achieved by previous works, including a study to remove methylene blue from dye wastewater by Ab Hamid et al. (2019), as well as by  in their research on concentrated glucose solution [23,40]. Somehow, for the oily bilgewater sample, the percentage was increased from 72.5% (at −8 • C) to 77.8% (at −10 • C), yet it was still lower than the highest peak (79% at −6 • C).
From this analysis, the final oil and grease values obtained for the oily bilge water samples at −6 and −10 • C (13.2 and 14 mg/L) were found to have complied with the 15 mg/L limit set by IMO. On the other hand, for oily kitchen wastewater samples, the lowest oil and grease value after the FF process was 57 mg/L, which was obtained at 3 • C.
As for the statistical analysis, the studies for both bilgewater and kitchen wastewater samples achieved p-values of 0.0037 and 0.0011, respectively. The results are shown in Table 3 below.
A similar trend was recorded for the FFA content in the separated oil from the kitchen wastewater. As shown in Figure 3, the FFA content decreases from 5 to 3 • C; however, it increased back as the temperature decreased from 3 to 2 • C. The phenomenon might be due to the process's inefficiency as the coolant temperature increases since lesser oil and grease are recovered at a higher temperature. On the other hand, the higher value of FFA obtained at 2 • C can be possible because of more wastewater trapped in the solid crystal formed since the crystal's growth rate is very high at the lowest temperature; hence, the solid freezes faster. The lowest FFA content obtained is at 3 • C, which is 0.06%, and the highest is 0.1% at 5 • C. It can be seen that the values obtained obey the limit set by PORAM.
The above statements were positively related to the results obtained in this study. As shown in Figure 2, the plots for both samples are more likely the same. As the coolant temperature decreased from the highest value to the lowest value, the percentage of oil and grease removal/recovery increased, indicating that both samples' final values of oil and grease decreased. However, the percentage started to decline back when the coolant temperature was further reduced, indicating that there were impurities trapped in the solid phase, hence decreasing the purity of the solid formed. The phenomenon is due to the high crystallization rate, since the temperature was too low. The same trend was also achieved by previous works, including a study to remove methylene blue from dye wastewater by Ab Hamid et al. (2019), as well as by  in their research on concentrated glucose solution [23,40]. Somehow, for the oily bilgewater sample, the percentage was increased from 72.5% (at −8 °C) to 77.8% (at −10 °C), yet it was still lower than the highest peak (79% at −6 °C). From this analysis, the final oil and grease values obtained for the oily bilge water samples at −6 and −10 °C (13.2 and 14 mg/L) were found to have complied with the 15 mg/L limit set by IMO. On the other hand, for oily kitchen wastewater samples, the lowest oil and grease value after the FF process was 57 mg/L, which was obtained at 3 °C.
As for the statistical analysis, the studies for both bilgewater and kitchen wastewater samples achieved p-values of 0.0037 and 0.0011, respectively. The results are shown in Table 3 below.   A similar trend was recorded for the FFA content in the separated oil from the kitchen wastewater. As shown in Figure 3, the FFA content decreases from 5 to 3 °C; however, it increased back as the temperature decreased from 3 to 2 °C. The phenomenon might be due to the process's inefficiency as the coolant temperature increases since lesser oil and grease are recovered at a higher temperature. On the other hand, the higher value of FFA obtained at 2 °C can be possible because of more wastewater trapped in the solid crystal formed since the crystal's growth rate is very high at the lowest temperature; hence, the solid freezes faster. The lowest FFA content obtained is at 3 °C, which is 0.06%, and the highest is 0.1% at 5 °C. It can be seen that the values obtained obey the limit set by PORAM.

Effect of Freezing Time
An optimum freezing time in the FF process is necessary to achieve the highest purity of solid crystal. A longer freezing time will lead to a higher amount of crystal being formed. However, when the freezing takes longer than the optimum time, the formed crystal's purity could be decreased. The results can be proved from a study conducted by Jusoh et al. (2009) to determine the optimum freezing time for sugar solution removal in wastewater treatment [41]. The percentage of solute reduction was satisfying at 10 to 15

Effect of Freezing Time
An optimum freezing time in the FF process is necessary to achieve the highest purity of solid crystal. A longer freezing time will lead to a higher amount of crystal being formed. However, when the freezing takes longer than the optimum time, the formed crystal's purity could be decreased. The results can be proved from a study conducted by Jusoh et al. (2009) to determine the optimum freezing time for sugar solution removal in wastewater treatment [41]. The percentage of solute reduction was satisfying at 10 to 15 min of freezing time from the results obtained. However, the percentage started to decrease when the experiment was run for 20 min. It was observed that ice contamination had occurred, since the crystallizer was almost occupied by the ice layers formed as well as the high saturation of solute in the remaining liquid [41]. Figure 4 shows the percentage of oil and grease removal and recovery in both samples against freezing time. The result shows that the percentage of oil and grease removal for the oily bilgewater decreased, and then it increased again. The higher amount of ice crystal that has been formed as the freezing time increases can explain this trend. Hence, the inclusion of unwanted particles or impurities in the solid might occur, since increasing the freezing time would lead to a highly saturated solution, leading to lower crystal purity [42][43][44]. This trend is supported by studies that have been conducted by Ab Hamid et al. (2019) and Amran et al. (2018). In their studies, K values decreased before increasing again as the time increased. This indicates the lower purity of crystal formed after their optimum freezing time due to the entrapment of solutes in the ice layer [40,42].  Figure 4 shows the percentage of oil and grease removal and recovery in both samples against freezing time. The result shows that the percentage of oil and grease removal for the oily bilgewater decreased, and then it increased again. The higher amount of ice crystal that has been formed as the freezing time increases can explain this trend. Hence, the inclusion of unwanted particles or impurities in the solid might occur, since increasing the freezing time would lead to a highly saturated solution, leading to lower crystal purity [42][43][44]. This trend is supported by studies that have been conducted by Ab Hamid et al.
(2019) and Amran et al. (2018). In their studies, K values decreased before increasing again as the time increased. This indicates the lower purity of crystal formed after their optimum freezing time due to the entrapment of solutes in the ice layer [40,42]. Unlike removing oil and grease in oily bilgewater, the oil recovery in kitchen wastewater shows an increasing trend as the freezing time increased. This indicates that more oil was solidified as the time increased from 30 to 50 min even though there was a slight decreased percentage recovery from 50 to 60 min. It proves that the oil continuously solidified as the concentration time increased at a fixed coolant temperature and stirring rate. The highest percentage removal of 91.7% was obtained from the bilgewater sample at 20 min. However, the final values of oil and grease for oily bilgewater for all the freezing time managed to achieve lower than the IMO limit, indicating that high separation occurred between 20 and 60 min of freezing time. On the other hand, the lowest oil and grease value obtained for the kitchen wastewater sample was 55 mg/L, which was recorded at 50 min.
In addition, low p-values were obtained for both samples, in which it was 3.85 × 10 −6 for the bilgewater sample and 0.001 for the kitchen wastewater sample. Table 4 demonstrates the results.  Unlike removing oil and grease in oily bilgewater, the oil recovery in kitchen wastewater shows an increasing trend as the freezing time increased. This indicates that more oil was solidified as the time increased from 30 to 50 min even though there was a slight decreased percentage recovery from 50 to 60 min. It proves that the oil continuously solidified as the concentration time increased at a fixed coolant temperature and stirring rate. The highest percentage removal of 91.7% was obtained from the bilgewater sample at 20 min. However, the final values of oil and grease for oily bilgewater for all the freezing time managed to achieve lower than the IMO limit, indicating that high separation occurred between 20 and 60 min of freezing time. On the other hand, the lowest oil and grease value obtained for the kitchen wastewater sample was 55 mg/L, which was recorded at 50 min.
In addition, low p-values were obtained for both samples, in which it was 3.85 × 10 −6 for the bilgewater sample and 0.001 for the kitchen wastewater sample. Table 4 demonstrates the results. For this set of study, the lowest value of FFA (%) content (0.05 %) in the separated oil from the oily kitchen wastewater was obtained at 50 min of freezing time. Figure 5 shows that the FFA (%) content decreased as the freezing time increased from 20 to 50 min, before it increased back at 60 min. The phenomenon is due to the sufficient time, which helps the ice growth rate. However, there is always an optimum time for ice growth because prolonged crystallization will result in a high number of solutes being trapped in the ice, leaving less solute in the unfrozen solution, increasing the FFA (%), and disrupting the quality of oil recovered [45].  For this set of study, the lowest value of FFA (%) content (0.05 %) in the separated oil from the oily kitchen wastewater was obtained at 50 min of freezing time. Figure 5 shows that the FFA (%) content decreased as the freezing time increased from 20 to 50 min, before it increased back at 60 min. The phenomenon is due to the sufficient time, which helps the ice growth rate. However, there is always an optimum time for ice growth because prolonged crystallization will result in a high number of solutes being trapped in the ice, leaving less solute in the unfrozen solution, increasing the FFA (%), and disrupting the quality of oil recovered [45].

Effect of Stirring Rate
In this study, stirring is introduced to reduce the solute accumulation in the samples near the solid-liquid interface due to the stirrer's constant flow distribution. The higher stirring rate will reduce the solidification rate of the solid front, resulting in higher purity of the solid formed [26,38,46].
From the result shown in Figure 6, it can be seen that the trend of the percentage of oil and grease removal for oily bilgewater is different from the percentage recovery of oily kitchen wastewater. There is a slight inconsistency, where the percentage decreased from 50 to 150 rpm; then, it increased and decreased back to 200 and 250 rpm, respectively. The increment of the percentage at 200 rpm is due to the higher heat transfer rate for crystallization, causing the solid to be formed from the cooling wall of the crystallizer, and removing the unwanted solutes from the solid crystal [23,38]. The entrapment of solute in the solid can also be prevented, since a high stirring rate produces a high shear force that can

Effect of Stirring Rate
In this study, stirring is introduced to reduce the solute accumulation in the samples near the solid-liquid interface due to the stirrer's constant flow distribution. The higher stirring rate will reduce the solidification rate of the solid front, resulting in higher purity of the solid formed [26,38,46].
From the result shown in Figure 6, it can be seen that the trend of the percentage of oil and grease removal for oily bilgewater is different from the percentage recovery of oily kitchen wastewater. There is a slight inconsistency, where the percentage decreased from 50 to 150 rpm; then, it increased and decreased back to 200 and 250 rpm, respectively. The increment of the percentage at 200 rpm is due to the higher heat transfer rate for crystallization, causing the solid to be formed from the cooling wall of the crystallizer, and removing the unwanted solutes from the solid crystal [23,38]. The entrapment of solute in the solid can also be prevented, since a high stirring rate produces a high shear force that can carry away the solute from the solid crystal and remove the trapped solute between the dendritic ice structure [23,29]. In addition, according to Jusoh et al. (2014), in their study of concentrated coconut water, they found that increasing the circulation flowrate leads to the decrement of effective partition constant (K) value, indicating that higher ice purity is formed. This is due to the slower speed of the ice front formation, since the contact time between the solution and the cooling surface is low; hence, there is less time for the ice growth to trap unwanted solute at the ice-liquid interface [29]. Meanwhile, the decrement of the percentage removal at 250 rpm might be due to solid growth interruption when the stirring rate is set at a vigorous speed. This finding is also discussed in the studies conducted by Amran et al. (2018) and Ab Hamid et al. (2015). The vigorous flowrate of the solution has washed away the slushy form of the ice's front growth [26,42]. carry away the solute from the solid crystal and remove the trapped solute between the dendritic ice structure [23,29]. In addition, according to Jusoh et al. (2014), in their study of concentrated coconut water, they found that increasing the circulation flowrate leads to the decrement of effective partition constant (K) value, indicating that higher ice purity is formed. This is due to the slower speed of the ice front formation, since the contact time between the solution and the cooling surface is low; hence, there is less time for the ice growth to trap unwanted solute at the ice-liquid interface [29]. Meanwhile, the decrement of the percentage removal at 250 rpm might be due to solid growth interruption when the stirring rate is set at a vigorous speed. This finding is also discussed in the studies conducted by Amran et al. (2018) and Ab Hamid et al. (2015). The vigorous flowrate of the solution has washed away the slushy form of the ice's front growth [26,42]. However, the trend of the percentage oil and grease recovery for oily kitchen wastewater is consistent, in which as the stirring rate increased, the percentage of recovery decreased, indicating an increasing final value of oil and grease. In other words, the slow stirring rate is preferable for this oily sample. This can be possible because of the dispersion of the vegetable oil away from the crystallizer wall while stirring occurred, causing high content of the oil in the unfrozen fraction by the end of the experiment.
The lowest oil and grease value obtained for the oily bilgewater treatment was 14 mg/L at 200 rpm with the highest percentage removal (77.8%); meanwhile, the highest oil and grease value is 31.9 mg/L at 150 rpm with the lowest percentage removal (49.4). Thus, out of five set values of the stirring rate, the bilgewater sample that has been conducted at 50 rpm and 200 rpm managed to comply with the 15mg/L discharge limit. It can be said that this parameter is less significant for the efficiency of the oily bilgewater treatment, since the lowest final oil and grease value obtained in this study is 14.mg/L, which is highest compared to the final values achieved in the study of coolant temperature and freezing time effects. On the other hand, for the separation of oily kitchen wastewater, the lowest oil and grease value is 54 mg/L at 50 rpm (74% recovery), and 96 mg/L is the highest value at the highest stirring rate, 200 rpm (51% recovery).
For the statistical analysis, the p-values obtained for the bilgewater sample and kitchen wastewater sample were 0.0004 and 0.0014, respectively. Table 5 shows the results obtained from the experiment and statistical analysis conducted. However, the trend of the percentage oil and grease recovery for oily kitchen wastewater is consistent, in which as the stirring rate increased, the percentage of recovery decreased, indicating an increasing final value of oil and grease. In other words, the slow stirring rate is preferable for this oily sample. This can be possible because of the dispersion of the vegetable oil away from the crystallizer wall while stirring occurred, causing high content of the oil in the unfrozen fraction by the end of the experiment.
The lowest oil and grease value obtained for the oily bilgewater treatment was 14 mg/L at 200 rpm with the highest percentage removal (77.8%); meanwhile, the highest oil and grease value is 31.9 mg/L at 150 rpm with the lowest percentage removal (49.4). Thus, out of five set values of the stirring rate, the bilgewater sample that has been conducted at 50 rpm and 200 rpm managed to comply with the 15mg/L discharge limit. It can be said that this parameter is less significant for the efficiency of the oily bilgewater treatment, since the lowest final oil and grease value obtained in this study is 14.mg/L, which is highest compared to the final values achieved in the study of coolant temperature and freezing time effects. On the other hand, for the separation of oily kitchen wastewater, the lowest oil and grease value is 54 mg/L at 50 rpm (74% recovery), and 96 mg/L is the highest value at the highest stirring rate, 200 rpm (51% recovery).
For the statistical analysis, the p-values obtained for the bilgewater sample and kitchen wastewater sample were 0.0004 and 0.0014, respectively. Table 5 shows the results obtained from the experiment and statistical analysis conducted.  Figure 7 illustrates an increase in FFA content with increasing stirring rate. Increasing the stirring rate beyond the optimum point might cause an erosion to the solid layer, leading to less concentrated solution produced, reduced efficiency [42], and an increased FFA value. The lowest FFA value obtained with 0.07% at 50 rpm shows that the recovered oil quality is good and is close to the standards set.   Figure 7 illustrates an increase in FFA content with increasing stirring rate. Increasing the stirring rate beyond the optimum point might cause an erosion to the solid layer, leading to less concentrated solution produced, reduced efficiency [42], and an increased FFA value. The lowest FFA value obtained with 0.07% at 50 rpm shows that the recovered oil quality is good and is close to the standards set.

Conclusions
The novel FF method is proven to have the potential to separate oil-water for the oily bilgewater and oily kitchen wastewater efficiently. The separation process is affected by all the parameters tested in this study: coolant temperature, freezing time, and stirring rate. Determining the optimum conditions related to these parameters is essential for efficient oil-water separation. The step is crucial so that the treated bilge water discharge can comply with the discharge limit, and the recovered used vegetable oils can be used as value-added products, thus reducing the oily wastewater discharge to the environment.
As a whole, the most optimum conditions for the coolant temperature, freezing time, and stirring rate that resulted in the lowest oil and grease values for oily bilgewater in this particular study are at −6 °C, 20 min, and 200 rpm, respectively. On the other hand, for the oily kitchen wastewater, the best conditions are at 3 °C (coolant temperature), 50 min (freezing time), and 50 rpm (stirring rate). However, despite the proven ability of this method in oil and water separation, further detailed research can be done to produce more data in order to validate the significance and accuracy of the parameters involved in this study.

Conclusions
The novel FF method is proven to have the potential to separate oil-water for the oily bilgewater and oily kitchen wastewater efficiently. The separation process is affected by all the parameters tested in this study: coolant temperature, freezing time, and stirring rate. Determining the optimum conditions related to these parameters is essential for efficient oil-water separation. The step is crucial so that the treated bilge water discharge can comply with the discharge limit, and the recovered used vegetable oils can be used as value-added products, thus reducing the oily wastewater discharge to the environment.
As a whole, the most optimum conditions for the coolant temperature, freezing time, and stirring rate that resulted in the lowest oil and grease values for oily bilgewater in this particular study are at −6 • C, 20 min, and 200 rpm, respectively. On the other hand, for the oily kitchen wastewater, the best conditions are at 3 • C (coolant temperature), 50 min (freezing time), and 50 rpm (stirring rate). However, despite the proven ability of this method in oil and water separation, further detailed research can be done to produce more data in order to validate the significance and accuracy of the parameters involved in this study.