Vous êtes sur la page 1sur 4

17 WORLD PETROLEUM CONGRESS BLOCK 2: Excelling in Refining and Delivering Quality Petrochemicals FORUM 4: The Future of Aromatics in Motor

Fuels and its Impact on Refining and Petrochemistry


Dauzacker, E. R.1,2; Pinto, J. C.1 and Monteiro, J. L.1 1- Programa de Engenharia Qumica/COPPE, Universidade Federal do Rio de Janeiro, Brazil; 2- CENPES/PETROBRAS S.A., Brazil INTRODUCTION
The rational usage of energy resources and the reduction of environmental pollution become more important every day. There is a claim for the production of better quality fuels and the increase of the efficiency of automotive vehicles. Considering these two aspects, the use of diesel fuel is advantageous when compared to gasoline, because diesel engines allow greater compression ratios, resulting on better thermodynamic efficiencies. It is then possible to consume less fuel, which leads to reduction of CO2 emissions and consequently to reduction of global warming effects. Regarding the specifications of diesel fuels, large reduction of sulfur contents was imposed in the past. Nowadays, the allowed sulfur content keeps diminishing, while additional restrictions, which include the aromatic content, are being defined. The hydrodearomatization (HDA) process is an alternative to meet the recent diesel fuels specifications. In general, HDA studies are performed on bench-scale differential plug flow reactors, using mixtures of key aromatic compounds as models to analyze reaction rates. In this work, however, actual refinery products are used to build a mathematical pseudo-phenomenological kinetic model, which is employed to simulate and optimize the HDA process. The elaborated methodology allows the prediction of mono-, di- and polyaromatics concentration on products as functions of the feedstock composition and of the reactor operation conditions.

Six real industrial feedstocks obtained from distinct oil sources and refinery processes were used in this work: kerosene, light gas oil (LGO) and heavy gas oil (HGO) from atmospheric distillation; light and middle gas oil (LGO/MGO) from delayed coking; light vacuum gas oil (LVGO) from vacuum distillation; and light cycle oil (LCO) from fluid catalytic cracking. Commercial Ni-Mo-S/Al2O3 catalysts were used for HDA tests. Catalysts were sulphided in-situ on the pilot plant, by passing a stream of hydrogenated spindle doped with CS2 (20000 ppmw) for 8 h. The reactor was heated by six independently controlled electric furnaces, which guaranteed the isothermal operation over the whole catalyst bed. The reactor was loaded with 310 cm3 of catalysts and an up-flow arrangement was used to minimize back-mixing problems. The liquid product was collected into a refrigerated recipient for analysis. The gas product was analyzed in-line with a wet gas meter (WGM) and a process chromatographer. A total of 79 tests were performed in a completely automated pilot plant. Replicates and blank tests were carried out to guarantee that catalyst remained active. 22 tests were used for independent validation of the model and were not used for parameter estimation (1). The oil compounds were identified by group type analysis (GTA) and grouped into eight distinct families: poly-aromatics (PA); di-aromatics (DA); mono-aromatics (MA); naphthenes (NP); paraffins (PF); dibenzothiophehes (DT); benzothiophenes (BT) and tiophenes (TP).


Based on available literature information proposed to describe the HDA reactions:

, the reaction mechanism presented in Figure 1 was

kPA kPAr

kDA kDAr









Figure 1 Proposed reaction mechanism for the HDA process. Based on Figure 1, the following set of pseudo-phenomenological mass balance equations may be written:

dC PA (1) = k PA C PA p H 2 + k PAr C DA dt dC DT = k DT C DT p H 2 (2) dt dC DA = k DA C DA p H 2 + k DAr C MA + k PA C PA p H 2 k PAr C DA + k DT C DT p H 2 dt


dC BT = k BT C BT p H 2 (4) dt dC MA = k MA C MA p H 2 + k MAr C NF + k DA C DA p H 2 k MAr C MA + k BT C BT p H 2 dt


dC NF = k NF C NF p H 2 + k MA C MA p H 2 k NF C NF dt dCTF = kTF CTF p H 2 dt dC PF = k NF C NF p H 2 + kTF CTF p H 2 dt

(6) (7) (8)

where C are concentrations, t is the reaction time, k are the kinetic constants and pH2 is the partial hydrogen pressure. Rate equations were integrated from the reactor inlet to the reactor outlet. It is assumed that the kinetic constants follow the Arrhenius form as:

k i = k io e




Therefore, the total set of model parameters comprises: pre-exponential factors: kPAo , kPAo,r , kDAo , kDAo,r , kMAo , kMAo,r , kNPo , kDTo , kBTo and kTPo. 2

activation energies: EaPA , EaPA,r , EaDA , EaDA,r , EaMA , EaMA,r , EaNP , EaDT , EaBT and EaTP.

The kinetic parameters were estimated with a previously validated parameter estimation procedure (4), based on the Maximum Likelihood technique (5), using the Gauss approach (6) to minimize the objective function. The ordinary differential equations were solved numerically with a 4th order Runge-Kutta method (7).


Figure 2 shows that the proposed model structure leads to excellent agreement between observed and predicted concentrations of mono-, di-, poly- and total aromatics. Therefore, the model structure may be used with confidence for optimization of the HDA process at plant site. It is interesting to observe that at certain typical operation conditions used for removal of sulfur and nitrogen of diesel fuels the total aromatics concentration in the product may be much higher that the total aromatics concentration of the feed. This effect becomes more pronounced as the reactor temperature becomes higher, while keeping hydrogen partial pressure and spatial time constant, as shown in Figure 3. This effect is caused by the reversible reaction steps presented in Figure 1. The reversible reaction steps impose maximum aromatics conversion limits that can be attained at plant site. These limits depend on the oil characteristics and on the reaction conditions. This also indicates that the HDS, HDN and HDA processes should be optimized simultaneously at plant site.

40 Model Predictions (%Volume) .




0 0 10 20 30 Experimental Values (%Volume) 40

Figure 2 Comparison between model predictions and experimental results for concentrations of mono- (), di- (o), poly- (+) and total aromatics ( ).

20% Conversion (% Volume) . 10% 0% -10% -20% -30% -40% -50% -30 -20 -10 0 10 20 30 40

Temperature - Tref (C)

Figure 3 Total aromatics conversion as a function of operating temperature.

A pseudo-phenomenological model for the HDA process was developed and validated, based on real process data. The model shows that reaction reversibility may be an important issue in industrial sites, as maximum aromatics conversion limits depend on the feed characteristics and operation conditions. This encourages the implementation of the model at plant site, in order to allow the optimization of HDA industrial plants as a function of feed properties. This may allow the improvement of diesel fuels, leading to lower emissions and higher combustion efficiency.

1. 2. 3. 4. 5. 6. 7. Dauzacker, E. R., Estudo do Processo de Saturao dos Compostos Aromticos em Destilados Mdios, M. Sc. Thesis, PEQ/COPPE, UFRJ, Rio de Janeiro (2001). Stanislaus, A. and Cooper, B.H., Aromatic Hydrogenation Catalysis: A Review, Catal. Rev. Sci. Eng. 36 (1), 75-123 (1994). Cooper, B. H., Donnis, B. B. L. , Aromatic Saturation of Distillates: An Overwiew,Appl. Catal. A. 137 (1996) 203. Noronha, F. B., Pinto, J. C., Monteiro, J. L., Lobo, M. W. and Santos, T. J., ESTIMA - The User's Guide, Internal Report, PEQ/COPPE/UFRJ, Rio de Janeiro (1993). Anderson, T.F., Abrams, D.S. e Grens II, E.A., 1978, AICHE J. 24,20-29. Bard, Y., Nonlinear Parameter Estimation, Academic Press, New York, 1974. ME Davis, Numerical Methods and Modeling for Chemical Engineers, John Wiley & Sons, New York, 1984.