Prediction of reaction kinetic of Al-Doura Heavy Naphtha Reforming Process Using Genetic Algorithm Al-Khwarizmi Engineering Journal Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61(2014) Prediction of Reaction Kinetic of Al- Doura Heavy Naphtha Reforming Process Using Genetic Algorithm Ramzy H. Saihod* Zaidoon M. Shakoor** Abbas A. Jawad*** * Department of Petroleum Technology / University of Technology ** Department of Chemical Engineering / University of Technology ***Al-Doura Refinery / Middle Refinery Company * E-mail: ramze_eng@yahoo.com ** E-mail: dr_zaidoon@yahoo.com ** E-mail: ahmed_198494@yahoo.com (Received 31 October 2012; accepted 2 March 2014) Abstract In this study, genetic algorithm was used to predict the reaction kinetics of Iraqi heavy naphtha catalytic reforming process located in Al-Doura refinery in Baghdad. One-dimensional steady state model was derived to describe commercial catalytic reforming unit consisting of four catalytic reforming reactors in series process. The experimental information (Reformate composition and output temperature) for each four reactors collected at different operating conditions was used to predict the parameters of the proposed kinetic model. The kinetic model involving 24 components, 1 to 11 carbon atoms for paraffins and 6 to 11 carbon atom for naphthenes and aromatics with 71 reactions. The pre-exponential Arrhenius constants and activation energies were determined after fine tuning of the model results with experimental data. The input to the optimization is the compositions for 21 components and the temperature for the effluent stream for each one of the four reactors within the reforming process while the output of optimization is 142 predicted kinetic parameters for 71 reactions within reforming process. The differential optimization technique using genetic algorithm to predict the parameters of the kinetic model. To validate the kinetic model, the simulation results of the model based on proposed kinetic model was compared with the experimental results. The comparison between the predicted and commercially results shows a good agreement, while the percentage of absolute error for aromatics compositions are (7.5, 2, 8.3, and 6.1%) and the temperature absolute percentage error are (0.49, 0.5, 0.01, and 0.3%) for four reactors respectively. Keywords: Heavy Naphtha, Reforming, Genetic Algorithm, Optimization, Reaction Kinetic. 1. Introduction The catalytic reforming process is one of the most critical operations in petroleum refineries. This process involves the reconstruction of low- octane hydrocarbons in the naphtha into more valuable high-octane gasoline components without changing the boiling point range, production of aromatic feedstock for petrochemical industries also hydrogen and lighter hydrocarbons are obtained as side products [1, 2] . The catalytic reforming of naphtha involves reactions such as dehydrogenation, dehydrocyclization, hydrocracking, isomerization, and dehydroalkylation. The naphtha feed to reformer is very complex usually consisting of about three hundreds components with carbon number range from C5 – C12[3]. Recent environment legislation in the United States has banned the use of lead as an additive for boosting antiknock properties of motor fuel. Coupled with these stricter environmental regulations, there has been a consistent increase in the demand for higher octane number gasoline. This can be achieved by reforming the naphtha under more severe conditions, but this will also cause an increase mailto:ramze_eng@yahoo.com mailto:dr_zaidoon@yahoo.com mailto:ahmed_198494@yahoo.com Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 48 in the rate of coke deposition, resulting in the reduction of cycle lengths of catalyst [4]. Concerning the kinetic modeling of the naphtha processes smith 1959 [5] firstly proposed four lumps model by considering naphtha as three group regents, paraffins, naphthenes, and aromatic hydrocarbons. Due to its simplicity, this model has been used in some recent reformer modeling work. Krane et al., 1960[6] recognized the presence of various carbon numbers from C6 - C10 as well as the difference between paraffins, naphthenes, and aromatics within each carbon number group. The model derived contained a reaction network of 20 different components. Kmak 1972 [7] used Langmuir kinetic model for reactors as first time. Marin and coworkers 1982 [8] presented the reaction network for the whole naphtha, containing hydrocarbons in the carbon number fraction from C5-C10. The reaction network included 23 pseudo components and used Hougen-Watson type rate equations. Ramage et al., 1987 [9] decided to develop a comprehensive kinetic model which (involving a reasonable number of group components and pathway) would capture the reactivity differences between particular raw materials. Their studies led to the construction of Mobil kinetic model of the reforming process (KINPtR start of cycle and deactivation kinetics). Taskar 1997 [10] developed a detailed kinetic scheme involving 35 pseudo components connected by a network of 36 reactions in the C5 – C10 range. Deactivation of the catalyst was modeled by including the corresponding equations for coking kinetics. Jorge 2000 [11] proposed a kinetic model for the naphtha catalytic reforming process (mathematical representation of the reaction that take place) and carbon number ranging from 1- 11 atoms for paraffins , 6-11 for naphthenes, and aromatics. The kinetic parameters values were estimated using experimental information obtained in three fixed-bed pilot plant. Weifeng 2003 [12] developed mathematical model for simulation and optimization of commercial naphtha catalytic reformers with four reactors in series. The model along with deactivation function of catalyst can monitor the reformer performance with time on stream. Arani 2010 [13] simulated a dynamic model of the catalytic naphtha reformer process by MATLAB software. The kinetic parameters of model are based on the steady state condition and obtained from a commercial plant data furnished by a domestic petroleum refinery. Askari 2012 [1] developed model for simulating catalytic reforming unit with four reactors in series by using Hysys-refinery software. The results are validated by operating data, taken from the Esfahan oil refinery catalytic reforming unit. In the industrial applications of reaction kinetics, the estimation of parameters in kinetic expressions from data series is essential for the design, optimization, and control of many chemical systems. The use of process models as a tool for simulation of commercial process has been growing rapidly. The advantage of utilizing rigorous mathematical models as compared to empirical approaches is related to the fact that the prediction accuracy of rigours models can be significantly superior over a wide operating range. To design new catalytic reforming process or to optimize the existing ones, an appropriate kinetic model to represent the reactions within the process is needed. Therefore the aim of this study is to determine the values of kinetic model parameters for Iraqi heavy naphtha reforming process by using differential optimization technique (genetic algorithm). 2. Process Description The process flow diagram of the reforming process is shown in Figure 1. The commercial semi-regenerative catalytic naphtha reforming contains four fixed-bed adiabatically operated reactors in series. The naphtha used as feedstock which contains a mixture of paraffins, naphthenes, and aromatics in the carbon number range C4-C11 was combined with a recycle gas stream containing 60 – 90 mol% hydrogen. The used catalyst is Pt-Re/Al2O3 which is bifunctional bimetallic catalyst providing the metal function and the acid function. Usual catalytic reforming consists of multiple reactors (three or four) with heaters between the reactors to maintain reaction temperature at operable levels, since the major reactions in the first reactor dehydrogenation of naphthenes, which are endothermic and very fast, causing a very sharp temperature drop in the first reactor. As the total reactor charge passes through the sequence of heating and reacting, the reactions become less and less endothermic and temperature differential across the reactors decreases. The product from the fourth reactor is cooled and then enters to the product separator. The flashed vapor circulates to combine the naphtha Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 49 feedstock as recycle gas. Extra hydrogen from the separator is sent to other hydrogen consuming units in the refinery. The separated liquid that mainly included the desired aromatics together with heavy paraffins and light gases is sent to the reformate stabilizer. Reformate off the bottom of the stabilizer is sent to storage for gasoline blending. Table (1) shows the design parameters and operating conditions of the catalytic reformers of Al-Doura refinery in Baghdad. The operating conditions of this unit were: 470 °C inlet temperature, 27.5 bar reactor pressure, and the feedstock flow rate of 33.5 m 3 /hr. Fig. 1. Process flow of four reactors naphtha reforming process. Table 1, Operating conditions of heavy naphtha reforming process. Reactor number 1 2 3 4 Catalyst weight kg 2700 4500 4750 5750 Inlet Temperature C 470 468 468 468 Reactor Length m 6 6 6 6 Reactor Diameter m 2.4 2.4 2.4 2.4 Reactor 2 Reactor 1 Reactor 3 Reactor 4 Header 1 Header 2 Header 3 Header 4 Recycle Gas Compressor Separator Feed Reformate Cooler Heat Exchanger Off Gas Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 50 ak tAj ii p p TTR E kk                          11 3. Mathematical Model The following assumptions were considered in the present mathematical model: 1. The system was in a steady state. 2. The variation in the radial direction was negligible. 3. All reactions were in homogenous phase. 4. All reactions are pseudo first order with respect to hydrocarbon. The equations of mathematical model results from application of material and energy balance principles in a differential volume [14] .   i m i i r WHSVz WtM dZ dC    1 . . …(1)          m i Pi m i Rii i CF Hrs dZ dT 1 1 . …(2) The Ergun equation was used for computing total differential pressure drop in axial flow reactor [15] : G de me G de e dZ dP PP t 23 2 32 3 5 )1( 10*5.1 1 10*75.1       …(3) In order to evaluate the heat capacity the following correlation was used [16] ; 3 i 2 iiii TDTCTBACp  …(4) The effect of temperature and pressure on the kinetic constants was expressed in equation (5), Jenkins et al. 1980 [17] ; …(5) The values of pressure effect factors on reaction rate are given in Table (2). Table 2, Values of pressure effect reaction rate [17] . Reactions ak isomerization 0.37 dehydrocyclization -0.7 hydrocracking 0.433 hydrodealkylation 0.5 dehydrogenation 0.0 4. Kinetic Model In this study, 24 lump kinetic model which proposed by Ancheyta et al 2000 [11] was used. According to this model, the naphtha feed to reforming process contain paraffin’s, naphthenes, and aromatics with carbon number from 1 to 11 carbon atoms for paraffin’s (P1-P11) and from 6 to 11 carbon atoms for naphthenes (N6-N11) and aromatics (A6-A11). The extended kinetic model employs a lumped mathematical representation of the seventy-one chemical reactions for all 24 lumps that taken place can be shown in Table (3). All reaction steps are combined into twenty-four rate reaction equations (ri), one for each component. Each reaction rate equation is a function of the kinetic constant (ki) and the component partial pressure (Pi). . Table 3, Reactions of the kinetic model (Ancheyta et al. (2001)). Number of Reactions Paraffin’s Naphthenes Aromatics Pn → Nn Pn → Pn-j + Pj subtotal 6 26 32 Nn → An Nn → Nn-j + Pj Nn → Pn subtotal 6 11 7 24 An → An-j + Pj An → Pn An → Nn subtotal 7 5 1 13 Total 71 n: Number of atoms of carbon (1 ≤ i ≤ 5) Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 51 5. Optimization Optimization can be for minimization or maximization of the desired objective function with respect to decision variables subject to process constraints and bounds on the variables. An optimization problem can have a single optimum or multiple optima, one of which is the global optimum and the others are local optima. A global minimum has the lowest value of the objective function throughout the region of interest. Most of the traditional optimization algorithms based on gradient methods have the possibility of getting trapped at local optimum depending upon the degree of non-linearity and initial guess. Unfortunately, none of the traditional algorithms are guaranteed to find the global optimum solution. In the recent past, nontraditional search and optimization techniques (Evolutionary Computation) based in natural phenomenon such as Genetic Algorithms (GAs) has been developed to overcome these problems. Arx et al., 1998 [19] used genetic algorithm technique for finding a global minimum for the error function. Tongcheng et al., 2005 [20] combine Numeric Genetic Algorithm (NGA) and Tabu Search (TS) to optimize the estimated rate constants of a consecutive reaction. Zhao et al., 2006 [21] used a nonlinear least squares regression to fit the kinetic profiles. Genetic Algorithms (GAs) are powerful and widely applicable stochastic search and optimization methods based on the concepts of natural selection and natural evaluation. Genetic Algorithm works on a population of individuals represents candidate solutions to the optimization problem. These individual consists of a strings (called chromosomes) of genes. The genes are a practical allele (gene could be a bit, an integer number, a real value or an alphabet character,…,etc depending on the nature of the problem). GAs applying the principles of survival of the fittest , selection , reproduction , crossover (recombining) , and mutation on these individuals to get , hopefully , a new butter individuals (new solutions). Genetic Algorithm generates new population of chromosomes by selecting the better fit solutions from existing population and applying genetic operators to produce new offspring of the solutions. The process is repeated successively to generate new population iteratively. This process is repeated until some criterion is met or a reasonably acceptable solution is found. The Outline of the Genetic Algorithm is given below [18] . 1. [Start] Generate random population of n chromosomes (suitable solutions for the problem). 2. [Fitness] Evaluate the fitness f(x) of each chromosome x in the population. 3. [New population] Create a new population by repeating following steps until the new population is complete. I. [Selection] Select two parent chromosomes from a population according to their fitness (the better fitness, the bigger chance to be selected). II. [Crossover] with a crossover probability cross over the parents to form a new offspring (children). If no crossover was performed, offspring is an exact copy of parents. III. [Mutation] with a mutation probability mutate new offspring at each locus (position in chromosome). IV. [Accepting] Place new offspring in a new population 4. [Replace] Use new generated population for a further run of algorithm 5. [Test] if the end condition is satisfied, stops, and returns the best solution in current population 6. [Loop] Go to step 2 Proportional selection, ranking, and tournament selection are the most popular selection procedures. Table 4, Contains the parameters of genetic algorithm. Population size 10 Maximum generation 3000 Crossover probability 0.8 Mutation probability 0.05 Neighborhood size 0.05 6. Numerical Computation For each individual reactor within the process, numerical integration method was used to integrate the component mass balance, energy balance and pressure drop differential equations (1, 2 and 3) to obtain concentration, temperature and pressure profiles along the reactor as follows. The rate equations (system of simultaneous differential equations) of the reaction were solved to get the analytical concentration versus reactor length profiles. Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 52 Fourth order Runge-Kutta integration command named ode45 was used to integrate 24 stiff ordinary differential equations for mass balance and two other equations for heat balance and pressure drop. Genetic Algorithm was used to predict the parameters of kinetic model by minimizing the objective function J in equation 6, which is the sum of squares of errors between the predicted and measured values for all of the state variables.               nr i nc j i i pred i ji pred ji T TT yy n J 1 1 2 exp exp 2exp ,, exp )()( 1 …(6) Where: exp , ji y is the composition of component j for the effluent stream of reactor No i. Using the differential optimization algorithm to optimize the system through a sequence of optimization-evaluation, the objective function [Eq. (6)] was minimized and the global optimum set of kinetic parameters was found out. It is important to mention here that [Eq. (6)] was used in evaluation the fitness (objective function) for each chromosome x in the population. Values of the Frequency factors (A1, A2 to A71), Activation Energies (E1, E2 to E71) were found by minimization of the sum of the squares of the deviations between the plant and the calculated results of the key variables (the compositions and temperature of effluent from each one of the four reactors). The fitness for each generation of chromosomes was calculated and the minimum fitness represents the best chromosomes within the chromosomes and the average fitness was also calculated. Figure (2) shows a plot of the best and mean fitness (J) with respect to generation number. Genetic algorithm has one disadvantage which is a huge computation time in the case of complex systems. In the present case study a PC with 4.12 GHz and 4GB RAM. 5 runs take more than 120 hr to reach produced results. Finally the better chromosomes was selected which represent less objective function for all chromosomes and all generations. The kinetic parameters of obtained reaction rate with the genetic optimization procedure are presented in the Table 5. 0 500 1000 1500 2000 2500 3000 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Generation F it n e ss , J Minimum fitness Mean fitness Fig. 2. Minimum and mean fitness with respect to generation number. 7. Results and Discussion 7.1. Validation of Reaction Kinetics To validate the predicted kinetic model, the model results using the predictive kinetic model were compared with actual results collected from AL-Doura refinery (catalytic reforming process). Figure 3 (a, and b) shows the comparison between the actual and simulated reformate composition (run 1 means the data collected in 1/12/2012, while run 2 in 1/1/2012). It can be observed that the calculated reformate compositions of all (paraffins , naphthenes, and aromatics) for the four reactors in catalytic reforming unit agree very well with experimental information with average deviation less than 2% as shown in table (6 and 7). . . Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 53 Table 5, Kinetic constants of the kinetic model. Reaction Step Ko EA (cal/mol) Reaction Step Ko EA (cal/mol) Reaction Step Ko EA (cal/mol) P11 N11 0.073082 53553.8 P8 2P4 0.001478 65074.0 N8 N7+P1 0.000016 27622.9 P10 N10 0.037751 37965.0 P7 P6+P1 0.000704 58826.6 N11 A11 1.146035 24527.0 P9 N9 0.055255 9439.8 P7 P5+P2 0.003188 39451.4 N10 A10 0.903828 23982.5 P8 N8 0.033905 53951.7 P7 P4+P3 0.000376 65101.8 N9 A9 0.403788 6489.7 P7 N7 0.004895 39023.9 P6 P5+P1 0.003237 19289.4 N8 A8 0.395175 29318.6 P6 N6 0.000004 41629.3 P6 P4+P2 0.000070 64572.5 N7 A7 0.286847 31262.2 P11 MCP 0.009867 36845.6 P6 2P5 0.007834 11940.1 N6 A6 0.090934 33819.9 P11 P10+P1 0.055585 33531.3 P5 P4+P1 0.000668 60910.7 A11 P11 0.014602 13735.3 P11 P9+P2 0.070800 29879.0 P8 P3+P2 0.020922 14733.7 A10 P10 0.014531 11859.7 P11 P8+P3 0.090772 22115.8 N11 P11 0.033542 16723.6 A9 P9 0.015793 11816.5 P11 P7+P4 0.013195 58984.6 N10 P10 0.047576 47431.2 A8 P8 0.010064 9059.9 P11 P6+P5 0.057271 63624.4 N9 P9 0.051031 35413.1 A7 P7 0.001665 27438.8 P10 P9+P1 0.003473 27543.6 N8 P8 0.024635 12039.8 A11 A10+P1 0.002544 21381.4 P10 P8+P2 0.000517 39959.3 N7 P7 0.009329 11474.4 A11 A9+P2 0.005768 37776.0 P10 P7+P3 0.001024 61173.5 N6 P6 0.195584 17660.9 A10 A9+P1 0.003530 43959.9 P10 P6+P4 0.000822 64218.0 MCP P6 0.001872 35936.4 A10 A8+P2 0.000541 40180.9 P10 2P5 0.000389 62410.0 N11 N10+P1 0.082099 58751.5 A10 A7+P3 0.000005 40148.4 P9 P8+P1 0.008461 16828.6 N11 N9+P2 0.114900 24140.5 A9 A8+P1 0.001916 39779.4 P9 P7+P2 0.000367 61631.1 N11 N8+P3 0.066425 18506.9 A9 A7+P2 0.001371 42474.0 P9 P6+P3 0.001029 61813.5 N10 N9+P1 0.131747 15698.4 A8 A7+P1 0.000059 32571.2 P9 P5+P4 0.000012 48938.2 N10 N8+P2 0.087315 46817.6 A6 N6 0.013302 24076.0 P8 P7+P1 0.000228 60915.0 N10 N7+P3 0.001255 38799.8 MCP N6 0.234651 36084.8 P8 P6+P2 0.015421 64298.5 N9 N8+P1 0.056925 64391.1 N6 MCP 0.014943 24377.5 P8 P5+P3 0.000375 29372.1 N9 N7+P2 0.002137 61898.0 Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 54 Table 6, Comparison between actual and simulated reformate compositions (run 1). Composition Feed Reactor 1 Reactor 2 Reactor 3 Reactor 4 Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. n-P4 0.0036 0.0023 0.0000 0.0023 0.0014 0.0000 0.0014 0.0026 0.0000 0.0026 0.0014 0.0000 0.0014 n-P5 0.0106 0.0128 0.0130 0.0002 0.0135 0.0192 0.0057 0.0200 0.0242 0.0042 0.0177 0.0278 0.0101 n-P6 0.0802 0.0797 0.0890 0.0093 0.0879 0.0936 0.0057 0.0969 0.0962 0.0007 0.0908 0.0955 0.0047 n-P7 0.1243 0.1201 0.1242 0.0041 0.1139 0.1162 0.0023 0.1040 0.1062 0.0022 0.0894 0.0945 0.0051 n-P8 0.1023 0.1194 0.1156 0.0038 0.1003 0.0980 0.0023 0.0756 0.0748 0.0008 0.0548 0.0597 0.0049 n-P9 0.1782 0.1130 0.1191 0.0061 0.0913 0.0682 0.0231 0.0666 0.0510 0.0156 0.0452 0.0432 0.0020 n-P10 0.1214 0.0928 0.1107 0.0179 0.0754 0.0756 0.0002 0.0569 0.0513 0.0056 0.0424 0.0364 0.0060 n-P11 0.0020 0.0040 0.0009 0.0031 0.0043 0.0001 0.0042 0.0064 0.0000 0.0064 0.0063 0.0000 0.0063 MCP 0.0033 0.0021 0.0047 0.0026 0.0005 0.0049 0.0044 0.0005 0.0050 0.0045 0.0036 0.0052 0.0016 N6 0.0214 0.0000 0.0055 0.0055 0.0000 0.0034 0.0034 0.0000 0.0041 0.0041 0.0000 0.0047 0.0047 N7 0.0554 0.0046 0.0160 0.0114 0.0035 0.0021 0.0014 0.0036 0.0019 0.0017 0.0032 0.0017 0.0015 N8 0.0699 0.0078 0.0141 0.0063 0.0076 0.0070 0.0006 0.0072 0.0065 0.0007 0.0055 0.0056 0.0001 N9 0.0406 0.0183 0.0169 0.0014 0.0152 0.0090 0.0062 0.0098 0.0063 0.0035 0.0055 0.0052 0.0003 N10 0.0542 0.0000 0.0022 0.0022 0.0000 0.0022 0.0022 0.0000 0.0016 0.0016 0.0000 0.0012 0.0012 N11 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 A6 0.0035 0.0080 0.0078 0.0002 0.0106 0.0114 0.0008 0.0135 0.0155 0.0020 0.0161 0.0215 0.0054 A7 0.0255 0.0632 0.0664 0.0032 0.0812 0.0907 0.0095 0.1001 0.1053 0.0052 0.1215 0.1238 0.0023 A8 0.0762 0.1379 0.1338 0.0041 0.1707 0.1705 0.0002 0.2059 0.2027 0.0032 0.2478 0.2305 0.0173 A9 0.0182 0.1427 0.1030 0.0396 0.1485 0.1514 0.0029 0.1536 0.1611 0.0075 0.1659 0.1572 0.0087 A10 0.0091 0.0713 0.0570 0.0143 0.0742 0.0763 0.0021 0.0768 0.0860 0.0092 0.0829 0.0861 0.0032 A11 0.0000 0.0000 0.0002 0.0002 0.0000 0.0003 0.0003 0.0000 0.0003 0.0003 0.0000 0.0002 0.0002 Para- ffins 0.623 0.544 0.574 0.030 0.488 0.476 0.012 0.429 0.413 0.016 0.348 0.371 0.023 Naph- thenes 0.245 0.033 0.059 0.026 0.027 0.028 0.002 0.021 0.025 0.004 0.018 0.023 0.005 Arom- atics 0.132 0.423 0.366 0.057 0.485 0.496 0.011 0.550 0.562 0.012 0.634 0.606 0.028 Temp. (K) 743.150 698.150 697.454 0.696 732.150 729.251 2.899 738.150 738.414 0.264 745.150 742.418 2.732 Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 55 Table 7, Comparison between actual and simulated reformate compositions (run 2). Composition Feed Reactor 1 Reactor 2 Reactor 3 Reactor 4 Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. Exp. Pred. Abs. diff. n-P4 0.0025 0.0021 0.0000 0.0021 0.0025 0.0000 0.0025 0.0033 0.0000 0.0033 0.0036 0.0000 0.0036 n-P5 0.0059 0.0097 0.0080 0.0017 0.0150 0.0138 0.0012 0.0223 0.0189 0.0034 0.0257 0.0229 0.0028 n-P6 0.0378 0.0462 0.0492 0.0030 0.0590 0.0618 0.0028 0.0753 0.0718 0.0035 0.0782 0.0778 0.0004 n-P7 0.1225 0.1110 0.1219 0.0109 0.1016 0.1134 0.0118 0.1046 0.1030 0.0016 0.0983 0.0911 0.0072 n-P8 0.1182 0.1375 0.1327 0.0048 0.1058 0.1121 0.0063 0.0877 0.0843 0.0034 0.0733 0.0661 0.0072 n-P9 0.2035 0.1196 0.1357 0.0161 0.0851 0.0763 0.0088 0.0683 0.0562 0.0121 0.0500 0.0471 0.0029 n-P10 0.1093 0.1004 0.1012 0.0008 0.0766 0.0704 0.0062 0.0596 0.0483 0.0113 0.0484 0.0343 0.0141 n-P11 0.0016 0.0030 0.0007 0.0023 0.0034 0.0001 0.0033 0.0049 0.0000 0.0049 0.0053 0.0000 0.0053 MCP 0.0009 0.0009 0.0023 0.0014 0.0015 0.0031 0.0016 0.0023 0.0036 0.0013 0.0026 0.0042 0.0016 N6 0.0199 0.0000 0.0046 0.0046 0.0000 0.0021 0.0021 0.0000 0.0029 0.0029 0.0000 0.0036 0.0036 N7 0.0507 0.0042 0.0156 0.0114 0.0035 0.0021 0.0014 0.0037 0.0019 0.0018 0.0034 0.0017 0.0017 N8 0.0793 0.0085 0.0168 0.0083 0.0088 0.0078 0.0010 0.0083 0.0072 0.0011 0.0072 0.0061 0.0011 N9 0.0469 0.0363 0.0191 0.0172 0.0268 0.0100 0.0168 0.0201 0.0069 0.0132 0.0152 0.0056 0.0096 N10 0.0588 0.0000 0.0020 0.0020 0.0000 0.0020 0.0020 0.0000 0.0015 0.0015 0.0000 0.0012 0.0012 N11 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 A6 0.0017 0.0046 0.0054 0.0008 0.0075 0.0077 0.0002 0.0102 0.0103 0.0001 0.0119 0.0149 0.0030 A7 0.0244 0.0547 0.0607 0.0060 0.0789 0.0842 0.0053 0.0946 0.0985 0.0039 0.1064 0.1165 0.0101 A8 0.0858 0.1473 0.1487 0.0014 0.1911 0.1890 0.0021 0.2110 0.2238 0.0128 0.2323 0.2530 0.0207 A9 0.0200 0.1427 0.1158 0.0269 0.1553 0.1691 0.0138 0.1492 0.1782 0.0290 0.1588 0.1720 0.0132 A10 0.0100 0.0713 0.0594 0.0119 0.0776 0.0748 0.0028 0.0746 0.0824 0.0078 0.0794 0.0816 0.0022 A11 0.0000 0.0000 0.0001 0.0001 0.0000 0.0003 0.0003 0.0000 0.0002 0.0002 0.0000 0.0002 0.0002 Para- ffins 0.601 0.530 0.551 0.022 0.449 0.453 0.004 0.426 0.391 0.035 0.383 0.353 0.029 Naph- thenes 0.257 0.050 0.060 0.010 0.041 0.027 0.014 0.034 0.024 0.011 0.028 0.022 0.006 Aro- atics 0.142 0.421 0.389 0.032 0.510 0.521 0.010 0.540 0.585 0.045 0.589 0.625 0.036 Temp. (K) 743.15 698.15 694.69 3.456 732.15 728.49 3.660 738.15 738.22 0.078 745.150 742.442 2.708 Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 56 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 Experimental mole fraction K in e ti c M o d e l m o le f ra c ti o n Reactor 1 Reactor 2 Reactor 3 Reactor 4 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 Experimental mole fraction K in e ti c M o d e l m o le f ra c ti o n Reactor 1 Reactor 2 Reactor 3 Reactor 4 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 690 700 710 720 730 740 750 Catalyst weight (kg) T e m p e ra tu re ( K ) 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 690 700 710 720 730 740 750 Catalyst weight (kg) T e m p e ra tu re ( K ) a (run 1) b (run 2) Fig. 3. Comparison between actual and simulated reformate compositions (symbols actual ,lines predicted). 7.2. Reactor Temperature Simulation Results The temperature decreases along the catalyst bed in the reactors of reforming process, because the process reactions are, overall, endothermic. For this reason, commercial catalytic reformers are designed with multiple reactors and with heaters between the reactors to maintain reaction temperature at operable levels. The simulated temperature profile with the actual reactors temperature is plotted in Figure (4), it can be seen that in the first reactor the temperature decrease very sharp, since the major reactions are endothermic and very fast reaction, such as dehydrogenation of naphthenes components to aromatics and the temperature decreasing drop is less in other reactors especially in the last two reactors, which is due to the exothermic hydrocracking reaction. The comparison between the simulated temperature and the actual temperature of commercial reforming unit shown in table (6 and 7), which shows a good agreement results and the accumulated difference is 6.5 ° C between the predicted and actual reactors temperature. . Figure (5) shows both the actual and the simulated pressure drop with respect to accumulative catalyst weight within the four reactors in commercial reforming process, it can be seen that a good agreement between these results. a (run 1) b (run 2) Fig. 4. Comparison between actual and simulated temperature profile (symbols actual, lines predicted). Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 57 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 Catalyst weight (kg) M o le f ra c ti o n P 4 P 5 P 6 P 7 P 8 P 9 P 10 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 0.25 Catalyst weight (kg) M o le f ra c ti o n P 4 P 5 P 6 P 7 P 8 P 9 P 10 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 2.2 2.3 2.4 2.5 2.6 2.7 2.8 x 10 6 Catalyst weight (kg) P re ss u re ( P a ) 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 2.2 2.3 2.4 2.5 2.6 2.7 2.8 x 10 6 Catalyst weight (kg) P re ss u re ( P a ) 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Catalyst weight (kg) R e fo rm a te C o m p o si ti o n m o l% Paraffins Naphthenes Aromatics 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Catalyst weight (kg) R e fo rm a te C o m p o si ti o n m o l% Paraffins Naphthenes Aromatics a (run 1) b (run 2) Fig. 5.Comparison between Actual and Simulated Pressure Drop (symbols actual ,lines predicted) 7.3. Reformate Composition Simulation Results Reformate composition of total paraffins, naphthenes, and aromatics obtained from actual catalytic reforming unit with the proposed model are shown in Figure (6), the results show a good agreement between the actual and the simulated results. Tables (6 and 7) show comparison between the actual and the simulated reformate compassion of all paraffins components (4 – 11) carbon number, naphthenes and aromatics components from (6 – 11) carbon number through the four reactors in catalytic reforming process. It can be observed that total aromatics hydrocarbons yields are higher as goes from the first reactor to the last reactor, therefore the total amount of aromatics increase from 13.2 mol% to reach 42.3%, 48.5%, 55%, and 63.4% for feedstock of run 1. While for feedstock of run 2 increasing from 14.2 mol% to 42.1%, 51%, 54%, and 58.9%. a (run 1) b (run 2) Fig. 6. Comparison between actual and simulated total Paraffins, naphthenes, and aromatics in reforming process (symbols actual ,lines predicted). Figure (7) shows a comparison between the actual and the simulated results of the heavy and the light paraffins along the reactor as a function of catalyst weight. It can be seen that light paraffins increased because they are produced by hydrocracking reaction, while heavy Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 58 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 Catalyst weight (kg) M o le f ra c ti o n MCP N 6 N 7 N 8 N 9 N 10 N 11 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 Catalyst weight (kg) M o le f ra ct io n MCP N 6 N 7 N 8 N 9 N 10 N 11 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 Catalyst weight (kg) M o le f ra c ti o n P 4 P 5 P 6 P 7 P 8 P 9 P 10 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 0.25 Catalyst weight (kg) M o le f ra c ti o n P 4 P 5 P 6 P 7 P 8 P 9 P 10 paraffins decreasing because exhibited high levels of conversion this is because the increase in the probility of ring formation is high as the molecular weight of paraffins increases also in the last two reactors since dehydrocyclization and reaction take place. a (run 1) b (run 2) Fig. 7. Comparison between actual and simulated Paraffins composition (symbols actual, lines predicted). The naphthenes and aromatics reformate comparison results shown in Figures (8, and 9). Naphthenes components which are the most desirable in reforming feed stocks react and converted to aromatics components through dehydrogenation reaction which take place in the first two reactors and these reactions goes to completion. Also it is important to mention her that all aromatics comparison in reformate are increased as feedstock passes through the catalytic reforming reactors especially lighter aromatics (A6, A7, A8, and A9), while heavy aromatics (A10, and A11) increasing very low or nearly remains unchanged. . a (run 1) b (run 2) Fig. 8. Comparison between actual and simulated naphthenes composition (symbols actual, lines predicted). Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 59 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Catalyst weight (kg) M o le f ra ct io n A 6 A 7 A 8 A 9 A 10 A 11 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Catalyst weight (kg) M o le f ra ct io n A 6 A 7 A 8 A 9 A 10 A 11 a (run 1) b (run 2) Fig. 9. Comparison between actual and simulated aromatics composition (symbols actual ,lines predicted). 8. Conclusions The kinetic parameters of model are based on the steady state condition and are obtained from a commercial plant data furnished by a domestic petroleum refinery. In this work a mathematical model of semi-regenerative catalytic reforming unit with four reactors in series it has been developed. The model parameters were estimated on the basis of data obtained from industrial unit (Al-Dura refinery). The effects of different feed composition on product properties are evaluated, and simulation results were compared with the actual data, there for the absolute percentage error of aromatics compositions range between (2% to 8.3%) and the temperature absolute percentage error range between (0.01% to 0.5%) for four reactors respectively. Nomenclature A Aromatics ( - ) N Naphthene ( - ) P Paraffin ( - ) MCP Methylcyclopentane ( - ) n-P Normal Paraffin ( - ) ki Reaction rate constant hr -1 k ◦ i Pre-exponential factor ( - ) EA Activation energy kcal/mole R Gas constant J/mole.K T Reaction temperature °C T • Initial temperature °C Pt Total pressure bar p◦ Partial pressure bar αk Pressure effect ( - ) Ci CP Concentration of species i Specific heat mole/cm 3 J/mole.K Z, z Length of reactor Cm M.Wt Molecular weight g/gmole WHS V Weight hour space velocity hr -1 ri Reaction rate of species i mole/gcat. hr S Cross sectional area of reactor m 2 ∆H˚r Heat of i th reaction J/ mole Fn Molar flow rate of species n mole/hr Cp Specific heat J/mole.°C dp Equivalent diameter of a catalyst particle m e Void fraction of reactor bed m 3 /m 3 m Viscosity pa.s G total mass flux of fluid kg.s/m 2  density kg/m 3 Ramzy H. Saihod Al-Khwarizmi Engineering Journal, Vol. 10, No. 1, P.P. 47- 61 (2014) 60 9. References [1] Askari A., Karimi H., Rahimi M.R., Ghanbari M., "Simulation and Modeling of Catalytic Reforming Process", J. Petroleum and Coal, vol 54, No (1), p 76-84, 2012. [2] Seif Mohaddecy S.R., Zahedi S., Sadighi S., Bonyad H.," Reactor Modeling and Simulation of Catalytic Reforming Process", J. Petroleum and Coal, vol 48, No (3), p 28-35, 2006. [3] Yong H., Hongy S., Jian C.,” Modeling, Simulation and Optimization of Commercial Naphtha Catalytic Reforming Process ”, Proceeding of the 42nd IEEE Conf, USA, (Dec), p 6206, 2003. [4] Jin L., Tan Y., Liangcai L.,” Modeling and Optimization of Semi-regenerative Catalytic Reformer”, Proceeding of the 2005 IEEE Conf, Canada, (Aug), p 28-31, 2005. [5] Smith R.B.,” Kinetic Analysis of Naphtha Reforming with Platinum Catalyst ”, J. Chem. Eng. Prog, vol 55, No6, P 76-80, 1959. [6] Krane H.G., Croh A.B., Schulman B.L., and Sinfelt J.H.,” Reactions in Catalytic Reforming of Naphtha’s,” 5th World Petroleum Congress, vol.3, p 39, 1970. [7] Kmak W.S.,”A Kinetic Simulation Model of the Powerforming Process ”, AIChE, Meeting, Houston TX, 1972. [8] Marin G.B., Froment G.F.,” Reforming of C6 Hydrocarbons on Pt-AL2O3 Catalyst ”, J. Chem. Eng. Sci, vol 37, No 5, p 759-773, 1982. [9] Ramage M.P., Graziani K.R., Krambeck F.J., Choi B.C.,” MObils Reforming Model: A Review of Mobil’s Industrial Process Modeling Philosophy ”,J. Chem. Eng. Sci, vol 35, P 193-266, 1987. [10] Taskar U.M., Riggs J.B.,” Modeling and Optimization of a Semiregenerative Catalytic Naphtha Reformer ”, AIChE. J, vol 43, No 3, p 740-753, 1997. [11] Ancheyta J.J., Eduardo V.M., “ Kinetic Modeling of Naphtha Catalytic Reforming Reactions ”, J. Energy Fuels, vol 14, No 5, p 1032-1037, 2000. [12] Weifeng H., Young H., Hongye S., Jain C.,” Simulation, Sensitivity Analysis and Optimization of a Continuous Catalytic Reforming Process ”, Proceeding of the 5th World Congress on Intelligent Control and Automation, June (15-19), 2004. [13] Arani H.M., Shokri S., Shirvani M., " Dynamic Modeling and Simulation of Catalytic Naphtha Reforming", J. Chem Eng and Appl, vol 1, No (1), August 2010. [14] Villafertre E., Jorge A.J.,” Kinetic and Reactor Modeling of Naphtha Reforming Process ”, J. Petroleum and Coal, vol 44, 1-2, p 63-66, 2002 [15] Froment G.F., Bischoff K.B.,”Chemical Reactor Analysis and Design”, John Wiley and Sons, 1990. [16] Reid R.C., Prausnitz J.M., Sherwood T.K.,” The Properties of Gases and Liquid”, 3rd Edition, McGraw-Hill Co, 1977. [17] Jenkins J.H., Stephens T.W.,” Kinetics of Catalytic Reforming”, J. Hyd. Proc, Nov, p 163-167, 1980. [18] Namita K., Anju R., Akshatha.P.S,”Genetic Algorithm: A Search of Complex Spaces” International Journal of Computer Applications (0975 – 8887), Volume 25– No.7, July 2011. [19] Arx, K.B.V., J.J. Manock, W.H. Scott and M. Messina, 1998. “Using limited concentration data for the determination of rate constants with the genetic algorithm” Environ. Sci. Technol., 32: 3207-3212, (1998). [20] Tongcheng, C., Z. Zhongliang, H. Xiaofeng and L. Tonghua, 2005”Simultaneous determination of reaction order and rate constant of consecutive reactions by union optimization algorithm of NGA-TS” Can. J. Anal. Sci. Spectros., 50: 182-191, (2005). [21] Zhao, Y., G. Wang, W. Li and Z.L. Zhu,”Determination of reaction mechanism and rate constants of alkaline hydrolysis of phenyl benzoate in ethanol- water medium by nonlinear least squares regression” Chemometr. Intell. Laborat. Syst., 82: 193-199, (2006). (2014) 47- 61، صفحة 1العدد، 10دمجلة الخوارزمي الهندسية المجل رمزي صهيود حميد 61 تفاعالت تهذيب النفثا الثقيلة لوحدة حركيةاستخدام النظرية الجينية إليجاد مصفى الدورة ***عباس عالء جواد **زيدون محسن شكور *رمزي صيهود حميد الجامعة التكنولوجية / قسم تكنولوجيا النفط* الجامعة التكنولوجية / م الهندسة الكيمياويةقس ** طشركة مصافي الوس/ مصفى الدورة *** ramze_eng@yahoo.com : االلكتروني البريد* dr_zaidoon@yahoo.com : ونيااللكتر البريد** ahmed_198494@yahoo.com : االلكتروني البريد*** الخالصة وحدة التهذيب تتكون من . تفاعالت وحدة تهذيب النفثا الموجودة في مصفى الدورة ببغدادحركية في هذا البحث تم استخدام النظرية الجينية إليجاد الموديل الرياضي . تم تطوير موديل رياضي لمحاكاة عملية التهذيب ذاتية التنشيط للنفثا في الحالة المستقرة ولبعد واحد. مفاعالت على التوالي اربع .يصف تغير التراكيز والضغط ودرجة الحرارة على طول المفاعالت االربعة المستخدمة لعملية التهذيب االيزو )مادة وهي البرافينات 42التفاعالت التي تتضمن وصف حركيةتجميع نتائج عملية بظروف مختلفة وتم استخدامها لوصف ثوابت تم 11من حركيةذرة كاربون باالعتماد على 11-6ذرة كاربون والنفثينات والمواد االروماتية التي تحتوي من 11الى 1التي تحتوي من( والنورمال مادة وكذلك درجة الحرارة لتيار التدفق لكل مفاعل من المفاعالت االربعة المستخدمة ضمن عملية 41لداخل لالختيار االفضل هو التركيب ل ا .تفاعل تيار واستخدمت النظرية التفاضلية لالخ ,تفاعل ضمن عملية التهذيب 11ثابت للموديل الرياضي المحسوبة المتكونة من 124التهذيب بينما الخارج هو .االفضل لحساب ثوابت الموديل الرياضي بنفس ماخوذة عملية نتائج مع المقترحة التفاعل حركية على المعتمد الرياضي الموديل من المستحصلة الرياضية النتائج مقارنة تم تصف المقترحة التفاعل ةحركي ان على يدل العملية وهذا والنتائج الرياضي الموديل نتائج بين جيدجدا انطباق هناك كان حيث الظروف والنسبة (and 6.1% ,8.3 ,2 ,7.5) العطرية للمواد للخظأ المطلقة المئوية النسبة وكانت النفثا تهذيب وحدة في التفاعالت جيد بشكل .التوالي على االربعة للمفاعالت( (and 0.3% ,0.01 ,0.5 ,0.49 هي الحرارة لدرجة للخطأ المئوية mailto:ramze_eng@yahoo.com*البريد mailto:ramze_eng@yahoo.com*البريد mailto:dr_zaidoon@yahoo.com**البريد mailto:dr_zaidoon@yahoo.com**البريد mailto:ahmed_198494@yahoo.com***البريد mailto:ahmed_198494@yahoo.com***البريد (2014) 47- 61، صفحة 1العدد، 10دمجلة الخوارزمي الهندسية المجل رمزي صهيود حميد 64