I. Litanibarzilai et al., ONLINE REMOTE PREDICTION OF GASOLINE PROPERTIES BY COMBINED OPTICAL METHODS, Analytica chimica acta, 339(1-2), 1997, pp. 193-199
On-line prediction of 10 gasoline properties, such as research and mot
or octane numbers, vapor pressure, API gravity, aromatic contents, etc
., are carried out on-line by a remotely operated detector coupled to
the main control unit by communication optical fibers. This informatio
n is of considerable importance since it is needed for process monitor
ing and for the preparation of final petrochemical products of well de
fined properties. The currently available spectroscopic methods for pr
edicting gasoline properties, which are based on near infrared (NIR) (
and recently on Fourier transform infrared (FTIR) spectrometry) couple
d with chemometric algorithms, provide limited performance (e.g. 0.4 s
tandard error of prediction (SEP), for octane number). We propose an i
mprovement of the performance by applying combined optical methods. In
this study we combine on-line information from a short-wave NIR photo
diode array spectrometer (700-1000 nm) with laser induced fluorescence
(LIF) spectra obtained with a PC-plugged in linear CCD spectrometer.
UV excitation is performed by harmonics of a compact and low-cost Nd:Y
AG laser, and the system is remotely operated through optical fibers.
A comparison between the octane number predictions by NLR spectrometry
and by LIF from third and fourth harmonics, is provided. It is shown
that the addition of the fluorescence information improves octane numb
er prediction (<0.2 SEP). These two setups (NIR and LIF) can be combin
ed into one integrated system, based on common optical fibers and dete
ctor.