The focus is with the experimental methods for generating the discrimination variables that are required for performing the discriminant analysis for characterization and identification of crude oil samples. In order to achieve this objective, the samples are weathered in the laboratory to simulate as much as is possible the weathering conditions in the seas. Also, we shall present the results of the weathering experiments performed on six Nigerian export crudes.
Test samples were obtained from the Niger Delta region of Nigeria. The crude oil samples are the following.
1. Bonny Light
2. Bonny Medium
5. Forcados Blend.
6. Delta field.
These are representative samples from the two halves of the Niger Delta. Four representative samples were drawn from the Eastern zone of the delta alluvia while the other two samples were from the Mid-Western alluvia region. These representative samples are the most probable sources of marine pollution in the Nigerian waters since the export crudes are Forcados Blend, Bonny Light and Bonny Medium. They are the most transported crude in our regional waters.
Several environmental factors were considered for the evaluation of weathering effects on sample chemical fingerprint indices. These included the degree of water washing of the oil, exposure to sunlight, wind effects and contact time. For all samples and test variables, the Lagos Bar Beach sea water was used under atmospheric temperature conditions.
The time that an oil sample is exposed to the environment (contact time) is important in its measure of the rate of weathering. The effect of contact time was incorporated into the study by varying the experimental exposure levels of the test, the two levels applied were ten (10) days and twenty one (21) days and the washing rates of 6-8 litres/min and 12-16 litres/min.
A high washing level increases the leaching rate of partially water soluble components from the oil sample and increases the tendency of sample to emulsify, which further accelerates leaching by providing a higher oil-water interracial area. Water-oil contacting may also control chemical processes which are diffusion limited. Thus, the level of water washing was selected as the second weathering variable for the experimental tests.
Wind speed and sunlight intensity were also considered important environmental factors. Wind speed can greatly affect the rate of evaporation of volatile components and ultraviolet light from the sun can affect chemical oxidation changes. However, the intensity of these weathering factors changes substantially in nature in the short time intervals and their use as experimental variables, though of interest was considered to have lower practical value. A KDK table fan (Model E40BK4) and a UV lamp (150 Watts) were provided for the simulation of sunlight and wind in the experimentation but were not used as experimental weathering variables.
The responses used to determine the nature and magnitude of any effects of these variables were selected compound concentration ratios of candidate fingerprint tags. Compound indices that showed no change ( or minimal change) during weathering and which had the best discriminating capability for the oils are used to develop the complete fingerprint functions.
The main function of the environmental simulation test facility schematic is to provide a means of simulating and controlling the environmental (weathering) variables being tested during the experiments. The main components of this facility included:
A continuous water recirculation system.
A reservoir for holding sea water.
A 15-litre baffled tank for the oil samples.
A magnetic stirrer.
A UV lamp to simulate exposure to the sun, and
A fan to simulate wind currents.
The continuous water recirculation system is equipped with the 120 litre drum furnished with galvanized sheet baffle raised to about 2 metres above ground level. The sea water was continuously pumped from the 120 litre drum into the weathering tank. In the 15 litre tank, the water continuously “washed” the test oil sample and then was returned to the drum by gravity. The underflow baffle in the weathering tank allowed the passage of water but prevented the oil sample which remained on the sea water surface from leaving the tank. The control valve is used to regulate the water recirculation rate.
The 15 litre tank which contained a magnetic stirrer provided additional mixing between the sea water and the oil sample in the tank.
A UV lamp is mounted about 1 metre above the weathering tank and shone directly on its surface. A KDK table fan is placed above the tank directly on the sea water surface.
The weathering tests were conducted by establishing steady state condition of sea water recirculation rate and magnetic stirrer speed and then adding 500 ml of test sample onto the sea water in the weathering tank.
Lagos Bar Beach sea water (salinity = 0.2838 moles solute/1000 grammes water) and atmospheric room temperature was maintained throughout the experiments. Temperature variation was usually between 30-34 degrees centigrade even at high sea water recirculation rate. Lamp heat was efficiently removed by the heat sink effect of the recirculation water. UV lamp was on for 12h each day and the test sample was continuously faned throughout the duration of the test. High and low water mixing (washing of oil) conditions were regulated using the control valves. The high and low water washing rates were regulated at 12-15 and 6-8 litres/min. The flow rates and stirrer speed were adjusted daily to maintain the desired levels.
Weathered oil samples were removed from the tank at 10 and 21 day intervals. In addition to the weathered samples, portions of unweathered oil of each type under test were submitted for analysis. The analytical processing and separation of the oil samples evaluated here precede weathering tests.
Sample Separation Methods.
The weathered and unweathered oil samples were subjected to an analytical processing scheme to obtain the necessary fingerprint data. Two samples were taken from each oil under study for each set of experimental conditions.
One sample was used for nickel, vanadium, sulphur and nitrogen analyses. The gas chromatographic analysis required considerable sample processing and separation before these techniques could be applied. The processing is best described with reference to the analysis schematics already depicted.
Removal of Water and Light Ends.
Samples were first distilled in a glass still to remove any water(which was always present with weathered samples) and to remove light ends with boiling points up to 240 degrees centigrade. The light ends were not considered reliable fingerprint indices and tended to make subsequent separation difficult. They were thus removed in this preliminary distillation step. This generally follows the procedure described for distillation of crude petroleum of ASTM test D285-68. In this procedure, 100 ml of test sample was charged into 1litre distillation flask placed in a heating jacket capable of low heat retention with attachments for close heat control adjustable from 0 to 300 W. The heating unit was equipped with a reflux condenser, support, still head with thermometer and recovery flask. With a steady water recirculation rate through the reflux condenser, heat was gradually applied by means of the heat control attachment. As the sample boils, the distillate was collected over the reflux condenser in a receiving conical flask. When the thermometer reads the predetermined temperature, the heater was switched off. This procedure was repeated many times until no more distillate collects. The distillation process was thereafter discontinued. The residue from the distillation step was then treated to remove insolubles. The procedure for effecting the removal of the insolubles is as follows: 2.0 g of the sample was added to 20 ml n-pentane. This mixture was charged to a filtering assembly with fine porosity filter and suction for the aspiration of test samples. Then, three successive additions of 10 ml portion of n-pentane solvent were used to rinse the flask, inner walls of filter funnel and precipitate and the aspiration was continued until the precipitate on the filter became dry.
The amount of n-pentane insolubles was determined by weight difference of the insoluble precipitate predried in the oven and cooled in a desiccator to a constant weight.
Removal of Polar Compounds.
After the insolubles were removed, the sample dissolved in n-pentane from above was fed to a clay separation column to remove polar compounds. The clay column used in the separation followed the ASTM procedure D2007-65T.
In this procedure, the clay column polar compound was packed with 30 g of the porocel clay (coarse size) carefully weighed to the nearest milligram. The column was prewetted with10 ml portion of the n-pentane. As soon as nearly all of the permanent had percolated into the clay, the sample dissolved in n-pentane from the insolubles removal was charged into the column. This was followed with the addition of 8 separate 5 ml portions of n-pentane which were used to wash the sample flask to the column. The n-pentane eluant from the clay column will contain the sample saturate compounds (paraffins and naphthenes) and aromatics.
After the n-pentane eluent had essentially drained from the column, 40 ml of 70% benzene – 30% acetone mixture was added to the clay column. The eluent was collected in a graduated conical flask. The column was finally cleaned with 15 ml portion of acetone. The eleunt (acetone -benzene and acetone) will contain the Polar fraction of the sample. The Polar compounds were not considered as candidate chemical fingerprints because of their high solubility.
Separation of Aromatics from Saturates.
The n-pentane eluent from the clay separation step ( containing saturate and aromatic fractions of the original sample) was stripped of its pentane and charged to a silica gel column to separate out the aromatics.
The silica gel column was packed with 10g of silica gel (30-120 mesh size). The saturate aromatic cut residue of the clay separation step was dissolved in 5 my pentane and charged into the column after prewetting the column with 10 ml n-pentane solvent. Addition of 8 separate 5 my portions of n-pentane was added to the column to elute the saturate fraction of the sample from the adsorbents. Finally, the column was cleaned using 20 ml portion of acetone. The acetone eluent will contain the aromatic compounds.
GC Analysis of the saturate fraction.
The saturate fraction contains n-paraffins and these compound indices were resolved using gas chromatographic analysis. Pure components were used in the identification of the sample peaks while constituent concentration of n-paraffins were normalized with compound range.
HPLC Analysis of Aromatic hydrocarbons.
The aromatic cut was resolved into mono-, di-, and polynuclear aromatic compounds using the HPLC. Waters Associates Model-ALC 200 HPLC equipped with a Model 6000 solvent delivery system and a Model-440 Absorbence detector was used for the fractionation of the aromatic cut. The HPLC columns used here were two in series (each 4mmx30cm) and the solvent used was n-hexane. The column was calibrated using standard aromatic compounds and relative quantities of the compounds present in synthetic mixtures were determined.
Infrared Absorbance Analysis.
Infrared spectra were recorded from 4,000 to 650cm for each crude oil sample on a Perkin Elmer Model SP3-100 spectrophotometer equipped with 0.1 mm, fixed-path, sealed, sodium chloride sample cells. All spectra were as “neat” (undiluted) samples.
Chemical Type Analyses
Nickel and Vanadium Analysis.
Nickel and vanadium were determined by atomic absorption/flame emission spectroscopy following ASTM D3327-79 method for the analysis of waterborne oils. In this method, the sample was initially ashed in the presence of benzene-sulphonic acid, followed by additional heating to burn off carbon leaving an inorganic residue. The residue was dissolved in hydrochloric acid and the nickel and vanadium were measured by atomic absorption using air-acetylene flame for nickel and by flame emission using nitrous-acetylene flame for vanadium.
Sulphur was determined by x-ray spectroscopy following ASTM D-3327-79 method for analysis of waterborne oils. In the x-ray procedure, sulphur was determined by placing the sample in an X-ray beam and measuring the sulphur characteristics radiation, the intensity of which is a quantitative measure of the sulphur concentration.
Kjeldahl Nitrogen Analysis
Kjeldahl nitrogen analysis for waterborne oils, ASTM D3327-79 was used for the determination of nitrogen in experimental samples. The sample was digested in a mixture of concentrated sulphuric acid, potassium sulphate and mercuric oxide. After digestion, sodium sulphide was added to precipitate mercury, the mixture was made alkaline with sodium hydroxide and the nitrogen as ammonia was distilled into boric acid solution and titrated with standard sulphuric acid to methyl purple end point.
Crude oil samples included three leading export crudes from Nigeria, namely, Bonny Light, Bonny Medium and Forcados Blend. Other crudes from the East and West were selected to test the ability of the identification system to distinguish closely related oils and also to examine the peculiar characteristics of crude fingerprints in relation to source formation.
The extent of experimental weathering conducted and the physical changes in the character of oil samples observed were in the theological properties of tested samples. The crude became more glossy and viscous as the lower molecular weight components had been washed away. Therefore, the weathered samples became frothy and more resistant to flow. This trend in the viscosity and specific gravity measurements of the weathered and unweathered oil samples.
The loss of volatile components is indicated by the decrease in boiling range for each weathered oil and higher boiling temperature for an equivalent volume percent distilled. The boiling point distribution data obtained for weathered oil samples. At 10 and 21 days intervals show little difference in boiling point composition. This indicates that virtually all the light ends were lost within the 10-day weathering interval. The weathering simulation scheme used subsequently thus provide for significant evaporative weathering effects.
The Characteristic IR spectra of Bonny Light crude oil and for comparative purposes, that of Lagos Bar Beach Tar balls. The methyl group asymmetric and symmetric stretching and blending absorption were observed at the frequencies of 2930, 2850,1460, and 1380 cm, respectively. This characteristic spectra show clearly that the test samples are petroleum hydrocarbons with predominant C-H stretching and bending vibrations. The infrared spectra absorbance results for Forcados, Bonny Light and Bonny Medium crude oils. Frequency band absorbances do not show any distinctive trend, but all tested samples give rise to common characteristic spectra at or near the same frequencies suggesting that all samples have similar structural forms. IR spectra of other crude oil sample show similar results.
The absence of infrared carbonyl absorbances at 1700 CM shows that no significant oxidative effect was induced due to weathering even after 21 days of weathering in the test facility. The selected fingerprint indices were thus virtually unchanged during the test.
Sample Separation Methods.
The sample separation processing step is dependent on the performance of the clay and silica gel columns. For effective separation, both the clay and the silica gel were first activated before use. The derivation was conducted at 250 degrees centigrade in a vacuum electrical oven for 16 hours before the columns were packed for the sample separation experiments.
The components eluting in time with n-pentane are the saturate compounds, while those eluting with 70% benzene -30% acetone mixture contain the aromatic cut. However, there was no sharp separation between the fractions as the fractions tend to overlap.
The analysis schematic was aimed at providing relatively weathering insensitive data. The weathered and unweathered oil samples were subjected to a sample processing scheme to obtain the necessary fingerprint data. The samples were first distilled to remove any water which was always present with weathered samples and to remove light ends with boiling points up to 240 degree centigrade. Light ends were not considered reliable fingerprint indices and tended to make subsequent separation more difficult.
The next step was the removal of insolubles by filtration. The Polar compounds were removed in a clay separation column using n-pentane as eluent. The Polar compounds were not considered suitable chemical fingerprint candidates because of their high solubility. These sample processing steps preceded the sample separation into saturate and aromatic compound fractions. The weight percent of saturate and aromatic fractions remained constant for each crude oil sample. For Delta field crude oil, for example, the weight percent for saturate fraction for both weathered and unweathered crude samples varied only between 56.30% and 57.85%, and the aromatic fraction between 21.90% and 22.50%. For the Polar ends and the n-pentane insolubles, the variation is significant, justifying their rejection as candidate chemical fingerprints.
For a measured property to be important for the differentiation of crude oils or other related samples, the variation in the property for a large population of samples must be significantly larger than the analytical uncertainty. The criterion has been used successfully to evaluate IR absorption bands for oil identification.
To evaluate the importance of n-paraffin weight percent data for differentiation, the average percent standard deviation in the n-paraffin weight percent data was determined to be 1.5% based upon replicate determinations of the n-paraffin weight percent for selected crude oil samples. This value represents a typical uncertainty which can be expected in any analytical laboratory. The weight percentage distribution in the carbon range were staggered, although a pronounced maximum was exhibited by the Delta Field crude. The other crude samples exhibited this phenomenon somewhat in the carbon range. This general feature of n-paraffins distribution can hardly be used for differentiation as the maximum was only pronounced for Delta Field crude and at a specific location. It was neither pronounced nor its location well defined for the other crudes.
Generally, uniform trend is observed for all crude samples whether the samples were weathered or unweathered. The sample which shows some n-paraffin distribution of Tar balls exhibit some important features, viz: low weight percent composition in Carbon range and a preponderance of higher carbon number. This clearly reflects the fact that with considerable weathering, there is further reduction of the lower molecular weight materials in the carbon range investigated. The weight percentage distribution of n-paraffins in the carbon range lies between 4.0% and 12.0%. The maximum range per chemical fingerprint was 2.63 observed only in Forcados bled crude, the range for all crude oils tested were generally less than 1.0. This stability exhibited in the tags justifies their selection for further evaluation in the discriminant analysis.
Isoprenoid compounds are built up from a wide range of linear and cyclic compounds comprising isoprene groups. In crude oils, they occur mostly as fully saturated aliphatic or alicyclic molecules. A regular isoprenoid is an acyclic, branched, saturated molecule with methyl group on every fourth carbon atom. Such an arrangement implies a head to tail linkage of isoprene groups. The commonest isoprenoids are pristine (2,6,10,14 tetramethyl pentadecane) and phytane ( 2,6,10,14 tetramethyl hexadecane). In general, the medium isoprenoids are present in most sediments and oils and are readily analyzed and reported.
The pristine and phytane are the most abundant isoprenoids. These isoprenoid compounds were recorded using the flame ionization detector simultaneously with the n-paraffins. Two dominant ratios have been selected for inclusion in the discriminant test. The ratios of isoprenoid compounds were included in the discriminant analysis as candidated fingerprints for characterization and source identification purposes.
Compound Distribution of Aromatic Fraction.
Fractionation of the aromatic cut into individual components was obtained using HPLC/GC analysis. A series of preliminary studies were carried out using column chromatography, gas chromatography and HPLC analysis to resolve the test samples into individual components. These analysis schemes were employed in the preliminary stage of the analysis to determine the scheme that gives the best result.
The first scheme involved sample elution in a chromatogarphic column packed with porocel clay in the upper half and a 60-100 mesh size silica gel at the lower half of the column. The eluent solvents were n-pentane, 70% benzene-acetone mixture and acetone. The n-pentane eluant is the saturate fraction while the 70% benzene-acetone mixture contains the aromatic compounds. The asphaltenes and heavy molecular weight materials were elites with acetone.
The aromatic fraction was concentrated and charged into the liquid chromatograph for group compound type analysis and was subsequently resolved into individual compound type using gas liquid chromatography.
The two other schemes included direct HPLC analysis of test samples followed by gas-liquid chromatographic analysis of test samples and HPLC/GC analysis of sample distillation fraction. In the second scheme, the crude oil sample was dissolved in n-pentane and filtered. The HPLC runs on the filtrate was followed by GC analysis. In the third scheme, the crude oil sample was distilled and the distillate fractions were subjected to HPLC/GC analysis.
Comparison of the chromatograms from the three schemes shows an increased resolution of components when using the chromatographic column as in scheme one prior to HPLC or GC analysis. In the second scheme an attempt was made towards optimizing the fractionation of the test samples. Test samples were fractionated using HPLC for 1 ring, 2 ring, 3 ring, and 4 ring aromatic fractions. This was made possible by the UV detector with the wavelength at 254 nm. This experiment was carried out several times to collect several portions of the same group of aromatic compounds were pooled together and concentrated by evaporation of the solvent (n-hexane) used in the HPLC analysis. Each of these portions was reinjected into the liquid and Gas chromatograms for detailed resolution of the components. The complexity experienced however, are in the identification of that resolved peaks with known synthetic mixtures. The first scheme was found to provide the best analytical procedure for the quantification of the the compounds necessary for source identification of unknown crude oil sample.
Quantification of Aromatic Compounds.
The universal refractive index (RI) detector provides information on the overall composition of a sample. It also provides indication of the boiling point range of the saturates. Except for some compounds, hydrocarbons with low boiling points have a lower refractive index than hexane. Also, all higher boiling saturated hydrocarbons have a higher refractive index. Thus, a negative peak indicates the presence of hydrocarbons lower and a positive peak, hydrocarbons. If both groups are present an inverted peak is obtained.
The 254 nm UV detector, on the other hand is selective for olefins and aromatics. Moreover, at this wavelength, the extinction coefficient of aromatics increases with the number of aromatic rings and the detector is thus sensitive even to traces of polyaromatics. Because of these large differences in extinction coefficients, the UV response in the aromatic region cannot be directly related to the actual concentrations without prior calibration. Calibration standards were run on the UV detector using benzene, naphthalene, phenanthrene, benzo(a) of analytical quality. Using the calibration graph, the relative abundance of 1 ring, 2 ring, 3 ring and 4 ring aromatic compounds were estimated.
The distribution of aromatic compounds reflect the degree of weathering of oil in the test facility. The baseline being the compound of tar balls used here. The variation in the value of indices resolved about a mean value. The average percent standard deviation in the compound range (2.6%) though higher than the value observed in the n-paraffin compounds (1.5%) was generally within the range expected in any analytical work. The profiles of the weight percent distribution of group aromatic compounds observed in this work is typical of the profiles observed by other workers. No direct comparison was made since the chemical characteristics of the crude oil samples uses in the discriminant analysis ate slightly different.
Generation of Compound Indices.
The discriminant analysis test was conducted using two sets of compound matrices of experimental samples. The first consists of trace elements resolved group compounds of aromatic fraction and selected ratio of dominant n-paraffin compound: ratio of isoprenoids and realized individual compounds of n-paraffin saturates. For the metals, there is generally the prevalence of vanadium over nickel. The values vary from as low as 1.96 to as high as 7.121. The ratios of these tags were taken to avoid problems of compositional changes during weathering. These broad ranges are immediately identified: Delta field crude oil and Forcados blend crude oil have the highest ratios ranging from 6.010 to 7.121. Bonny Light and Umuechem crude oils fall into the second obvious group with ratios in the range of 3.990 to 4.496 while Bonny medium and Elelenwo field crude oils have the lowest range of 1.961 to 2.240. These broad ranges correlate with the theological properties of specific gravity and viscosity of the crude oils. The ratio of these tags is imputed into the discriminant analysis for further differentiation of the samples. Nickel and Vanadium content analyses were not carried out on the tar balls due to limited size for the test and were precluded in the discriminant test.
The total sulphur and nitrogen content were determined in the analysis scheme. Nitrogen content was usually lower than sulphur content in the crude oils test. The ratio of sulphur to nitrogen was taken to avoid the problems of compositional changes during weathering. The variational trend in sulphur/nitrogen ratio is quite divergent. With the exception of Bonny Light crude oil, all the other crude oils from the Eastern Zone exhibit ratios in the range of 6.340 to 7.072. Sulphur/Nitrogen ratio of Bonny Light is in the neighborhood of 1.726 to 1.893. The crude oils from the western alluvia of the Niger Delta were in the range of 1.826 to 1.938. Although the differentiation is sparse, there is however a subtle grouping characteristic of an individual crude oil sample. The ratio was therefore selected for inclusion in the discriminant analysis.
The ratios at selected absorbances were incorporated in the generation of the twelve (12) fingerprint indices for inclusion in the discriminant analysis. Additional compound base was used to test the ability of the discriminatory character of the developed identification system. The data matrices of each of the five export crudes to the United States was used and the data base each comprises of 26 chemical fingerprint indices. This assemblage was used for the discriminant test to differentiate between the crude oils. The characteristic discriminatory power of the chemical fingerprints were compared.