MODIFIED MAT AND GC HELP PREDICT FCC GASOLINE QUALITY

March 26, 1990
Carolyn A. Yatsu, Donald A. Keyworth Akzo Chemical Division Bayport, Tex. The modified micro-activity test (MAT), which differs from the ASTM MAT, has higher reactor temperatures, shorter oil feed durations, and better reactor design. In combination with improved gas chromatographic (GC) analysis of the MAT liquid, the modified MAT is a powerful approach to predicting the effects of FCC variables on FCC gasoline quality. The approach can be used to provide GC-RON, GC-MON gasoline front-end

Carolyn A. Yatsu, Donald A. Keyworth
Akzo Chemical Division
Bayport, Tex.

The modified micro-activity test (MAT), which differs from the ASTM MAT, has higher reactor temperatures, shorter oil feed durations, and better reactor design.

In combination with improved gas chromatographic (GC) analysis of the MAT liquid, the modified MAT is a powerful approach to predicting the effects of FCC variables on FCC gasoline quality.

The approach can be used to provide GC-RON, GC-MON gasoline front-end olefinicity, mid-cut branching, aromaticity, and GC-bromine number.

Refiners and FCC catalyst manufacturers have a considerable incentive to extend the capability of the MAT (microactivity test) so that the test can provide information about FCC gasoline quality as measured by octane performance factors and bromine number.

Older versions (ASTM D-3907-80) of the MAT, which employed low reactor temperatures (4801 C.) and long feed delivery times (75 sec), do not make a gasoline fraction in the MAT liquids product that is characteristic of gasoline from today's commercial FCC units. These older versions of the MAT, except for equilibrium catalyst characterization, have been generally replaced by the modified MAT (Mod MAT).1 2

The Mod MAT used in this work has been discriminative not only in activity and selectivity performance differences when catalyst composition changes are studied, but the test has also been predictive of differences in the refiner's FCC gasoline quality resulting from the change in the refiner's FCC feed or FCC catalyst.

An example of the prediction of an octane performance change with the MAT used in this study, if a refiner changed from one catalyst to another, is shown in Table 1. The magnitude between the predicted and actual refiner's octane number change is for the two catalysts being compared, which is, at best, a simulation to that which will occur in the customer's FCC unit.

Differences between the test feedstock and the refiners feedstock, and other differences, such as those between a fixed-bed MAT unit and a commercial scale fluidized FCCU, also affect predictions between the MAT simulation and the actual change the refiner will experience.

The modification of the MAT, used in these studies, increased the MAT reactor temperature to 510 C., used catalyst-to-oil ratios of 3, 4, and 5, and shortened syringe drive times to 25 sec. The Mod MAT reactor design also avoided a plug of catalyst in the reactor. The plug is associated with poorly controlled bed-temperature gradients from the endothermic cracking reaction.

In the MAT modification used, the catalyst is retained as a thin layer between a hot core (dead-man) and the reactor walls. In this modification of the MAT test, as well as in other modifications, approximately 1 g of total liquid product is obtained.

A great amount of information about the gasoline quality in the MAT liquid product can be accessed, even on such a small amount of product, using GC analysis and correlations that are extensions of a GC procedure published over a decade ago by Anderson, Sharkey, and Walsh.3

Anderson and coworkers divided the GC chromatogram for the gasoline into 31 groups.

Each group was defined by boiling range and usually contained compounds of similar chemical class. When the groups were multiplied by factors, which were experimentally determined by regression, the products were then summed.

The sum could be correlated to give GC-RON (research octane number). The approach was extended to the present work to include GC-MON (motor octane).

In our further extensions of Anderson's correlations, determinations of front-end olefinicity of the gasoline, branching of the paraffins in the mid-cut, light aromatics, and bromine number could also be made.

The importance of changes in the levels of classes of hydrocarbon types such as olefins, branched and cycloparaffins, and aromatics on gasoline quality is shown from octane number performance of compounds within such classes (Table 2).

A study of such tables leads to generalities as follows:

  1. C5's are good octane performers irrespective of olefinicity or branching.

  2. C6's and C7'S often (but not always) benefit from olefinicity, and always from branching.

  3. Multi-branching, which is not a characteristic of FCC catalysis, is needed for good octane number (ON) performance of C8's.

  4. C9+ paraffins and olefins in FCC gasoline are very poor ON performers.

  5. Cycloparaffin C8+ counterparts are much better, but still poor compared to aromatics.

  6. All chromatics are good ON performers, and C8 and C9 aromatics are best.

  7. Paraffins and olefins ON performance gets worse with increasing boiling point and molecular weight.

GC OCTANES

The original Anderson GC octane method divided the chromatogram for the refiner's pool gasoline into 31 groups.1

The coefficients were developed for each group which, when multiplied by the total GC area for the group, gave the contribution of that group to gasoline octane performance.

When the octane contributions for the groups were summed, the GC-RON (GC research octane number) and GC-MON (GC motor octane number) of the gasoline could be calculated.

The 31 Anderson groups can also be used to characterize the FCC gasoline in other ways. For example, FCC gasoline octane performance may benefit from front-end olefinicity, mid-cut branching, and light aromatics.

All of these gasoline components can be estimated by selective use of the Anderson groups.

The light aromatics, as estimated with the Anderson groups, relate to total aromatics in the gasoline because of the binomial distribution relationship of methyl benzenes 5 and other aromatic series in FCC gasoline.

Front-end olefinicity = G9/G10 (1)

where: Group G9 contains C6 olefins and Group G10 contains hexane.

Degree of mid-cut branching = (G14 + G19 +G25 + G27)/(G22 + G16) (2)

where: G14 is C7 isoparaffins, G19 is C8 isoparaffins, G20 is C8 isoparaffins, G25 and G27 are C8 isoparaffins and naphthenes, G16 is n-heptane, and G22 contains n-octane.

Light aromatics = G12 + G18 + G24 + G26 + G28 (3)

where: G12 is benzene and methylcyclopentene, G18 is toluene, G24 is ethylbenzene, G26 is p-xylene and mxylene, and G28 is o-xylene.

GC OCTANE PROCEDURE

Anderson's coefficients were initially obtained by regression of GC and engine ON data for gasoline blend pools from various refineries. Anderson used 19 marker compounds to define the groups in terms of GC retention times.

Many modifications of the original method have been attempted by further subdivisions of the groups, and some laboratories have gone so far as to attempt to assign a coefficient to each GC peak.

Using a statistical software package, RS-1, coefficients were obtained for 28 groups by regressing engine octane data and GC-octane data on 205 FCC gasolines from Akzo's pilot plant. Using these coefficients, much better agreement was obtained between the GC-octane number and engine octane number over the case where the original Anderson coefficients were used.

Only 28 groups were used in our regression because the first three Anderson groups are for C4's. We defined gasoline as C5 through C12.

The group composition originally reported by Anderson has been shown with modern chromatographic resolution to be sometimes different than originally thought. For example, Group 12 was thought by Anderson to be composed of a single member, benzene.

In FCC gasoline, Group 12 is usually composed of two members, benzene and methylcyclopentene, in about equal amounts.1

The significance in Equations 4 and 5 of the plus or minus before the coefficients is that the group's contribution exceeds, or is less than, the average value of the 205 gasolines for all the components-88.5 for RON and 78.6 for MON.

The size of the coefficient is the measure of the degree of enhancement or detraction to octane performance by the group over the average value.

The equations developed by our regressions for GCRON and GC-MON using these 28 groups are as follows:

GC-RON = 0.090922*G4

+ 0.067878*G5

- 0. 191578*G6

+ 0.248133*G7

- 0. 157002*G8

+ 0.391013*G9

- 0.338010*G10

+ 0.305454*G11

+ 0.326178*G12

- 0.0.682025*G13

- 0.431378*G14

- 0. 260932*G15

+ 1.114104*G16

+ 1.156157*G17

- 0. 171916*G18

- 1.022795*G19

- 1.09423*G20

- 0.048515*G21

- 0.601929*G22

- 0.215253*G23

+ 4.752491*G24

+ 5.599282*G25

+ 0.470219*G26

+ 0.080736*G27

- 0.56108*G28

- 0.213607*G29

+ 0.018005*G31

+ 88.511047 (4)

GC-MON = 0.009832*G4

+ 0.039251*G5

- 0.049166*G6

+ 0.040748*G7

- 0.147158*G8

- 0.541236*G9

- 0.306311*G10

+ 0.34681*G11

+ 0.164487*G12

- 0.846784*G13

+ 0.071437+X14

+ 0.203711*G15

- 0.887515*G16

- 0.022054*G17

- 0.22866*G18

+ 0.258998*G19

- 0.401056*G20

- 0.494713*G21

+ 0.33126*G22

- 0.653744*G23

+ 0.627641*G24

+ 0.608849*G25

+ 0.674251*G26

+ 0.247123*G27

- 0.053825*G28

- 0.003586*G29

+ 0.079357*G30

+ 0.017698*G31

+ 78.619465 (5)

The correlation for pilot plant FCC gasoline GC-RON to engine test RON and MON is shown in Figs. 1 and 2. The standard deviation for the GC-RON and MON was 0.35 and 0.53 respectively. The standard deviation for the ASTM engine RON and MON is 0.25 and 0.30 respectively.

GC-RON and MON with the correlation procedure can be run on the MAT liquid product; the ASTM engine RON or MON needs a bigger sample and therefore requires separation of the gasoline portion from the liquid product. The GC method can be applied to the MAT liquid product, pilot plant gasolines or undistilled liquid products, and FCCU gasolines.

The correlation of octane performance, as measured by the engine ON or the GC ON, to light aromatics is particularly striking. For FCC gasoline generated in a pilot plant under a variety of conditions, including different reactor temperatures, catalyst-to-oil ratios, and using different catalyst types, when light aromatics increased, so did octane performance (Fig. 3).

EQUIPMENT USED

Two chromatographs were used interchangeably; the Hewlett Packard HP5890 with a carousel and auto-injector; and the Varian Model 6000, similarly equipped. Both used the FID detector and a split/splitter ratio of 1:100. A nitrogen carrier gas with a flow of 28 ml/min was used.

The GC column was a 30 m SPBL (cross-linked methyl silicone) fused silica capillary column. The temperature program follows: 4 min at 0 C.; then ramp to 150 C. at a rate of 5 C./min; then ramp to 300 C. at a rate of 7 C./min; and hold at 300 C. for 10 min.

Integration and preliminary calculations were made using a Hewlett Packard data station with an HP 3359 data processor. A software program in Fortran scrolls through the GC data using pattern matching to define the groups and sum the peak areas from C5 through C12.

C12+ peaks are summed for the balance of the liquid product. The data are transmitted to the IBM network where GC-RON's and GCMON's are calculated.

BROMINE NUMBER CORRELATION

Olefins are of concern to West Coast refiners because olefins from gasoline evaporative emissions have been identified as contributing to air pollution. Particularly the lighter olefins in gasoline can evaporate and, in the gaseous state, are photochemically active.

Bromine number reflects gasoline olefinicity. Bromine number (ASTM Di 159) is measured by adding bromine in an inert solvent to the gasoline sample. The number of grams of bromine added per 100 g of sample is defined as the bromine number.

Olefins as measured by MS or GC do not correspond exactly to bromine number because stearically hindered olefins, n-alpha olefins, conjugated diolefins, and substituted cyclo-olefins, may show low bromine numbers; and because sulfur and N compounds in gasoline may react with bromine, thereby increasing the measured bromine number.

In spite of these considerations, bromine number tends to reflect olefinicity as an MS analysis showed for olefins in a light naphtha and heavier naphtha fraction of an FCC gasoline. The bromine value for the cuts within this fractionated FCC gasoline all correlated fairly well with bromine number (Table 3) when an MS bromine value was calculated using a simple equation:

Bromine value

MSO/Cn * 14

where: MSO is mass spectrometry olefins (ASTM D2789), and Cn is the number of carbon atoms in the molecule, estimated from the mean molecular weight, which is calculated from the boiling range.

As the fraction analysis shows, FCC catalysis makes mainly mono-olefins. Because diolefins and higher MW mono-olefins tend to crack to lighter products most of the olefins which survive secondary reactions are found for FCC gasoline predominately in the lower boiling fractions.

There are some assumptions involved in assigning Cn values to the boiling ranges, but when this is done the bromine number measured by the ASTM D1159 method and the bromine value calculated from MS shows reasonably good correspondence for all of the fractions.

Because some of the GC Anderson groups contain mainly olefins, a correlation was sought to calculate a GC bromine number like that calculated from MS data. Anderson Groups 5, 7, and 9 contain C5 and C6 olefins; Group 11 contains C6 and C7 olefins (as well as cycloparaffins and isoparaffins). Several combinations of groupings were tried to optimize the GC correlated bromine number prediction.

For 18 pilot plant gasolines, the following equation gave a correlation coefficient of 95%:

GC-Bromine Number 2.046*G7 + 14.95*G9 + 3.734*G11 - 11.43

A comparison of GC bromine values from this equation to actual bromine numbers is shown in Table 4. The standard deviation based on 56 determinations for the GC bromine number is 5.26; for the ASTM method (D-1159) for the same bromine number range (20-90) the standard deviation is 0.5-1.8.

A second set of 36 different FCC gasolines was selected and analyzed by the GC bromine number method to further test the accuracy of the equation. The GC bromine number results for both sets are compared in Fig. 4 to the ASTM D1159 bromine number values, where the circles represent the first data set comparison, and the squares represent the second data set comparisons.

Because many test facilities will want to obtain GC RON and MON results for all gasolines made in their MAT units, and will already have Anderson group data for these calculations, it becomes an easy matter to get GC bromine numbers.

The Mod MAT then becomes a very useful tool to refiners and catalyst manufacturers for studying the effect of feedstocks and catalyst changes on octane performance, benzene in the gasoline, olefinicity, aromaticity, mid-cut branching, and bromine number.

BROMINE NUMBER FACTORS

Some major influences on FCC gasoline olefinicity are feedstock and catalyst type and FCCU operating factors which include reactor temperature, oil partial pressure, and residence time. Catalytic and feed effects can be studied in either the Mod MAT or the FCC pilot plant.

Pilot plant studies can generate enough product so a bromine number and a GC-bromine number can be obtained on the same sample. When such studies were done, important feed and catalytic effects on the FCC gasoline's bromine number were observed (Table 5).

Some observations from this study, and from results from Table 3, are that FCC gasoline's bromine number usually is in the range of 35-95. The GC bromine number usually, but not always, agrees well with the ASTM bromine number,

Increasing catalyst-to-oil ratio (and conversion which results in overcracking of the gasoline range olefins) decreases bromine number. CREY catalysts make less olefinic gasoline than octane catalysts (probably because of higher hydrogen transfer rates and the overcracking effect).

FCC gasoline from an atmospheric gas oil with high sulfur had a higher chemical bromine number than the GC-bromine number. The GC method can discriminate between light olefins, which are thought to be of more concern than heavier olefins because of their volatility. The bromine number method does not make this discrimination.

REFERENCES

  1. Hartkamp, M. B., "The Micro-Simulation Test," Akzo Catalysts Symposium, F-3, The Netherlands. June 1988, pp. 1-9.

  2. Carter, G. D., and McElhiney, G., "FCC Catalyst Selection. A Modification of the ASTM microactivity Test (MAT) Provides Useful Information," Hydrocarbon Processing, September 1989, pp. 63-64.

  3. Anderson, P. C., Sharkey, J. M., and Walsh, R. P., "Calculations of the Research Octane and Motor Gasolines from Gas Chromatographic Data and a New Approach to Motor Gasoline Quality Control," Journal of the Institute of Petroleum, Vol. 52, 1972, pp. 83-93.

  4. API Technical Data Book, Petroleum Refining, Vol. 1, 4th ad., Table 1C1.1 - 1C1.17, pp. 1-22 1-91, from API Research Project 44, "Data on Hydrocarbons and Related Compounds," American Petroleum Institute, Washington D. C., 1982.

  5. Keyworth, D. A., Reid, T. A., and Wilson, J., "Legislative and Market Trend's Effect on Catalyst Selection," Paper No. AM-88-69, National Petroleum Refiners Association annual meeting, March 1988.

  6. Unzelman, G. H., "U.S. gasoline pool octane increase may be limited," OGJ, Apr. 4, 1988, p. 35.

Copyright 1990 Oil & Gas Journal. All Rights Reserved.