REV. CHIM. (Bucureºti) ♦ 60♦ Nr. 9 ♦ 2009 http/www.revistadechimie.ro 949 Derivation of Operating Region Runaway Boundaries for the Vapour Phase Catalytic Reactor used for Aniline Production DRAGOS-NICOLAE STEFAN, GHEORGHE MARIA* University Politehnica of Bucharest,Department of Chemical Engineering, 1-3 Polizu Str., 011061, Bucharest, Romania The safety in operation of a fixed-bed catalytic reactor remains a sensitive issue when a highly exothermic reaction is conducted and various process development elements such as controllability, stability, risk, and economic aspects have to be considered. Nominal operating conditions are set at a certain distance from safety limits in order to control the hot spot in the tubular reactor and to avoid excessive thermal sensitivity to random variations in process parameters. In the present study, a robust and precise sensitivity criterion, i.e. the model-based z-MV generalized criterion of Morbidelli & Varma, is used to establish the runaway boundaries in the operating variable space. The method sets the global runaway conditions as corresponding to the maximum sensitivity of the temperature peak in the reactor vs. operating parameters. A concrete example is provided for the case of the fixed-bed catalytic reactor for aniline production. Because the process operating variables, such as the inlet temperature or fed ratio, inlet pressure, cooling agent temperature, etc., are subjected to random fluctuations within a certain range, confidence intervals of the derived process runaway boundaries are also predicted, to be further considered in the optimal location of the reactor set-point. Keywords: runaway boundaries, confidence region, catalytic reactor, aniline production * email: gmaria99m@hotmail.com In an industrial chemical plant, the reactor is the core equipment on which the optimization efforts are focused due to the high value of raw materials and products related to the production cost, but also due to its high sensitivity to operating conditions, risk, and stability problems. Optimization procedures are usually employed to set the reactor’s nominal operating conditions within a stable and economic region in the parametric space, while elaborated control schemes are usually implemented to keep the reaction synthesis within the safety limits [1]. For the tubular reactor cases, safe operation tries to limit the hot spot and avoid excessive sensitivity to variations in the process parameters. Various optimization criteria have been proposed taking state variable sensitivity as constraints [2,3], keeping a certain distance vs. the runaway boundaries [4], or explicitly accounting for the uncertainty in the operating parameters [5-7]. However, frequent perturbations in the operating parameters, raw-material recycle conditions, catalyst replacement or reactivity modifications, all require periodical updates of the safety margins for the operating variables. Unsafe conditions correspond to sensitive operating regions when “the reactor performance becomes unreliable and changes sharply with small variations in parameters” [8]. Risky operating conditions for highly exothermic primary or secondary reactions are determined by using simple shortcut techniques, or by using more elaborated model-based methods to appreciate the high thermal sensitivity of the reactor to operating conditions [11]. For a catalytic tubular reactor, the runaway sensitivity analysis can be developed for a single particle at a certain reactor location (usually at the reactor inlet, i.e. the so-called local runaway conditions), or extended over the whole reactor length looking for the hot-spot sensitivity to concomitant variations of parameters (the so-called global runaway conditions). Approximate risk assessment of the reactor derives the safe conditions based on certain inequalities constructed for every reactor type by using engineering numbers (such as Damköhler-Da, Stanton-St, or Lewis- Le), or safety indices that replace the systematic model- based safety analysis of the process [9,11,12,28]. Such a risk assessment tries to evaluate interactions between primary and secondary reactions expressed by means of risk measures, such as: adiabatic induction time to explosion (τ ad ); reaction violence index B [ΔT ad E/(R g T o 2 ), where E = activation energy of the reaction; T o = initial temperature of the reaction; R g = universal gas constant; DT ad =(-ΔT ad )c j,o / (ρ g c pg ) = adiabatic temperature rise; (- ΔH)= heat of reaction; c j,o = initial concentration of key species; ρ g = reacting mixture density; c pg = average specific heat of gas]; rate of generating the reaction heat compared to the rate of heat removal by the cooling system; adiabatic temperature limits of reactions ΔT ad , ADT 24 , reaction T onset levels from dual scanning calorimeter DSC measurements, etc. Values of reaction heat of (- ΔH) > 10 kcal/mol, TMR ad <8h, Δ T ad >50K, and B>5 indicate potential dangerous reactions, presenting a fast evolution and a significant exothermicity (where TMR ad = time to maximum rate under adiabatic conditions; ADT 24 = sample initial temperature for an adiabatic decomposition within 24 h). More precise methods for predicting the operation safety limits are based on the process/reactor model, of complexity depending on the available information on kinetics, thermodynamics, and reaction pathway. For the tubular reactor case, including the catalytic fixed-bed operated with a hot spot (HSO) or pseudo-adiabatic (PAO), the thermal sensitivity conditions are identified mainly by using three types of methods[8]: explicit, geometrical, and sensitivity methods. Explicit criteria derive the approximate critical operation conditions based on simple and explicit