Design and Control of Integrated Styrene-Aniline Production Plant

DESIGN AND CONTROL OF INTEGRATED STYRENE - ANILINE PRODUCTION PLANT Octavian Parteniea, Vincent van der Lastb, Costin Sorin Bildeaa, Pietro Altimaric University “Politehnica” Bucharest, Polizu 1-7, RO-011061, Bucharest, Romania Delft University of Technology, Julianalaan 136, 2628BL, Delft, The Netherlands c Università “Federico II” Piazzale Tecchio 80, 80125 Napoli, Italy a b Abstract: The paper illustrates the operational difficulties arising from simultaneously performing exothermic and endothermic reactions, and demonstrates that the plant can be built and safely operated by integrating the design and plantwide control issues. The behaviour of reactor – separation – recycle systems carrying the coupled reactions A →P + Q (endo) and B + Q → R (exo) is investigated. Irrespective of the control structure, state multiplicity cannot be removed if the intermediate component Q is recycled. Therefore, the chemical reactor should be designed such the recycle of Q can be avoided without economic penalty. The theoretical findings are applied to the design and control of a plant coupling ethylbenzene dehydrogenation and nitrobenzene hydrogenation for simultaneous production of styrene and aniline. After plant design, a rigorous dynamic model is developed using AspenDynamics. A plantwide control structure is implemented and shown to be able to achieve large production rate changes and to effectively reject various disturbances. . Keywords: Reaction coupling; Design and control; Nonlinear dynamics; Styrene; Aniline 1. INTRODUCTION Direct coupling of endothermic and exothermic reactions leads to improved thermal efficiency and, for reversible reactions, to increased equilibrium conversion and reaction rate due to equilibrium displacement (Towler and Lynn, 1994). As a result, energy savings and reduced reactor size can be achieved. However, while enhancing reactor performance, coupling endothermic and exothermic reactions in a single unit may require additional separation units and recycles. In practice, energy savings and reduced reactor investments must outweigh the cost of required additional units. Furthermore, it is important to remark that recycle of unconverted reactants can lead to undesired nonlinear phenomena in plantwide systems, with important implications for design and control (Bildea and Dimian, 2003). The research on reaction systems coupling endothermic and exothermic reactions has been mainly focused on the efficient design and analysis of stand-alone reactor units, with the simultaneous ethylbenzene dehydrogenation and nitrobenzene hydrogenation as a frequent example demonstrating the effectiveness of the idea (Qin et al, 2003, AboGhander et al, 2008). To assess the feasibility of coupling endothermic and exothermic reactions, operational and control difficulties arising from the more complex behaviour should be taken into account. Recent studies (Altimari and Bildea, 2008) addressed the problem of integrated design and control of plantwide systems coupling endothermic and exothermic reactions. Hypothetical processes were considered. For several flowsheets and different plantwide control strategies, multiple steady states were invariably detected. Singularity theory was exploited to divide the space of reactor-design parameters into regions characterized by qualitatively-different solution diagrams, implications of the observed behaviour on plant controllability being thoroughly discussed. In the first part of this contribution, we investigate the behaviour of reactor – separation – recycle systems carrying the coupled reactions A →P + Q (endo) and B + Q → R (exo) with the goal of identifying the conditions which guarantee a unique, stable steady state. To achieve this, we attempt to avoid the feedback arising due to material recycles by fixing the reactor-inlet flow rates (Bildea and Dimian, 2003). We show that, irrespective of the control structure, state multiplicity cannot be removed if the intermediate component Q is recycled. Therefore, we conclude that the operational difficulties can be solved not by control, but by considering the chemical reactor. This should be designed such that the reactor-outlet flow rate of the intermediate component Q is minimized, and its recycle can be avoided without a significant economic penalty. The second part of the paper applies the theoretical findings to the design and control of a plant coupling ethylbenzene dehydrogenation and nitrobenzene hydrogenation for simultaneous production of styrene and aniline. Compared to the classical processes, the main advantages are reduced sensitivity of the chemical reactor which can be operated adiabatically or at low heat-transfer capacities, and the reduced energy consumption due to low steam requirements. After plant design, a rigorous dynamic model is developed using AspenDynamics. A plantwide control structure is implemented and shown to be able to achieve large production rate changes and to effectively reject various disturbances. In conclusion, this paper illustrates the operational difficulties arising from simultaneously performing exothermic and endothermic reactions, and demonstrates that the integrated plant can be built and safely operated by integrating the design and plantwide control issues. 2. NONLINEAR BEHAVIOUR OF REACTOR – SEPARATION -RECYCLE SYSTEM 2.1 Mathematical model The system investigated in this section considers the following system, where the product Q of the first reaction is consumed in the second one. In turn, the exothermic second reaction provides the energy necessary for the first reaction to take place. A k P Q H 0 1 B Q k2 R H 0 Fig. 1 shows the general structure of the Reactor – Separation – Recycle system. Ideally, the reactants A, B and Q are recycled, while the products P and R are removed from the plant. A, B A, B, Q 0 3 1 P, R 2 4 Reactor Separation A, B, Q, P, R Fig. 1. General structure of the Reactor – Separation – Recycle system The steady state behaviour of the Reactor – Separation – Recycle system contains the reactor equations (1) - (6), which assume a plug-flow model and Arrhenius dependence of the reaction constants versus temperature. The reactor-inlet temperature (0) is fixed by feedback control of a heat exchanger placed upstream of the reactor at the value (0) = 1. Equation (7), assuming that the reactants A, B and the intermediate product C are recycled, while the products P and R are withdrawn from the plant, describes the separation, recycle and the mixing point. df A fA (1) Da exp 1 d f df B d d d K 21Da exp Da f B exp G21 1 1 fB f fA f fQ (2) f K 2,1H12 B exp G21 1 fB f fQ f c (3) fQ fQ ,1 f fA fA 0 f A1 f A,1 fA fB fQ fP f A1; fB 0 f A,1 f A ; fP f B ,1 (4) fB (5) fR f B1; f A 1 ; f B1 f A0 fB ; fR f B ,1 0 (6) 1 f B 1 ; fQ1 f B0 fQ 1 ; fQ 0 (7) The variables appearing in the mathematical model are the dimensionless axial coordinate , the dimensionless flow rates f, the temperature along the reactor, and the dimensionless recycle flow rates fK(1) = fK3. The model parameters are the Damkohler number Da, the ratio between the kinetic constants K21, the dimensionless adiabatic temperature rise B, the ratio between the reactions enthalpies H21, the dimensionless heat transfer capacity , the dimensionless cooling temperature c. The model contains three algebraic equations (7) containing six unknowns: fA0, fB0, fQ0, fA1, fB1, fQ1. In addition, the differential equations require the reactor-inlet temperature to be specified as initial condition. Therefore, a total of four specifications are needed. 2.2 Classification of the steady state behaviour In practice, a set of specifications needed to solve the model (1) - (7) corresponds to a feasible plantwide control structure. In the following, we will consider that reactants A and Q are recycled together. A feasible set of specifications is given by equations (8), where t and u are given values. fQ1 t ; f B1 f A1 u ; fQ0 = 0; 1 =0 (8) Fig. 3 presents the classification of the steady state behaviour of the reactor – separation – recycle system, with the parameter t as bifurcation parameter, and the conversion XA as variable representative for the state of the system. The space of the reactor design-parameters c – is divided into several regions, each region being characterized by a specific t – XA bifurcation diagram. Fig. 2 demonstrates the complex behaviour, including state multiplicity, isolated solution branches, and regions of feasibility. 1 1 I isola cusp 7 XA / [-] 5 6 7 8 9 10 11 0.2 0 0 1 2 3 4 VII 1 IV 0.013 θc 0.014 0.015 XA / [-] 0.012 9 10 11 0 1 2 3 4 5 6 7 8 9 10 11 t / [-] V 0.6 0.4 0.4 0.2 0 0 1 2 3 4 5 6 7 8 9 10 0.8 0.6 0.2 11 VI 1 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 t / [-] 1 5 6 7 8 9 10 11 t / [-] 1 VII 0.05 0.1 1 VIII 0.8 0.6 XA / [-] 0 XA / [-] 0.8 θc 8 VI III 0 7 0.4 0.2 0.6 0.4 0.2 0 1 2 3 4 5 t / [-] 6 7 8 9 10 11 0.6 0.4 0.2 0 0 IX 0.8 XA / [-] V 2 6 1 0.8 V t / [-] II 5 t / [-] IV III 5.4 0.011 0 -0.05 4 0.4 VIII 4 -0.1 3 0.6 I 5.8 6 2 t / [-] 6.2 I 1 XA / [-] 8 0.4 III 0.8 0 0 6.6 β IX 0.6 0.2 0 IX β 0.4 0.2 10 XA / [-] isola cusp XA / [-] 12 1 II 0.8 0.6 XA / [-] 0.8 0 0 1 2 3 4 5 t / [-] 6 7 8 9 10 11 0 1 2 3 4 5 6 7 8 9 10 11 t / [-] Fig. 2. Classification of the steady state behaviour. 2.3 Discussion: design and plantwide control implications From an operational point of view, the complex behaviour depicted in Fig. 2 is undesired. For this reason, we attempt to simplify it by decoupling the reaction and separation sections. This can be achieved by fixing all reactorinlet reactor flow rates (Bildea & Dimian, 2003). This can be done only freeing one specification, te corresponding variable being used as a manipulated variable in a control loop. In the following, we investigate two alternatives: 1, or c, corresponding to specifications (9) and (10), respectively. (9) f A1 t ; f B1 u ; fQ0 0 ; fQ1 v ; f A1 t ; f B1 u ; fQ 0 0 ; fQ1 v; (10) Fig. 3 shows representative t – XA bifurcation diagrams. Although the behaviour seems simpler that the one displayed in Fig. 2, state multiplicity and regions of unfeasibility still exists. We conclude that the desired pattern of behaviour cannot be achieved by control, but more profound changes are needed. Indeed, if the flowsheet is such that no intermediate reactant Q is recycled, then the reactor-inlet flow rates of A and B can be easily controlled. When the reactor-inlet temperature q1 is also fixed, the feedback effect of the separation section on the chemical reactor is effectively broken. From an economic point of view, this option is attractive when a limited amount of reactant Q leaves the plant, which requires a proper design of the chemical reactor. 1 1 (b) (a) 0.8 0.8 f Q,1 = 0.2 0.6 1 XA XA 0.6 f Q,1 = 0.2 1 0.4 0.4 5 5 10 0.2 10 0.2 0 0 0 1 2 3 4 0 5 1 2 3 4 5 t t Fig. 3. Fixing all reactor-inlet flow rates by using 1 (a) and bifurcation diagrams showing multiple steady states. c (b) as additional manipulated variables: t – XA 3. INTEGRATED STYRENE – ANILINE PLANT 3.1 Design Styrene (ST) is currently produced by dehydrogenation of ethyl-benzene (EB). The reaction is reversible, endothermic, and severely limited by the thermodynamic equilibrium. Reaction can proceed to high conversions only when a significant amount of steam is used, which acts as a heat-carrier and simultaneously lowers the EB partial pressure. After the reaction and before separation of the reaction mixture, steam must be condensed before being recycled. Therefore, the energy consumption of the process is high. Aniline (AN), the second product considered, is obtained by hydrogenation of nitrobenzene (NB). The reaction very exothermic and can easily lead to reaction runaway. In this section, we investigate the design and control of an integrated process for simultaneous production of styrene and aniline, according to the following reactions: C6 H 5 C2 H 5 C6 H 5 C6 H 5 C2 H 3 NO2 3H 2 C6 H 5 H2 NH 2 2 H 2O In absolute values, the enthalpy of EB dehydrogenation is 4 times lower than the enthalpy of NB hydrogenation. On the other hand, one mole of hydrogen is produced in the first reaction, while three moles are consumed in the second reaction. According to the previous section, the behaviour of the plant is simpler when the intermediate reactant (hydrogen) is not recycled, but removed from the plant. To avoid the economic penalty, there should be a small amount of hydrogen at the outlet of the reactor. This can be achieved by a suitable choice of reactor-inlet mixture and of ratio between amounts of catalysts. In our example, the reactor is fed with 38 kmol/h EB and 11.6 kmol/h of NB, at 640 ºC and 1 bar. The reactor has 150 tubes of 0.1 m diameter and 5 m lengths. Conservative values of 50 kcal/h/m2/K for the heat-transfer coefficient (meaning almost adiabatic operation) and a coolant temperature of 620 ºC were assumed. The kinetics of the reactions were taken from Lee (2005) and Amon et al (1999). The densities of catalysts were assumed to be equal. Fig. 4 shows temperature and mole fraction profiles along the catalytic bed. Although the heat-transfer capacity is low, the processes take place almost isothermally. In addition, the concentration of hydrogen is low. 0.8 650 T / [ºC] 645 0.6 EB y 640 ST 0.4 635 NB 0.2 630 AN H2 0 625 0 1 2 3 4 0 5 1 2 3 4 5 z / [m] z / [m] Fig. 4. Temperature and mole fraction profiles along the chemical reactor. Fig. 5 presents the entire flowsheet. After reaction, the mixture is cooled and water and gaseous products are separated (S-101). Column T-101 (30 stages) removes the light by-products, such as benzene and toluene. Column T-102 (20 stages) achieves the split between ST and EB (in the distillate) and AN and NB (in the bottoms). The products are obtained in columns T-103 (55 stages) and T-104 (stages), respectively. The whole process was simulated in AspenPlus. After all units were sized, a dynamic model was obtained using AspenDynamics. Lights Lights P-5 Gases EB Recycle PC PC LC FC M-101 EB P-7 PC TC T-103 FC LC S-40 R-101 FC LC S-39 P-2 P-2 LC TC FC TC S-45 TC PC P-5 S-101 TC PC NB T-101 P-3 S-46 CC T-102 TC P-5 CC TC CC LC S-41 Styrene P-1 I-70 LC I-69 S-42 P-6 P-4 Water PC M-102 LC Aniline S-38 LC T-104 FC S-37 LC NB Recycle Fig. 5. Flowsheet of the integrated styrene-aniline plant 3.2 Plantwide control and dynamics Fig. 5 presents the plantwide control system. The reactor-inlet temperature and flow rates of EB and NB are fixed, fresh reactants being fed on inventory control. Control of heat exchangers, vapour-liquid-liquid separator and distillation columns is standard. It should be remarked that dual-composition control of the column T-102 is required, because the AN which escapes in the distillate will end in the ST product, while the EB from the bottoms will be found in the AN product. Fig. 6 presents results of dynamic simulation. Starting from the steady state, different variables are changed at time t=5 hr, in order to investigate the effect on the plant throughput. In simulation (A), the EB reactor-inlet flow is increased by 10%. The effect on production rates is 4% (ST) and 4.8% (AN). The dynamics of the ST is quite fast, while more than 55 hours are needed until the AN rate stabilizes. In simulation (B) the reactor-inlet flow rate of NB is increased by 10%. The plant does not reach a new steady state, and sustained oscillations seem to emerge. In simulation (C), both reactor-inlet flowrates of EB and NB are increased by 10%, while in simulation (D) the reactor-inlet temperature is increased by 20 ºC. In both cases, new steady states are reached, the dynamics of the AN rate being significantly slower. 10 30 F Aniline / [kmol/h] F Styrene / [kmol/h] 31 (C) 29 (D) (A) 28 (B) 27 26 25 9.5 (A) (C) (D) 9 (B) 8.5 8 0 5 10 15 20 25 30 35 40 0 t / [h] 10 20 30 40 50 60 t / [h] Fig. 6. Dynamic simulation results 4. CONCLUSIONS In reactor – separation – recycle systems carrying the coupled reactions A →P + Q (endo) and B + Q → R (exo), irrespective of the control structure, state multiplicity cannot be removed if the intermediate component H is recycled. The operational difficulties can be solved not by control, but by considering the chemical reactor, which must minimize the reactor-outlet flow rate of the intermediate component Q and avoid its recycle without a significant economic penalty. The design and control of a plant coupling ethylbenzene dehydrogenation and nitrobenzene hydrogenation for simultaneous production of styrene and aniline is presented. Compared to the classical processes, the main advantages are reduced sensitivity of the chemical reactor which can be operated adiabatically or at low heattransfer capacities, and the reduced energy consumption due to low steam requirements. A plantwide control structure is implemented and shown to be able to achieve large production rate changes REFERENCES Abo-Ghander, N, J. Grace, S. Elnashaie, C. J. Lim (2008). Modeling of a novel membrane reactor to integrate dehydrogenation of ethylbenzene to styrene with hydrogenation of nitrobenzene to aniline. Chem. Eng. Sci. 63, 1817 Towler G, S. Lynn (1994). Novel applications of reaction coupling: use of carbon dioxide to shift the equilibrium of dehydrogenation reactions. Chem. Eng. Sci., 49, 2585. Bildea C.S., A.C. Dimian (2003). Fixing flow rates in recycle systems: Luyben’s rule revisited. Ind. Eng. Chem. Res., 42, 4578. Altimari, P., C.S. Bildea (2008). Coupling exothermic and endothermic reactions in plug-flow reactor – separation – recycle systems. Ind. Eng. Chem. Res., 47, 6685. Qin Z, J. Liu, A. Sun, J. Wang (2003). Reaction coupling in the new processes for producing styrene from ethylbenzene. Ind. Eng. Chem. Res., 42, 1329. Lee, W.J. (2005). Ethylbenzene dehydrogenation into styrene: kinetic modeling and reactor simulation. Texas A&M University, College Station, Texas, USA. Amon, B., Redlingshofer, H., Klemm, E., Dieterich, E., Emig, G. (1999). Kinetic investigation of deactivation by coking of a noble metal catalyst in the catalytic hydrogenation of nitrobenzene using a catalytic wall reactor. Chemical Engineering and Processing 38 (4), 395–404.