PARAMETRIC SENSITIVITY OF A GASOIL HYDROTREATMENT REACTOR Ricardo SARDELLA 1 , Jean-Marc SCHWEITZER 2 * , Hua WU 3 , Clementina LOPEZ GARCIA 2 , Massimo MORBIDELLI 3 1 PDVSA-Intevep. 76343, Caracas 1070 A, Venezuela 2 Institut Français du Pétrole, IFP Lyon, P.O. Box 3, 69360, Solaize, France. 3 Institute for Chemical and Bioengineering. Swiss Federal Institute of Technology Zurich. ETH. 8093 Zurich, Switzerland Summary Thermal stability analysis is imperative to ensure the safe operation of chemical reactors carrying out highly exothermic reactions. This work presents a thermal stability study based on stationary and dynamic criteria. A fixed bed gasoil hydrotreatment reactor working at industrial operating conditions was investigated. The stationary study is based on the van Heerden criterion and parametric sensitivity. The influence of the variation of the most important operating parameters was studied. Stability maps were traced and the parametrically sensitive regions were identified. Finally, a complementary dynamic stability criterion was applied to evaluate the dynamic stability close to the bifurcation points. This comprehensive stability study enables to determine the safe/unsafe operating regions of this reactor. Keywords Parametric sensitivity, hydrotreatment, thermal stability. Introduction * j-marc.schweitzer@ifp.fr Hydrotreatment is a catalytic process that aims to purify oil fractions at high hydrogen partial pressure to obtain improved quality products by eliminating heteroatoms and increasing their H/C ratio. Hydrotreatment accounts of more than one thousand units all around the world covering a capacity of about 44 millions of barrels per calendar day 1 . It is one of the most important processes in the oil refining industry. Gasoils are petroleum cuts issued from different operations and processes in the refining chain and must be hydrotreated to meet commercial diesel specifications. Gasoil hydrotreatment (HDT) reactions mainly consist on hydrogenation of aromatics, olefins and hydrodesulfurization reactions; all are exothermic. Hydrotreatment of gasoils that have a high aromatics content such as light cycle oils (in average 80 wt%) is generally carried out by dilution with other kinds of gasoils having a much lower aromatics content. This is a choice not only based on the poor light cycle oils (LCO) quality, but also based on safety concerns since hydrotreatment is extremely exothermic. Hence, this study aims at determining the safe operating regions of a LCO HDT reactor via a thermal stability analysis. Different approaches to realize a thermal stability study are reported in literature. Van Heerden 2 studied exothermic reactions under stationary conditions. This author showed that a diagram representing the heat consumption and production enables to determine if the reactor temperature is maintained (stable) or a multiplicity of steady states occurrence (unstable). A stationary stability criterion is deduced establishing that the slope of heat generated by the reactions ( 29 dT dQ gen / must be lower than the slope of the heat transferred ( 29 dT dQ trans / . Another approach for thermal stability analysis under steady-state conditions is the parametric sensitivity. Parametric sensitivity analysis consists to assess how the reactor temperature responds to changes of parameters controlling the reactive system 3-4 . The sensitivity S is calculated as j j d dT T S φ φ = ) ; ( , where the variable studied is temperature (T) and the model parameters are j φ . Finally, a complementary dynamic stability criterion 5 can also be applied. Consisting on the perturbation of the reactor model, the stability condition states that after solution of the perturbed model, the real part of all the eigenvalues must be negative. This work presents a detailed parametric sensitivity analysis as well as the complementary approaches (van Heerden and dynamic analysis). Results A dynamic reactor model was developed to carry out the thermal stability study. The detailed description and validation of this model has been reported elsewhere 6 . The most relevant points are indicated here. The model