Transient modelling and catalyst deactivation in reaction engineering

Transcription

1 Transient modelling and catalyst deactivation in reaction engineering Tapio Salmi, Dmitry Murzin, Johan Wärnå, Esa Toukoniitty, Fredrik Sandelin Åbo Akademi utline Modelling of transient experiments Why transient experiment Transient techniques Modelling of transient experiments Results (examples) Catalyst deactivation About catalyst deactivation Modelling of catalyst deactivation Case studies 1

2 Why transient experiments btain more information on mechanism Multiple reactions (sequence of steps, forward and backward rates, etc) Adsorption, reaction, desorption Information on dynamic behavior (essential if reactor works in transient regime) Information on start-up and shut down behavior Transient behaviour in chemical engineering Start-up and shut down of processes Continuous change in conditions, e.g. car exhaust converters 2

3 Transient behaviour: Matros reactor Transient techniques Transient techniques Step response experiments Pulse experiments Temperature programmed desorption Temperature programmed surface reaction Isotope exchange experiments Transient infrared studies Instruments MS / rapid GC MS / GC MS MS MS FT-IR Characterized features Inlet parameters are changed in a step Inlet parameters are changed in pulses Desorption is recorded as function of time T is changed after each steady-state A molecule is replaced by its isotope Analysis of composition in bulk phase and on catalyst surface 3

4 Modelling transient experiments A transient plug flow model can be volumetric flow used dc dt concentration vector mole fraction ( cv & ) + σρ r d = ε Bε dv 1 1 void fraction surface area and in dimensionless form d x 1 d x dδ σρ BτRT = ε δ + x + dθ dz dz εp dimentionless time bulk density dimentionless change in volumetric flow r dimentionless coordinate dθ dt j = αr j Modelling transient experiments For the surface intermediates * dc * = r dt and expressed in surface coverage dθ τ = r d c Θ * 4

5 Case studies of transient experiments Examples from e.g. Rahkama-Tolonen doctoral thesis Transient kinetics: Isotope exchange in NH 3 NH3 -> NH3 + D2 Experiment: Ar -> 1%NH 3 -> 1%NH 3 + 1%D 2 -> 1%NH 3, Pd/modified alumina catalyst at 155 C NH3 2.5E-11 Intensities, amu 2.E E-11 1.E-11 NH2 ND3, fragmentation H2 NH2D 5.E-12 NHD2 HD H.E+ Formation of H 2 time, s 5

6 Catalyst deactivation, general All industrial catalysts experience catalyst deactivation Rate can vary significantly, from seconds to many years Time-Scale of Deactivation Hydrocracking HDS FCC Catalytic reforming E C 3 dehydrogenation MA Formaldehyde Hydrogenations Aldehydes Acetylene xychlorination Time / seconds Fat hardening NH 3 oxidation TWC SCR Most bulk processes.1-1 year Batch processes hrs-days hour 1 day 1 year 6

7 Types of deactivation bservation of catalyst deactivation 1 A -> > B 8 6 c A c B c t Concentrations as a function of time in a batch reactor. How do you distinguish kinetics from deactivation? Multiple of experiments needed. 7

8 Batch reactor bservation of catalyst deactivation 1 c, ut c B Catalyst deactivation A -> > B t Concentrations as a function of time on stream in a fixed bed reactor. Deactivation observed in activity decay as function of time. 8

9 The elements of a reactor model Catalytic reactor model Kinetics Stoichiometry Equilibria Rate equations for adsorption, desorption and surface reactions Heat effects on rates Catalyst deactivation Mass and heat transfer Mass and heat transfer on reactor level Interfacial mass and heat transfer Intraparticle mass and heat transfer Flow pattern Continuous or batch Degree of backmixing Fixed bed, slurry, moving bed or fluidized bed Computational fluid dynamics Mass, energy and impulse balances Influence of Deactivation on Reaction Rate conversion or k obs initial level k obs process time = kintr NT η constant variable variable loss of surface area blocking of pores loss of active sites Fouling Sintering Poisoning 9

10 Two approaches to catalyst deactivation Semi-empirical separate function Typical approach in the past Simple to use and obtain, easy computation Semi-empirical separate function Reaction rate obtained by adjusting initial rate with time dependent function = a t r Semi-empirical functions a = 1 k t a = e k d t 1 a = 1 k d d t ( ) i r i linear, zero order exponential, first order hyperbolic, second order 1

11 What to do? Two approaches to catalyst deactivation Mechanistic approach (as any reaction) Treats deactivation as any reaction in system Involves many (complex) reactions in a network Dynamic models evolves, demanding to compute More information of the system can be obtained. 11

12 Mechanistic approach Identify reactions e.g. monomolecular reaction and bimolecular coking A* B* 2A* C* and develop rate expressions r r Isom Deact = k iso = k c dea A K A Θ V ( c K Θ ) 2 A A V Mechanistic approach Rearrange equations r r Isom deact k c + K ( ΘC ) ( c + c ) iso A 1 * = 1 = A B 2 2 k deaca( 1 ΘC* ) ( 1+ K( c + c )) 2 A B 12

13 Mechanistic approach Develop mass balances for reactor and for catalyst surface species ci 1 w ci = + ri ρ B t ε L z d dt Θ C * = αr deact The mass balance for the surface components Mass balance for adsorbed surface components * dc j ΔΑ = rjδmcat dt and dθ dt j = αr j where, α = Δm * cat c and Θ j = ΔΑ c * j c * 13

14 Dynamic reactor models Axial dispersion model 2 ci 1 w ci Da ci = + + r ρ 2 2 i B t ε L z L z Dynamic plug flow model dci 1 dci = ri ρ B dt ε dτ Batch reactor dci dt 1 = ri ρb ε Numerical methods Dynamic fixed bed Steady state fixed bed Batch reactor PDE + DE DE + (N)LE DE PDE DE (N)LE Partial differential equation rdinary differential equation (Non-)linear equation 14

15 Principle of numerical methods The PDEs are solved by discretization With finite differences rtogonal collocation The DEs are solved with routines suitable for stiff systems Backward differences, BD, e.g. LSDE Runge-Kutta, RK, e.g. SIRK Solving PDEs by discretization PDE ci 1 w ci = + ri ρ B t ε L z c t c t DEs 1 w ci ε L ( z ) + r ρ i = 1 i B 1 w ci ε L ( z ) + r ρ i = 2 i B 15

16 Case study I: Skeletal isomerization of 1-pentene A B A + * A* B* B + * 3 4 C * C * Semi-empirical model Mechanism reduces to A B Kinetics r = k c Isom A a 16

17 Semi-empirical model Deactivation functions k d t a = e 1 a = 1 + kdt Mass balance dci risomρ B dτ = Results, semi-empirical model 17

18 Mechanistic model Mechanism A* B* II 2A* C* IV r r Isom = k +2 c AK AΘV ( ) 2 Deact = k + 4 c AK AΘV Mechanistic model Kinetics k ca risom = K r deact ( 1 ΘC* ) ( c + c ) A B 2 2 k 4cA( 1 ΘC* ) ( 1+ K( c + c )) 2 = + A B 18

19 Mechanistic model Mass balance ci 1 w ci = + ri ρ B t ε L z d dt Θ C * = αr deact Results, mechanistic model 19

20 Results, mechanistic model Investigate deactivation mechanisms Isomerization A* B* II Deactivation mechanisms 2A* C* IVa A* C* IVb B* C* IVc A* C* IVd 2

21 Concentration (wt-%) Concentration (wt-%) 5, 45, 4, 35, 3, 25, 2, 15, 1, 5, 5, 45, 4, 35, 3, 25, 2, 15, 1, 5,, Investigate mechansim utlet, Model II a, pp=.5, Time (h) utlet, Model II c, pp= Time (h) 2A* 2C* = n-c5 olef. 2B*? 2C* = n-c5 olef. Concentration (wt-%) Concentration (wt-%) 5, 45, 4, 35, 3, 25, 2, 15, 1, 5,, 5, 45, 4, 35, 3, 25, 2, 15, 1, 5,, utlet, Model I a, pp= utlet, Model Time V a, pp=.5 (h) A*? C* = n-c5 olef Time (h) A*? C* = n-c5 olef. 6. Estimate coke yield Discretization point 1 (utlet), Model II a, pp=.5 5. coke (wt-%) = model estimate = coking at 18 ºC = coking at 3 ºC = kinetic experiment Time (h) 21

22 Conclusions Deactivation can be accounted for in many ways. If understanding of the deactivation phenomenon is desired a more rigorous model is needed. Time-on-stream is not allways a good variable for catalyst deactivation A dynamic mechanistical model is solvable with modern computational tools. The coke on catalyst was modelled and compared to experimental data H (I) H H (B) H (F) H [ R] [ S] [ R] + [ S] ee (%) = 1 % H H (E) (A) (D) Main product is 1 R H H H H H (H) (C) (G) 22

23 .25 c (mol/dm 3 ) dione 1-hydroxyketones 2-hydroxyketones diols (Batch reactor) time (min) Multi-centered adsorption model applied Co-adsorption of organic molecules and catalyst modifier Adsorption of hydrogen essentially non-competitive Passive spectators on the catalyst surface Kinetic model Impossible to obtain explicit rate-expressions, expressions, but fractional coverages solved numerically c A (mk 1 (1+ k 7 /k 8 )c (m-1) Θ m + nk 2 (1+ k 9 /k 8 )c (n-1) Θ n ) + c M (pk 3 c (p-1) Θ p + qk 4 c (q-1) Θ q ) + (m+p+f)k 1 K 3 K 5 (1+k 11 /k 12 )c A c M c (m+p+f-1) + (n+p+l)k 2 K 3 K 6 (1+ k 13 /k 14 )c A c M c (n+p+l-1) + Θ = 1 23

24 Continuous operation Catalyst deactivation: Dione H 2 uptake (mol dm -3 ) Deactivation model 88 s 44 s 3 s 22 s U H ( t) 2 = P + P e P 2 t.5 Hidden in batch operation time-on-stream (min) E. Toukoniitty, P. Mäki-Arvela, A. Kalantar Neyestanaki, T. Salmi, D. Yu. Murzin, Applied Catalysis A: General, 235 (22)

25 Transient modelling: Dione Kinetic model : steady or non-steady adsorption? adsorption equilibrium dynamics of adsorption Reactor model dynamic axial dispersion model for tube reactor Peclet number from impulse experiments with inert tracer Impulse experiments with inert tracer H 2 1 Pe = Injection 2 μl (H 2 + NaCl) H 2 E(t ) τ.6.4 Experiment t/τ response 25

26 Transient modelling: Dione Direct implementation of steady state kinetics c (mol/dm -3 ) B.8.6 C time Steady state kinetics: Describes well batch reactor data Adsorption quasiequilibria Addition of hydrogen RDS Transient modelling: Dione Non-steady state kinetics.1 c (mol/dm -3 ).. B C Reversible adsorption 5 μm time 26

27 Case study III: Transformation of epoxides in the Hydrogen peroxide process H R + 2 R + H 2 H THAAHQ (B) + 2 Epoxide (A) + H 2 (D) R + H R 2 R + H 2 H Epoxide (A) + THAAHQ (B) 2 THAAQ (C) + H 2 (D) Case study: Transformation of epoxides in the Hydrogen peroxide process 3 R R + 2 H R H 3 THAAQ (C) AAQ (E) + 2 THAAHQ (B) 27

28 Mechanism I A + B 2C + D IIa C + C C' + B IIb C' + C E + B III B + E F + C IV A + F E + C + D Kinetics 1 r k c c K cc 2 1 = + 1 A B C D 1 r 2 = k c 3 21 C k c + c 22 B C k = Ae E 1 / RT k = A e E 2 / RT

29 Modelling results Epoxide ºC THAAHQ Tim e / ( h) Modeling of a continuous pilot reactor (packed bed) 29

30 Modeling of a continuous pilot reactor (packed bed) Dispersion model c t 2 1 w c D c = + + riρ 2 2 B ε L z L z i i a i Reaction kinetics and deactivation function () t r j rj = a Stoichiometry ri = ν ijr j Semi-empirical description of catalyst deactivation r = k c c k c P + P v P P + * P* da dt ( ) = k a a n k t ( a a ) e, n 1 1 [( ) n ( ) ] a( t) = a + = 1 n at () = a + a a + k n 1 t, n 1 1 3

31 Reversible and irreversible deactivation 1 8 Choose the right model Choose the right model! for deactivation! 6 c 4 2 reversible irreversiblel t Modeling of a continuous pilot reactor (packed bed) 1.5 Catalyst B 1.4 poxide concentration (mass-%)