Académique Documents
Professionnel Documents
Culture Documents
www.fuelfirst.com
Abstract
A new simple no-linear calculation method of isomerisation gasoline Research Octane Number (RON) is presented. The calculation
method effectively utilizes the compositional data from high-resolution capillary GC analysis, and the measured pure and blending RON
values of various HCs, which are widely published. The method, which is a simple non-linear method, utilizes calculable weighting factors,
which are specific for each gasoline blend, and showed an excellent agreement with the RON values of various refinery isomerates samples as
they measured in a standard CFR engine.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Non-linear methods; Gasoline; Isomerates; Octane number calculation; Gas chromatography
The fast, easy and accurate determination of octane (iso-octane) is equal to 100 (the octane number of pure iso-
number is important for refiners today, since optimisation of octane).
refining process at a reasonable cost is an ever-increasing Eq. (2) is generally extended and used for a gasoline
requirement. consisting of k components as:
where y1 and y2 are the volume fractions of n-heptane and Recently, and with the continuous advancement of
iso-octane in the blend respectively, RON (n-heptane) is computational power, some new and quite promising non-
equal to 0 (the octane number of pure n-heptane) and RON linear utilisation methods, based on pattern recognition,
N. Nikolaou et al. / Fuel 83 (2004) 517523 519
Table 1
Octane number data for various HCs present in isomerization gasolines
Ordered by retention time Boiling point (8C) Type of H/C Molecular weight Pure RON Pure MON Blending RON
have been proposed [7] for the prediction of octane analysed by the GC, as long as there are data available for
number. their pure and blending octane numbers.
In the new calculation method proposed by the authors,
the following equations for the prediction of isomerisation
gasoline research octane number (RONG) are used. Similar 2. Experimental
equations can be equally applied for MON:
The demand for increased octane number in gasoline
X
N
production from refineries in consequence of the reduction,
RONG Ki RONi yi 4
under the new specifications, of high octane ratings
i1
aromatics and olefins led to the investigation of any possible
where yi ; are volume fractions of HCs present in the alternative processes for octane number improvements. For
gasoline and obtained by GC, RONi is the measured example, one of the best processing options for light straight
(published) pure HC research octane number and Ki is a run (LSR) has been determined to be the isomerisation.
weighting factor that is calculated from: Isomerisation units rearrange the molecules from straight-
chain, low-octane hydrocarbons to branched-chain, high-
XN octane hydrocarbons known as isomers. This is done for two
RONi yi BRONi
Ki Xi1
N 5 reasons:
BRONi yi RONi
i1
They create extra isobutane feed for alkylation.
where BRONi is the measured (published) blending They improve the octane number of straight run pentanes
research octane number of each HC. and hexanes and hence make them into better petrol
The product Ki RONi in Eq. (4), gives a calculated blending components. So, the resulting isomerate is a
effective or blending octane number, characteristic for the superior gasoline blending stock.
specific gasoline, which also depends, except on the known
pure and blending octane numbers, on the number, type and Twenty (20) pure refinery isomerates samples, with
concentration of the HCs present in the specific gasoline. different number and components concentrations were
Thus there is not limit in the number of the HCs present and collected and their research octane number (RON) was
520 N. Nikolaou et al. / Fuel 83 (2004) 517523
Fig. 1. A typical gas chromatogram obtained for an isomerization gasoline sample and column/detector characteristics.
carefully measured in a standard CFR engine. The same the gasoline was quantitatively determined. In Table 3 are
samples were analysed in a PERKIN ELMER 8700 GC with shown typical calculated weighting factors and effective
a Petrocol DH by Supelco, 150 m capillary column, which is research octane numbers for a specific sample. The
known to have an excellent HCs separation performance. A weighting factors Ki calculated values were close to one.
summary of the GC operating conditions and capillary As it would be normally expected, for a pure component
column characteristics is provided in Fig. 1. Standard yi 1 from Eq. (5) we obtain Ki 1 and subsequently
analysis calculations of area% (normalization) in a standard from Eq. (4) RONG is equal to the pure RON.
integrator incorporated into the chromatograph were used.
The GC TALK software available in the 8700 GC used for
the transfer of compositional data through an RS232 to a
PC, where they were fatherly processed with a Visual Basic Table 2
Typical composition (%w/w) of detected HCs
program.
1 n-C4 0.320 14 2,2,3-TM-C4 0.013
2 2,2-DM-C3 0.031 15 3,3-DM-C5 0.026
3 iso-C5 (2-M-C4) 41.596 16 Cyclo-C6 2.677
3. Results 4 n-C5 12.323 17 iso-C7 (2-M-C6) 0.076
5 2,2-DM-C4 25.818 18 2,3-DM-C5 0.032
6 Cyclo-C5 1.037 19 1,1-DM-cyclo-C5 0.020
A typical chromatogram from a full range isomerisation
7 2,3-DM-C4 2.971 20 3-M-C6 0.066
gasoline blend is shown in Fig. 1 and a typical composition 8 iso-C6 (2-M-C5) 5.398 21 1,3-DM-cis-cyclo-C5 0.020
with the HCs peaks allocated also shown in Table 2. The 9 3-M-C5 2.200 22 1,3-DM-trans-cyclo-C5 0.024
RON and various other data for these HCs are also present 10 n-C6 2.869 23 3-E-C5 0.030
in Table 1. 11 2,2-DM-C5 0.054 24 n-C7 0.029
12 M-cyclo-C5 1.221 25 M-cyclo-C6 0.493
A maximum number of 26 different HCs was detected
13 2,4-DM-C5 0.033 26 E-cyclo-C5 0.011
and allocated. In all the samples more than 98%w/w of
N. Nikolaou et al. / Fuel 83 (2004) 517523 521
Table 3
Example of calculated weighting factor and effective research octane numbers
In Fig. 2 are shown the RON values measured in the By employing Chebyshevs probability theorem [14],
CFR engine and those calculated from the GC data with which holds for any distribution of observations:
the new method proposed. The agreement was very good.
1
Considering the RON uncertainty of the ASTM method P0 2 ksDRON , DRON , 0 ksDRON $ 1 2 6
to ^ 1 RON, and the fact that the standard deviation k2
of the difference DRON RON(ASTM)-RON(GC) was and for an uncertainty ksDRON ^1 RON; we can safely
sDRON 0:23RON it is easily justified the consistency of conclude that at minimum 95% of our observations are
the estimation. falling within the ^ 1 RON interval, which is a very good
Fig. 2. Measured RON (ASTM) versus calculated RON(GC) from GC data for 20 isomerates. (Uncertainty of RON(ASTM) taken as ^1RON).
522 N. Nikolaou et al. / Fuel 83 (2004) 517523
[7] Meusinger R, Moros R. Application of genetic algorithms and neural [11] Durand JP, Boscher Y, Dorbon M. On line chromatographic analyser
networks in the analysis of multi-component mixtures using NMR for determining the composition and octane number of reforming
spectroscopy. 1995. www2.chemie.uni-erlangen.de/external/cic/ process effluents. J Chromatogr 1990;509(1):47 51.
tagungen/workshop95/meusinge/index.html. [12] Lugo HJ, Ragone G, Zambrano J. Correlations between octane
[8] http://www.process-nmr.com/chemo.htm (2003)-NMR Chemometric numbers and catalytic cracking naptha composition. Indust Engng
modeling. Chem Res 1999;38(5):2171 6.
[9] Protic-Lovasic G, Jambrec N, Deur-Siftar D, et al. Determination of [13] ASTM STP 225 -API research project 45, Knocking characteristics of
catalytic reformed gasoline octane number by high resolution gas pure hydrocarbons; 1994.
chromatography. Fuel 1990;69(4):525 8. [14] Walpole RE, Myers RH. Probability and statistics for engineers and
[10] Albright LF, Eckert RE. New equations help rapidly determine scientists, 5th ed. ; 1993. 002-424201-2.
alkylate octane numbers. Oil Gas J 1999;97(3):514.