INTRODUCTION
Nigeria oil reserve is abundant and it is estimated to be about 209 billion cubic feet with its net oil export of over 2.5 million barrels per day (Energy Information Administration, 2007). But one of her oil fieldsObagi oil field in Rivers, Port Harcourt is having emulsion problem. The Basic Sediment and Water (BS and W) of the crude oil is 212% which makes it higher than the specified 0.5%. This problem makes crude oil produced from this region to have low market value because it is difficult to meet international market specifications.
Waterinoil emulsions are common occurrences during crude oil production. They are formed when oil and water are coproduced with sufficient agitation or injected water in reservoir, at well bore, in pipelines during flow of the mixture from the reservoir to the manifold and separators and at surface facilities (AlSahhaf et al., 2009). Emulsion stability ranges from a few minutes to years depending on the nature of the crude oil mixtures (Bhardwaj and Hartland, 1998). Crude oil is a complex mixture of hydrocarbons (Aske, 2002; Rhee et al., 1989) ranging from the paraffins to naphthenes and aromatics. It varies in color from clear to tarblack and in viscosity, from that of water to almost solid. It exists in the reservoirs in most cases as gas at the top followed by the oil and then water at the bottom (Hyne, 2001). Hence, during the production of crude oil, water accompanies the oil being produced. It has been reported that an equivalent volume of water accompanies the daily production of some 60 million barrels of crude oil (Ivanov and Kralchevcky, 1997). Owing to the various factors that affect production, a relative amount of this water can become completely dispersed in the crude oil as tiny droplets to form water in oil emulsion where oil is the continuous phase and water the dispersed phase. In most cases, crude oil emulsions result from the natural surfactants such as asphaltenes and resins contained in the crude oil which, when mixed with water, emulsifies the water into the oil (Sjoblom et al., 1992). This produces stable waterincrude oil emulsion which often has a much higher viscosity than either the crude oil or water alone (Kokal and Wingrove, 2000; Kokal et al., 2001). The film formed by the adsorption of asphaltenes and resins around the water droplets, is generally strong and difficult to break. The amount and nature of the emulsifying agents determines the stability of the emulsion (Bhardwaj and Hartland, 1998). The stable crude oil emulsion is characterized by high viscosity and rigid film results with significant formation damage to the reservoirs which increases the Basic Sediment and Water (BS and W) of oil. It also increases the operational and capital cost. Crude oil emulsion occupies larger volume of pipes thereby reduce the pipes handling capacity and causes corrosion of processing equipment (Becher, 1985). This necessitates the need to prevent crude oil emulsions formation or break them so as to minimize the production and processing problems associated with such emulsions (Gafonova, 2000). Several methods have been used in the deemulsification of crude oil such as thermal, electrical and chemical methods. Chemical method which involves the use of chemical agents called deemulsifiers break the stabilized oil/water interface with small doses at lesser time. For deemulsification to occur, this interfacial film has to be broken (Aveyard et al., 1992). These specialty chemicals basically work by weakening the stabilizing film hence promoting coalescing and settling. A great number of deemulsifiers have been formulated and used. Many more are still being formulated because the performance of deemulsifiers is known to be crude oil specific and as field conditions change, the deemulsifier requirements also changes. What works excellently well for crude A may perform very poorly for crude B and vice versa. Sometimes, two or more deemulsifiers may have to be combined in certain ratios for meaningful deemulsification to take place. The formulation of commercial deemulsifiers is largely based on empirical approaches in an attempt to get the most effective which can work in shorter separation time and at smaller dosages (Selvarajan et al., 2001). Other factors that promote the effectiveness of a chemical deemulsifier include sufficient mixing at the oilwater interface, right dosage and temperature (Kokal and AlJuraid, 1999). A good deemulsifier must possess ability to partition into the water and oil phases. There must be sufficient concentration of the deemulsifier in droplets to ensure high diffusion flux to the interface (Krawczyk, 1990). Deemulsifiers that have been proposed for use include organic substances and/or surfactants such as sulphonates (Porter, 1994), polyglycol ethers (Saywer et al., 1994), epoxy resins, polyamines (Myers, 1992), polyols, oxylated phenols, e.g., alkanolamine and nonylphenol ethoxylate derivatives (Easton and Thomas, 1989). The increasing economic need to eliminate emulsion and reduce the Basic Sediment and Water (BS and W) in order to meet strict crude specification for the Nigerian oil industry has led to this research work, to develop varied phenolformaldehyde (phenolic) based deemulsifiers and tested against a commercial deemulsifier sample using the popular bottle test method in a laboratory setting. Factorial design to study the interplay and interrelationship of mole ratios, concentration and temperature (multi variables) for deemulsifier’s optimum performance was carried out.
MATERIALS AND METHODS
Materials: Crude oil emulsion sample used was obtained from an onshore oil field in Port Harcourt, Rivers State, Nigeria. The various chemicals used include: Phenol crystals (99% purity, Technical grade), Formaldehyde (37% purity, Baker analyzed), Sodium hydroxide (99% purity, Riedelde Haen), Sulfuric acid (98% purity, SigmaAldrich analyzed) and xylene (99.9% purity; Baker analyzed). All chemicals were used without further purification.
Apparatus/equipment: A batch reactor comprising of a 250 mL 3neck flat bottom flask to which were mounted a reflux condenser and a quick fit thermometer (0250°C range) was set up inside a fume cupboard. The third neck (opening) on the reactor served as the charging port for the reactants and catalyst. Other equipment used includes pH meter, centrifuge machine, graduated centrifuge bottles and water bath.
Procedure for preparing resole and novolak resins: The 30% solution of caustic soda was first prepared. Based on a molar ratio of 1.2:1 of formaldehyde to phenol (F:P), accurately weighed and melted phenol was added into measured formaldehyde solution in a beaker. The mixture was stirred slowly as the pH was adjusted by adding drops of caustic soda solution until the desired pH was reached. The mixture was then charged into the reactor set up and heated for 3 h at the desired temperature. The pH was monitored and adjusted when necessary every 30 min to maintain it at the desired set value. Uniform agitation was ensured to prevent caking. At the end of three hours, the reaction mixture was then cooled and stored in an appropriately labeled sample bottle. Six different molar ratios, 1.2:1, 1.3:1, 1.4:1, 1.5:1, 1.7:1 and 1.8:1 of F:P were used in the formulations to examine their effects on the crude oil emulsion system. Temperature was varied depending on the type of resin desired: For low mole ratio resins (1.21.3:1), reaction temperature used was 95°C and at pH of 7.5. The resins in the intermediate mole ratio range (1.41.5:1) were prepared at 75°C at pH of 8 while the resins in the high mole ratio range (1.71.8:1) were prepared at 55°C and pH of 9. Reaction time was constant at 3 hours for all the formulations examined. For the novolak resins, the reaction procedures followed the same process as described for the resoles except that two mole ratios of 0.1:1 and 0.8:1 at a desired pH of (1 or 5), reaction time and temperature of 4 h and 100°C were respectively used.
Factorial design of experiment: The Design of Experiment (DOE) helped to study multiple experimental factors or variables at multiple levels. It was necessary in order to know the optimized value for each of the deemulsifier formulated. Using two levels and three factors, a software called “Minitab 16” ran a 2^{3} design of experiment table. The three factors considered were mole ratio of formaldehyde to phenol, temperature of deemulsification and concentration (in ppm) used. From the various F:P ratios, the deemulsifiers obtained were coded for ease of reference (Table 1).
Table 1:  Formulated deemulsifier samples and their codes 

Table 2:  Low and high level factor Variables for deemulsifier pair Q and T 

The deemulsifiers were then paired to run the 2^{3} DOE table. The pairing for the resoles were deemulsifiers A and D, B and E and C and F while the pairing for the novolaks were deemulsifiers GandI and HandJ. Table 2 shows an example of the two levels of each factor for one of the pairs used. The full factorial designs were made for all the deemulsifier pairs with “Minitab 16” and the performance of the deemulsifier calculated as % volume of water separated using Eq. 1:
The full factorial designs carried out with their corresponding responses for each pair of deemulsifiers are shown in Table 37.
Basic Sediment and Water test (BS and W): The basic Sediment and Water test was first done to know the percentage of water in the crude oil emulsion. Crude samples were collected and agitated to homogenize. The sample was poured into centrifuge bottle to 50% of its level. Pure xylene was used to make it up to 100%. It was shaken vigorously to homogenize.
Table 3:  2^{3} Full factorial design and its responses for deemulsifier pair Q and T 

Table 4:  2^{3} Full factorial design and its responses for deemulsifier pair R and U 

Table 5:  2^{3} Full factorial design and its responses for deemulsifier pair S and V 

Table 6:  2^{3} Full factorial design and its responses for deemulsifier pair W and Y 

Table 7:  2^{3} Full factorial design and its responses for deemulsifier pair X and Z 

This was placed in a water bath at temperature of 60°C for 15 min. The sample was put in a centrifuge and spun at 1500 rpm for 10 min. The percentage water and bottom sediment in the tube was recorded as ‘x’. The BS and W (%) = 2x.
Bottle test method: The bottle test method was used for screening the deemulsifiers formulated. The screening process consists of a series of centrifuge bottles spun in a centrifuge machine to effect the emulsion separation. The bottles containing the emulsion were first immersed in a water bath at the required temperature for 5 min. Twenty percent solution of each of the formulated deemulsifiers was prepared using xylene as solvent. This was then dosed or injected at the required concentration of 20 or 50 ppm into each of the centrifuge bottle and uniformly agitated by overturning 100 times. The bottles were then returned to the water bath for 20 min to acquire the right temperature of 50 or 70°C. At the end of 20 min, the bottles were placed into the centrifuge bottles chamber and spun at 1500 rpm for 20 min. Then the total volume of water drop, appearance of the water/oil interface (filaments, turbidity and sludge) and clarity of the water were observed and recorded. The most effective single deemulsifier was obtained based on the water separated from emulsion system.
RESULTS
The screening process of the commercial and formulated deemulsifiers was done using crude oil emulsion. The results are as given in Table 37 where the factorial design of experiment were analyzed using software called "Minitab 16’’.
Factorial optimization: Optimization of the best conditions for deemulsification obtained from the factorial design was carried out using Minitab 16 to analyze its output or responses in Table 37.

Fig. 1:  Pareto chart for deemulsifier pair Q and T 

Fig. 2:  Normal plot for deemulsifier pair Q and T 
The analysis tools used were Pareto plot, Normal plot, Main effects plot and Interaction plot. The results are graphically presented in Fig. 114.

Fig. 3(ac): 
Main effects plot of deemulsifier Q and T 

Fig. 4(ac): 
Interactions plot of deemulsifiers Q and T 
Pareto chart (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): Pareto charts (Fig. 1) were made in other to know the statistical significance of the main and interaction effects of the three variables (concentration, temperature and mole ratio) investigated during the deemulsification process. The reference (red line) on the chart helps to indicate which effects are significant. The charts considered the effects of each variable and their combined or interaction effects of the deemulsifier pair on deemulsification. The same effects and trend were observed for the other two deemulsifier pairs R and U and S and V.
Normal effects plot (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): Normal effects plot (Fig. 2) was made to compare the relative magnitude and the statistical significance of both main and interaction effects.

Fig. 5(ac): 
Interactions plot of deemulsifiers R and U 

Fig. 6(ac): 
Interactions plot of deemulsifiers S and V 
Minitab 16 drew a line to indicate where the points would expectedly fall if all effects were zero. Points that do not fall near the line indicated significant effects. In Fig. 2 for deemulsifier pair Q and T, using the volume data from Table 4 and for deemulsifier pairs R and U and S and V, there were three significant effects (at α = 0.05) which includedconcentration (A), temperature (B) and mole ratio (C). 3.1.3. Main effects plot (analysis of resole deemulsifiers pairs Q and T, R and U and S and V).
Figure 3 gives the main effect plot which was used to show how the volume of water separated from the crude oil emulsion responds to one or more of the factors.

Fig. 7:  Pareto chart for novolak deemulsifier pair W and Y 

Fig. 8:  Pareto chart for novolak deemulsifier pair X and Z 

Fig. 9:  Normal plot for novolak deemulsifier pair W and Y 
A horizontal line (parallel to the xaxis) means there is no main effect present and a line that is not horizontal shows there may be a main effect present. The larger the slope is, the stronger the effect of the variable on the deemulsification process. The relative magnitude of the effects of the factors can be seen by comparing the slopes of the lines on the plots.
Interaction effects plot (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): The interactions plot (Fig. 4) is used to visualize the interaction effect of two factors on the performance or response of the deemulsifiers and to compare the relative strength of the effects.

Fig. 10:  Normal plot for novolak deemulsifier pair X and Z 
Minitab 16 was used to draw interactions plots for the three factorsconcentration, temperature and mole ratio. An interaction is present when the change in the response mean from the low to the high level of a factor depends on the level of a second factor. The greater the degree of departure from being parallel, the stronger the effect.
Factorial optimization for novolaks deemulsifier pairs W and Y and X and Z: Same optimization process was carried out for novolak deemulsifiers using the same tools that were used for resole and are described as follows.
Pareto chart (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): Figure 7 and 8 gave completely different analyses of the two deemulsifier pairs in which Minitab used an αlevel of 0.05 to draw the reference line. For the deemulsifier pair W and Y, reference line cuts across the three main factorsmole ratio, temperature and concentration being investigated (Fig. 7).
Normal plot (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): In Fig. 9, the three variables considered showed significant effects in the deemulsification process. Temperature extended the farthest from the reference line indicating that it has the greatest effect on the deemulsification and was closely followed by concentration (ppm) of the deemulsifier while mole ratio of formaldehyde to phenol in the novolak deemulsifier negatively influenced deemulsification.
Main effects (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): For the two deemulsifier pairs W and Y and X and Z shown in Fig. 11 and 12 the three variables mole ratio, temperature and concentration produced similar effects on the crude oil deemulsification.
Interaction effects plot (Analysis of novolaks deemulsifiers pairs X and Y and Y and Z): In Fig. 13 and 14, different effects of the variables were observed. From Fig. 13 the lines in the temperaturemole ratio plot and the concentrationmole ratio plot are parallel to each other.

Fig. 11(ac): 
Main effects plot for novolak deemulsifier pair W and Y 

Fig. 12(ac): 
Main effects plot for novolak deemulsifier pair X and Z 
However, the lines in the temperatureconcentration plot are not parallel to each other indicating that there exists an interaction effect between temperature and concentration. The negative slopes of the lines in the temperaturemole ratio and concentrationmole ratio plots indicate that the interactions between these variables do not enhance deemulsification.
DISCUSSION
Pareto chart (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): For Pareto charts (Fig. 1), Minitab 16 used an αlevel of 0.05 to draw the reference red line. The farther the graph of the variable extends from the red line the more significant the effects of such variable.

Fig. 13(ac): 
Interaction effects plot for novolak deemulsifier pair W and Y 

Fig. 14(ac): 
Interaction effects plot for novolak deemulsifier pair X and Z 
For deemulsifier pair Q and T shown in Fig. 1, it was observed that the effect of each variable had significant effects on deemulsification. Mole ratio (denoted by C) had the largest effect on deemulsification because it extended the farthest from the red line. This was minimally followed by temperature (denoted by B) and concentration (denoted by A) gave the least effect. The combinedconcentration and temperature interaction (denoted by AB, temperature and mole ratio interaction (denoted by BC), concentration and mole ratio (denoted by AC) and concentration, temperature and mole ratio (denoted by ABC) had no effect on the deemulsification process because they did not extend beyond the reference line. The same effects were observed for samples B and E and C and F.
Normal effects plot (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): In Fig. 2 for deemulsifier pair Q and T, using the volume data from Table 5 and for deemulsifier pairs R and U and S and V, there were three significant effects (at α = 0.05) which includedconcentration (A), temperature (B) and mole ratio (C). Again mole ratio (denoted C) was seen to have the largest effect for all pairs of deemulsifiers on the deemulsification process because it lies farthest from the reference line. This was also minimally followed by temperature (denoted by B) and concentration (denoted by A) having the least effect.
Main effects plot (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): From Fig. 3 the plots indicate that for concentration, the volume of water removed increased as concentration was increased from 2050 ppm. For Temperature effect, volume of water removed increased as temperature was increased from 5070°C and for mole ratio, the volume of water removed increased as the mole ratio increased from 1.2:11.8:1. The relative magnitude of the effects of the factors can be seen by comparing the slopes of the lines on the plots. It was observed from the plot that effect of mole ratio on deemulsification was more significant with a main effect of 42% followed by temperature of 35% main effect and 32% main effect for concentration. The plots of concentration and temperature showed no much difference in the magnitude of their effects compared to the plot of mole ratio where there was a significant magnitude of its effect on the deemulsification process.
Interaction effects plot (Analysis of resole deemulsifiers pairs Q and T, R and U and S and V): As seen in Fig. 4, the lines are slightly parallel to each other in the concentrationtemperature and temperaturemole ratio plots. This showed that there were no interaction effects between these factors. However, in the concentrationmole ratio plot, the lines are not parallel to each other; hence, there was an interaction present. This meant that at high concentration the volume of water separated, increased as mole ratio increased. For deemulsifier pair R and U (Fig. 5) the concentrationtemperature plot, the lines are not parallel to each other. This showed the presence of an interaction between these factors on deemulsification. While for deemulsifier pair S and V, the temperaturemole ratio plot gave lines that were not parallel to each other which, showed that there was an interaction present (Fig. 6). At higher temperature, the volume of water separated increased as mole ratio was increased.
Hence, from the plots obtained from Minitab 16, the overall optimized deemulsification conditions for the deemulsifiers gave 50 ppm concentration and 70°C temperature and mole ratio of F:P as 1.8:1.
FACTORIAL OPTIMIZATION FOR NOVOLAKS DEEMULSIFIER PAIRS W AND Y AND X AND Z
Pareto chart (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): Figure 7 showed significant effects of all the three variables. Mole ratio extended farthest from the reference line which meant that mole ratio had the highest effect on the deemulsification process. This was closely followed by temperature and minimally by concentration. But for deemulsifier pair Y and Z (Fig. 8), the reference line did not cut across any of the bars which meant that none of the variables was significant at an alpha level of 0.05.
Normal plot (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): From Fig. 9, the mole ratio (C), extended to the opposite side of temperature (A) and concentration (B). This means that mole ratio less than 1 of formaldehyde as in the case of novolaks cannot enhance deemulsification of waterincrude oil emulsions. The differences between the various novolak deemulsifiers were in their molar ratios and pH of preparation. Table 1 shows the novolak deemulsifier samples W, X, Y and Z. While the molar ratio of phenol was kept constant at 1, that of formaldehyde was either 0.1 or 0.8 and at pH of either 1 or 5. If the formaldehyde mole ratio is high, the resin will have high methylol content and enormous water solubility. Thus the deemulsifiers with lower formaldehyde mole ratio resins (novolaks) have much less methylol content and hence low water solubility. This was what accounted for the poor performance of the novolak deemulsifier (Fig. 9 and 10) in the deemulsification the crude oil emulsion used. The deemulsifier pair X and Z did not produce any noticeable effects on deemulsification (Fig. 10).
Main effects (Analysis of novolaks deemulsifiers pairs W and Y and X and Z): It was observed that the volume of water removed increased as the concentration (ppm) of deemulsifiers and temperature of deemulsification were increased, respectively. While the mole ratio of formaldehyde to phenol in the novolak deemulsifier adversely affected rather than to enhance deemulsification. The volume of water removed decreased as mole ratio of formaldehyde to phenol was increased from 0.1:1 to 0.8:1. This negated the observed trend in resoles where, the percentage water removed increased as the mole ratio of formaldehyde to phenol was increased from 1.2:1 to 1.8:1. The method of preparation of the two deemulsifiers may have been responsible for the observed difference in performance between them. While the resoles were prepared in basic or alkaline medium, the novolaks were prepared in acidic medium. Also, the molar ratios of the formaldehyde in the novolak deemulsifiers were all below 1 mole.
Interaction effects plot (Analysis of novolaks deemulsifiers pairs X and Y and Y and Z): From Fig. 13, both concentration and temperature enhanced deemulsification. That is, the volume of water removed increased with increasing concentration and temperature. But in Fig. 14, the lines in the temperatureconcentration plot are slightly parallel to each other indicating that there was no temperatureconcentration interaction effect on the deemulsification process. However, in the temperaturemole ratio and concentrationmole ratio plots, the lines are not parallel to each other. Both of these interacting variables had no overall positive effects on deemulsification. This was the reason for the negative slopes observed in these plots. Their interactions are inversely proportional to each other in the deemulsification process. Hence, from Fig. 14, decreasing mole ratio and increasing temperature, decreasing moleratio and increasing concentration of deemulsifiers and vice versa favored deemulsification. Though, the extent of water separation from the crude oil emulsion was far lower than that obtained from resole deemulsifiers.
Thus from Fig. 714, the optimized deemulsification conditions for novolak deemulsifiers pairs W and Y and X and Z are concentration 50 ppm, temperature 70°C and mole ratio of 0.1:1 (F:P).
In comparison to works carried out by other researchers TempleHeald et al. (2015), AlSabagh et al. (2009) and Pena et al. (2005), where base catalysed ethoxylated phenolic resins were mostly investigated, the phenolic resins promoted coalescence of droplets (water separation) giving optimum performance as the resins’ hydrophilic property (or water solubility), concentration, temperature and molecular weight (mole ratios) were increased.
CONCLUSION
For the resole deemulsifiers, performance increased as the molar ratio of formaldehyde to phenol was increased from 1.2:1 to 1.8:1.
For the novolak deemulsifiers, performance decreased as the molar ratio of formaldehyde to phenol ratio was increased from 0.1:1 to 0.8:1.
In the deemulsification process using resole deemulsifiers, the most significant singular factor is the mole ratio of formaldehyde to phenol.
Temperaturemole ratio interaction had the greatest combined effect on the deemulsification process.
The performance of all the formulated deemulsifiers increased as the temperature of deemulsification was increased.
ACKNOWLEDGMENT
The authors gratefully acknowledge the financial support from Management of Covenant University to undertake this research.