Method and application to determine the amount of oil or condensate in water or in water-based samples with the help of an extractive agent

The determination of the amount of oil in water using spectroscopic measurement methods first requires that the oil components be extracted from the water phase. This in turn requires that the extractive agent be insoluble in water while at the same time be able to effectively extract oil components from the water phase. Following this, the concentration is increased through, i.e., evaporation. The extractive agent must be relatively transient so that it may be evaporated without too much of the oil components being lost as a result of this transience. Following the concentration increase of the extractive agent, the sample is then ready for quantitative analysis with the help of vibration spectroscopy. The combination of Freon (Dichlorodifluoromethane) as an extractive agent and the so-called Fourier Transform Infrared Spectroscopy (FTIR) has been a very popular method for quantitative analysis of oil in water for a long time. Freon detaches the oil components quite well and does not interfere with the FTIR signal within the spectral area in question. This provides the opportunity for a quantitative analysis of oil in the FTIR even though the amount of Freon is still very high. The problem is that Freon is not environmentally-friendly as it breaks down the ozone layer, and is currently no longer in use or being phased out. (Please refer to The Vienna convention for the protection of the ozone layer (2001), ISDN 92-807-2121-6).

Phasing out Freon and other products containing halogen hydrocarbons as an extractive agent for oil in water analysis has created a problem in using vibration spectroscopic quantitative analysis. It appears that good and effective extractive agents used to get oil components out of the water phase also powerfully interfere with the oil components in the subsequent vibration spectroscopic analysis. For example, weak or slightly insoluble oil components with a boiling point greater that approximately O ⁰ C and below approximately 17O ⁰ C, are especially suited to extraction. Examples are butane, pentane, hexane, isooctane, cyclohexane and toluene.

However, an extraction with one of these oil components will give off a very powerful signal in a vibration spectroscopic analysis, which will entirely camouflage the oil containing components extracted from the water phase. Interference occurs especially when the solvent's molecular structure contains one or more carbon-hydrogen bonds, called hydrocarbons. A quantitative analysis of components containing oil in such an agent within selective spectral areas is therefore, according to available literature on the subject of spectroscopy and the general understanding of all professional groups, considered to be impossible. It is therefore advised that a 100% evaporation of hydrocarbon solvents be carried out prior to the spectroscopic analysis. A good example of such an analyzer is described by Wilks Enterprise for their TOG/TPH analyzer (www.wilksir.com).

The main purpose of this present invention is twofold: 1 ) to determine whether there exist spectral areas where it is possible to find selective information about other oil components besides the hydrocarbon-based solvent agent and, 2) to thereafter investigate whether this selective area can be used to quantify oil components with the help of FTIR spectroscopy even if the extractive agent containing hydrocarbons is present in the sample cell.

To investigate if selective spectral information existed, base crude from the North Sea was dissolved in distilled water, approximately (but as accurately as possible) 10, 20, 30, 50, 70 and 90 mg of base crude per litre of water. Then, 30 ml of isooctane was added to each water sample; the mixture was shaken and the isooctane phase was separated from the water phase. The isooctane phase was then evaporated to precisely 2 ml, and analyzed in an FTIR instrument. The spectral data was analyzed by the software MUST Analyze! (www.must.as). A traditional multivariate statistic method was applied as a principal component analysis to search for selective spectral areas. In addition to repeating these tests and at different concentration levels, the tests

were also verified by repeating the experiments with a pale, light condensate from the North Sea. Both series of experiments gave surprisingly good results by indicating a spectral selective area for oil components. A selective spectral area in the FTIR for oil components with a hydrocarbon- based solvent present in the test cell has so far not been mentioned in specialized literature. In addition to this, good quantitative connections were uncovered between the size of the signal in these spectral areas and amount of oil components (base crude or condensate) added to the water samples.

These two examples below detail exactly what was done, and how the problems involved in quantifying oil components through the use of FTIR spectroscopy were solved, once oil components were dissolved in a solvent containing carbon and hydrogen. The protective scope and the special characteristics belonging to this invention are described in the enclosed patent claim.

The main feature of this invention is quantitative determination of oil components through the use of FTIR, when the oil components are present in a solvent which includes one or more hydrocarbons.

A special feature of this invention is that it allows for the quantification of both pale, light condensates as well as heavier, darker oils.

The invention is further elaborated in connection with the following examples, one for oil and one for condensate. Both sample types were prepared as described above, and analyzed in the FTIR (0.1 and more NaCI cell) in an isooctane extract evaporated to precisely 2 ml. All the spectral data with all wavelength intensities ranging from 400 to 4000 cm ^'1 were analyzed.

Table 1: Weight-in of oils and condensates

* Three replica samples were run in the FTIR instrument Example 1

Table 1 presents the analysis spectra from eight samples with oil components and also two clean spectra of isooctane. As we can see from figure 1 , there appear to be no differences between the samples. In other words, the quantitative information regarding the oil components is drowned out by the signal from the isooctane solvent in the entire spectral area, as indicated by available literature on spectroscopy.

Surprisingly, using a principal component analysis, it was discovered there were nevertheless, systematic differences between the samples in the spectral area spanning for example, 650 to 1120 CM ^"1 . This region is shown in figure 2. One of the isooctane spectres (oil components = 0 ppm) is marked with a black dotted line together with the spectre of the number of oil components equal to 92 ppm of oil. As clearly indicated by the table, the other test results fall within these two values. At the same time it appears that the more the oil in the water sample, the more systematic elevated the signal.

By plotting the signal against concentration, a calibration line may be drawn as shown in Figure 3. An R ² equal to 0,996 gives a very good linear model.

Surprisingly, there were several local areas within the spectral area of 400 to 4000 cm ^"1 , which showed the same selectivity as the spectral area ranging approximately from 650 to 1120 cm ^"1 . Although we have elected not to show all the areas here, the conclusion is that it is entirely

possible to find quantitative information on oil components from the FT)R spectra, even when a hydrocarbon as is used as a solvent.

Example 2

The test described for the oil from the North Sea, was repeated on a light condensate from the North Sea. Spectra showing all samples are identical to those shown for isooctane and oil in Figure 1 , and are not shown here. As was the case for experiment spectra, these samples were analysed using an ordinary principal component analysis. Once again the patent applicant was surprised by the good selectivity ranging in the area from 650 to 1120 cm ^"1 . As was the case for spectra from oil components in isooctane, here too, other selective areas were discovered in the range from 400 to 4000 cm" ¹ ; conflicting with commonly accepted textbook learning. Figure 4 shows the quantitative information from the selective area ranging from approximately 650 to 1120 cm ^"1 . By plotting the signal against the concentration value, a very good linear model is achieved (R ² for the entire 0,967) (see Figure 5).

The two examples unexpectedly show that it is fully possible to identify selective areas for oil and condensate areas within the FTIR spectra, even when hydrocarbons are used as an extractive agent and solvent for the FTIR instrument.

Further, the patent description shows that it is relatively easy to develop quantitative models once the selective areas have been determined. Such models have been presented in a simple form, by selecting an average signal of 727 cm ^"1 +/- 10 wavelengths without any pre-processing of the spectra. Tests have shown that far more precise and robust methods can be achieved through the application pf suitable pre-processing in combination with multivariate regression methods. Though not shown here, the main purpose of this invention is to prove that there exists selectivity in oil-/condensate analyses on the FTIR, even when hydrocarbons are used as extractive agents

and solvents. Moreover, this invention shows that these selective areas are suited for the collection of quantitative information concerning oil/ condensate components in the area ranging from 0 to 100 ppm. In practice, oil/condensate samples should first be analyzed in an FTIR instrument pursuant to the prescribed method, in order to determine the optimal area for quantification. It will be possible to choose one or several areas ranging from 400 to 4000 cm ^"1 , and also a multivariate method to ensure optimal precision and robustness. Thereafter, a calibration ¹ model shall be prepared where the method is further modified and calibrated towards an approved/validated method, such as i.e., GC-FID.

We have shown the possibilities for quantitative determination through the use of isooctane as an extractive agent and solvent for the FTIR. Isooctane, i.e. 2,2,4 trimethylpentane, is a branched paraffin (the molecule does not contain a straight chained CH ₂ -CH ₂ fragment), and therefore, the selective signals observed in the spectral areas most likely stem from the components in oil condensate which are not 100 % branched paraffin. Correspondingly, other selective areas will appear when hexane, cyclohexane or toluene are used as solvents. We have shown though that even when using hydrocarbons as solvents, it is nevertheless possible to identify selective spectral areas in an FTIR spectre.