Intelligent FTIR system for efficient fuel analysis
From the very beginnings of fuel characterisation, infrared spectroscopy was widely applied in the field of fuel analysis. Fuel, be it gasoline, diesel or jet fuel, consists of a complex mixture of hydrocarbons of which many have no characteristic absorption in the IR.
Grabner Instruments, AMETEK Oil & Gas Business Unit
Viewed : 895
They constitute the “matrix” or “background”. In contrast, aromatics, olefins and the many additives which are used to compose a fuel show absorption bands and are used to characterise it. The content of aromatics like benzene is strictly monitored by authorities and IR standard procedures exist to assess their concentrations, for example ASTM D6277 or ISO EN238.
The amount of additives used to enhance fuel properties also requires regulation, firstly for optimised blending and secondly for environmental reasons. Various components that enhance the oxygen content and foster a better combustion have been designed and since the chemical bond between oxygen and carbon atoms is nicely accessible with IR radiation, IR spectroscopy became the method of choice to detect such substances (for example ASTM D5845). Currently the US Environmental Protection Agency (EPA) lists close to 7,400 different substances of different brands from different vendors used as fuel additives.
Usually a well defined number of specific additives are used to tune important fuel characteristics, such as the octane number (or cetane number in diesel fuel). Over time it became evident that the spectrum of a fuel must have a correlation with typical additives determining its properties. As the standardised methods to obtain information about fuel properties are time consuming and expensive (for example CFR Knock Engines), chemometric methods linking the IR method with such fuel properties became popular.
In the nineties, Dr. Werner Grabner, founder of Grabner Instruments in Vienna, Austria developed the IROX, the first portable FTIR spectrometer to measure the concentration of Oxygenates in gasoline by mid-IR. Another instrument specialised on Diesel followed in the year 2000. Both instruments are used to measure concentrations of different fuel components as well as predicting classical fuel properties by developing a robust prediction model for various regions around the world.
Towards the middle of the last decade, the classical benchtop IR producers identified the fuel market as a new business area and developed similar specific applications for fuel analysis, and also applied Raman IR spectroscopy as complement to classical IR. Other technologies like the attenuated total reflection (ATR) IR spectroscopy were introduced. FTIR, Raman and ATR technologies however are essentially different ways to look at the same thing from slightly different angels. There is no method better or worse that the other and each technology brings its own merits and drawbacks for the particular purpose it is applied for. ATR can be easily applied in process for online applications, FTIR instruments are typically designed very compact and robust and can be used in mobile fuel labs. The actual differentiators come from: the ease of use, the available support, the flexibility and application range, the quality of the design, the robustness and finally also from the precision and accuracy of the device, which is a direct function of the spectrometer design, its resolution, stability, the spectral range and intensity of the IR light source.
Concentration Measurement and Property Prediction
With modern IR equipment specialised on fuel analysis, taking a measurement and obtaining the concentration of several fuel components as well as properties like octane number, distillation properties or vapour pressure take only a few minutes in comparison to potentially expensive, and often difficult methods defined in the fuel standards. Measuring the concentration of substances, like additives, with infrared spectroscopy is defined in standards, such as ASTM D5845, D6277, EN 238 or EN 14078. In contrast, a correlation for fuel properties is a method very difficult to standardise. The issue is that fuel is such a complex mixture and so different throughout different countries so that establishing a standardised one-fits-all fuel model is not feasible. As a consequence, certain questions arise.
Is the Property Prediction a Measurement in the Scientific Sense?
No, it is not; spectroscopy uses electromagnetic waves that induce electronic transitions in matter, allowing in many cases to identify the matter and assess its concentration within an object composed of different substances. The octane number is not a concentration.
Is Property Prediction Precise?
It depends on the quality of the collected spectra and on the accuracy to which the actual properties where measured. For that reason, various numbers of fuel spectra are collected, analysed, sorted and stored in a database along with their known properties. Through principal component analysis (PCA) those components in the wavelength range can be identified that correlate most with a certain property allowing to establish a prediction model for those properties. If an unknown fuel is then measured the IR spectrometer can use the prediction model to assess RON or distillation properties among others.
Can Property Prediction Beat Standardised Methods in Accuracy?
Yes and no. If the reproducibility (R) of a certain property measured by the conventional standard is not better than Â±1, the R of the prediction cannot be better than Â±1. In contrast, the repeatability (r) which corresponds to the standard deviation of the results of the same sample measured over and again at the same instrument can be better than for the original standard. The precision of the property prediction also depends on the prediction model that was derived from the database. The better the identified areas of certain components of the full spectra correlate with a property the better the prediction. The robustness of the property prediction is also linked to the chemometric methods applied. Nowadays, cluster analysis, multi linear regression and partial least square methods are applied simultaneously to get the best results derived from the existing data.
How Can a Prediction Achieve Robustness?
For the robustness of prediction but also owing to the simpler design of the instrument, NIR devices have been introduced to the market over the past 15 years. Those devices can be made small and portable and can use a longer absorption path. Since the spectra of a fuel in this range shows mainly the overtones of molecule vibrations (especially C-H, O-H and N-H bonds) and different combinations of molecule rotation and vibration modes, distinct bands are difficult to measure, but within the overall feature of the spectrum a robust correlation to fuel properties can be established. Owing to be less informative, NIR devices usually do not measure the concentration of individual substances and for establishing a prediction model, a larger number of different fuels is required.
The mid-IR range is also known as the fingerprint region of the IR spectra. Here individual substances show most of the time, nicely separated bands, which help to identify the substance due to the location of the bands and to assess their concentration. Looking at only a limited number of substances or even just one specific substance like ethanol in gasoline or biodiesel (FAME) in diesel, filter instruments are widely used in the market. Their costs are low and their application limited but they are usually easy to handle, small and compact. They can also develop a prediction model using the correlation found between the spectral ranges their selected filters and typical fuel properties.
Again the limited number of ranges (corresponding to the number of filters) requires a larger amount of collected data to establish a reliable prediction model. Inherent to all filter based IR instruments is that they cannot measure the baseline of a spectral band. In determining the concentration of a substance the background coming from the fuel matrix, the optical components used in the instrument can influence the accuracy. As a consequence, IR spectrometers following this design use dual beam systems to measure at the same time the intensity of incoming (Iin) and transmitted (Iout) IR radiation.
You Cannot Fool Physics
Some physical facts dictate the spectral range of infrared spectrometers. The optical components and the heat of the IR source are essential components of the interferometer design and limit the achievable wavelength range and resolution of a spectrometer design. Firstly, the material of the beam splitter (and cell windows) must allow for the IR light to pass without itself absorbing any IR components. This becomes more difficult in spectrometers with extended wavelength ranges. Most standard FTIR spectrometers for the mid-IR range use the same design. As material for the beam splitter a salt of ZnSe (Zinc selenide) is used. This material allows to pass IR radiation between 2.5Âµm (4000cm-1) and 15.5Âµm (645cm-1). It basically works like a broad pass filter for IR waves. To extend the wavelength range, either the range of this “filter” or the “filter” has to be changed depending on the wavelength currently measured. The latter option is costly and tedious and not practical for routine measurements. Better is the use of special material which is both, inert over degradation and allows passing IR radiation from 1.25Âµm (8000cm-1) to 25Âµm (400cm-1).
Secondly, the best resolution of the spectrum is related to the maximum retardation (Î´ in cm) of the interferogram. The intensity of the beam will be at its maximum where both reflecting mirrors are the same distance from the beam splitter and there is no phase shift between the two parts of the IR light (zero retardation or zero path difference, ZPD). For a defined wavelength it will have its minimum where the moving mirror has passed a distance (Î´) of exactly half the wavelength (Î½) as then the IR wave maximum of one beam meets the wave minimum of its counterpart and they eliminate each other. In fact, the IR sources used to produce polychromatic IR radiation with a wide wavelength distribution.
Add your rating:
Current Rating: 3