Heterogeneous Reaction Engineering: Theory and Case Studies Module 4 Analysis of Local Transport Effects in Gas-Liquid-Solid Systems P.A. Ramachandran rama@wustl.edu 1 Outline • • • • Transport Effects Diagnostic plots for slurry systems Partial wetting and implications Slurries containing fine particles 2 Heterogeneous Liquid-Phase Reaction Phenomena ci catalyst Ni Ni ci T Liquid film Vapor film Bulk liquid Bulk vapor E E Liquid film T Challenges: 1. Identifying reaction(s) and their location(s) 2. Accounting for internal and external catalyst wetting / holdup phenomena 3 Mass Transfer Resistances in Gas-Liquid-Solid Systems 4 Local Rate of Reaction for GasLiquid-Solid Catalyzed Systems A (g) + b B (l) P (l) Gas-Liquid Mass Transfer: RA = kLaB (A* - AL ) Liquid-Solid Mass Transfer for A: RA = ksap (AL - As ) Liquid-Solid Mass Transfer for B: RB = ksap (BL - Bs ) Intra particle Diffusion with Reaction: RA = wc(As, Bs)kmnAsmBsn 5 Intraparticle Diffusion Limitations • Solution of the reaction-diffusion equations in the catalyst particle for some simple reactions results in effectiveness factorThiele modulus relationship similar to that represented by the enhancement factor-Hatta number relationship for gas-liquid reactions tanh k1 L De Other details: Froment & Bischoff (1979) 6 Observed Rate - 1st Order Reaction in a Gas-Liquid-Solid System • For linear kinetics and the slow reaction regime, an overall resistance can be defined that includes the gas-liquid and liquid-solid mass transfer and reaction terms, including intraparticle diffusion limitations 1 1 * 1 RA A kL aB kS aPw C wk1 -1 A* 1 rb rcr RA w rb ( s ) 1 k LaB rcr ( gcat s / m 3 ) 1 1 k s a p c k1 7 Diagnostic Plots: First Order Case A*/RA A*/RA 1/w 1/w Gas-Liquid Mass Transfer Controls the Process Negligible Gas-Liquid Resistance A*/RA Slope = rcr Intercept = rb 1/w Intermediate Case 8 Diagnostic Plots (contd) A*/RA A*/RA Increasing resistance to gas absorption Decreasing Particle Size 1/w 1/w m>1 A*/RA m=1 m<1 m=0 1/w Schematic Plots for other Higher Orders 9 Commonly-Used Kinetic Models for Gas-Liquid-Solid Systems A (g) + b B (liq) Mechanism P (liq) Rate Form 1. Single site adsorption of dissolved gas k11 A B 1 K A A 2. Dissociative adsorption of dissolved gas k11 A B 1 K A A 3. Adsorption of both A & B on single sites k11 AB 1 K A A K B B 4. General single site adsorption for N species k11 AB N 1 Ki Ci i 1 10 Overall Effectiveness Factor for a (m,n) order Reaction RA 0 m n wk (A*) BL = f(A , o) where: (1) 1 1 1 A* k a k a L s p A wk(A*)m BLn R (m 1)Sp k(A*) o 3 2De (2) m 1 n L 1 2 B (3) 11 Overall Effectiveness Factor for a Single-Site L-H Rate Form k11 AB Example: Glucose Hydrogenation A 1 KA A Ramachandran & Chaudhari, 1983 12 Analysis of External Mass Transport Resistance • Overall mass transfer of H2 (gas) to catalyst surface is : Rgas = MA(A*-As), A* - gas concentration in the liquid phase (using Henry’s Constant) MA ( 1 1 1 ) K LaB ks a p • Gas consumption by all reactions: RH 2 No . of rxn j 1 H2 , j rj • Considering As = 0 , we have M A A* No . of rxn r j 1 H2 , j j If LHS > > RHS, then no mass transfer resistance ! 13 Analysis of Internal Resistance (within the Catalyst Pellet) • Weisz-Prater criterion is used r L n1 We D C 2 2 2 obs i eff b n = reaction order, taken as unity robs = net rate of consumption of limiting Reactant (initial rate) Cb = concentration of limiting reactant in liquid Deff = effective diffusivity = DABp/τ p= particle porosity ~ 0.5 τ = totuosity = 2 DAB = binary diffusivity (Wilke Chang correlation L = (characteristic length) Vp/Sp Internal Resistance is considered negligible if We0.5 < 0.2 14 Parameter Estimation Method • • • • Step-by-step approach Start with a temperature data set Identify the reactions Identify the reaction form (reaction rate): rj 1 K C 1 Where, n w cat k j C gas Cj K 2C 2 K 3C 3 K 4C 4 K 5C 5 2 gas k j (or K j ) Aoj e E j RT Aoj and Ej are estimated ! 15 Kinetic Parameter Estimation (contd.) • Non-linear Optimization Problem N obj min(C i 1 C i , aut oc l av )2 , t ,runs e i ,pr edi c t ed s ubj: Aut oc lav eModel / Slurryreac t or • Identify and select the for the objective function • Identify the species adsorbed, if any (C1 to C5 here, as an example) • Develop parameter estimation program and the autoclave model / Slurry reactor model • Autoclave model predicts the species concentration at every instant (for the operating conditions) – set of differential equations can be solved by VODE routine from NETLIB libraries • Levenberg-Marquardt algorithm for parameter estimation – UNLSF routine from IMSL libraries 16 Trickle-Bed Reactors 17 Fixed-Bed Multiphase Reactors (a) Trickle - Bed Cocurrent downflow (b) Trickle - Bed Countercurrent flow (c) Packed - Bubble Flow Cocurrent upflow Semi-Batch or Continuous Operation; Inert or Catalytic Solid Packing 18 Trickle-Bed Reactors - Pros and Cons Pros • Plug-flow • Low liquid holdup less homogeneous reactions • • • high conversion Cons • Intraparticle diffusion resistance • Incomplete contacting/wetting • High pressure drop • Temperature control problems hot spots • Scale-up and design is complex • Attrition and crush resistant catalyst is required High specific reaction rate Temperature control possible by liquid vaporization High pressure operation possible • Minimal catalyst handling issues • Process flexibility, reasonable throughput limitations • Dirty process streams cannot be used plugged or fouled bed • Lower capital & operating costs • Catalyst loading is complicated 19 Fundamental Phenomena in Trickle Bed Reactors Macroscale • Axial & radial RTD’s • Flow regime • Pressure drop • Liquid holdup • Liquid flashing • Interphase transport • Liquid distribution • Heat transfer • Energy dissipation Microscale • LocaL texture of liquid flow (films, rivulets, stagnant pockets) • Local irrigation and wetting • Liquid holdup in pores • Local transport between gas and flowing and stagnant liquid, and solid • Local transport between flowing liquid, stagnant liquid, and solid • Local transport between gas and vapor-filled pores 20 Classification of TBR Processes Based on Volatility 1. Nonvolatile liquid reactant • Rate limiting reactant - Liquid - Gas - Both Reaction occurs only on wetted catalyst 2. Volatile liquid reactant • Rate limiting reactant - Liquid - Gas - Both Reaction occurs both on wet and dry catalyst 21 Key TBR Design Parameters • Flow regime • Pressure drop • Liquid holdup • Liquid - solid contacting • Interphase transport coefficients • Intraparticle diffusion • Extent of liquid volatilization • Reaction kinetics • Thermo - physical constants 22 Flow Regime Structures for Gas-Liquid Flow in Fixed-Beds Trickle-Flow Pulse-Flow Spray-Flow Bubble-Flow Mewes, Loser, and Millies (1999) 23 Three Key Factors Affecting Flow Regimes 1. Throughput of gas and liquid L - liquid mass velocity G - gas mass velocity L / G - ratio of mass velocities 2. Physical properties of the gas and liquid - viscosity - surface tension - density 3. Foaming or non-foaming characteristics of the liquid 24 Factors Affecting Choice of L / G • Stoichiometry of the reaction • Pressure drop limitations • Establishment of desired flow regime • Foaming characteristics of liquid • Heat removal requirement • Maximum allowed Tad 25 Flow Regime Map for Gas-Liquid Flow in Fixed-Beds Gianetto, Baldi, Specchia and Sicardi, AIChEJ (1978) 26 Effect of Bed Prewetting and Hysteresis Effects 250 non-prewetted bed prewetted bed Intensity of image 200 CCD Video Imagesof Liquid Flow in 2-D Beds Axial position: 6 cm down from top dp = 0.3 cm 150 100 50 0 0 3 6 9 12 15 18 21 24 X (dp) L = 3.52 Kg/m2.s channel flow film flow 28 Models for Trickling to Pulsing Flow Regime Transition • Macroscopic model - balance of inertial and capillary forces – Grosser, Carbonell & Sundaresan, AIChE J (1988) – Attou & Ferschneider, CES (1999) • Microscopic model - pore blockage by balance of inertial and capillary forces – Ka Ng, AIChE Jnl (1986) • Microscopic model - wave formation on surface of liquid film – Holub, Dudukovic & Ramachandran, AIChE J (1993) 29 Estimation of Pressure Drop for Two-phase Flow in Packed-Beds Various empirical correlations based on: • Lockhart -Martinelli parameter • Two - phase friction factor • Energy dissipation parameter • Relative permeability parameter • Other dimensionless parameters 30 Key Pressure Drop Equation Parameters • Single - phase pressure drop = P L 150 (1 - B ) 2 u = B 3 d pe 3 1.75 (1 - B ) u 2 + B 3 d pe • Lockhart -Martinelli parameter 1/2 L = G • Two - phase friction factor 1 f GL = LG d PE uG 2 G 2 Validity: • Low and high Interaction regimes • Non-foaming and foaming systems 31 Pressure Drop - Summary • Correlations based on single-phase gas and liquid P (Ergun equation) – Lockhart-Martinelli (1949), Larkins et al. (1961), Specchia & Baldi (1974) - separate for low and high interaction, Kan & Greenfield (1978) - hysteresis effect on P • Flow models – Relative permeability model: Saez & Carbonell, AIChE J (1985); Levec, Saez & Carbonell, AIChE J (1985); Saez, Levec & Carbonell, AIChE J (1985) – Slit model: Holub, Dudukovic & Ramachandran, CES (1992); AIChE J (1993); Al-Dahhan, Khadilkar, Wu, & Dudukovic IEC Res. (1998); Iliuta & Larachi, CES (1999) – Fluid- fluid interface model: Attou, Boyer & Ferschneider, CES (1999), Attou & Ferschneider, CES (1999) 32 Liquid Holdup - Key Definitions • Liquid holdup (HL , L ) is the fraction of reactor volume that is occupied by liquid (m3 liquid / m3 reactor). L = VL / VR • Liquid saturation (L , L ) is the fraction of external bed voidage (B ) occupied by liquid (m3 liquid / m3 voids). L = L / B • Fractional pore fill-up (Fi) is the fraction of catalyst pore volume occupied by liquid (m3 liquid / m3 pore volume). 33 Key Liquid Holdup Relationships Total Bed Voidage t = External Voidage + Internal Voidage = B + Total Liquid Holdup = External Holdup L = LE p ( 1 - B ) + Internal Holdup + L Internal Holdup for Liquid-Filled Catalyst Pores (Fi = 1) LI F i p ( 1 - B ) = External Liquid Holdup = Dynamic Holdup + Static Holdup LE = LD + LS 34 Typical External Holdup Values External Liquid Holdup = Dynamic Holdup + Static Holdup LE LD = + LS 0.1 < LE < 0.25 ( or higher at high L / G ) LS = NEo = 1 20 + 0. 9 NEo L g dp 2 B2 L (1 - B ) Gravity force Surface tension force = Eotvos Number 35 Liquid Holdup - Summary Contributions to the overall liquid holdup • Internal liquid holdup (inside particle) ~ equal to particle porosity • External liquid holdup – dynamic (flowing liquid) - depends on flow regime and is determined by viscous, gravity and inertial forces – static - volume fraction of liquid retained when a pre-wetted bed is drained, from balance of gravity and surface tension forces HL= HLD + HLSe+ HLi = HLD + HLSe+ ip(1- B) HL, HLD & HLe correlations for low & high interaction regime • Separate correlations for low and high interaction regimes • Empirical: Larachi et al. (1991), Lara-Marquez et al. (1992) • Phenomenological: Holub et al. (1992, 1993); Al-Dahhan & Dudukovic (1994) 36 Pressure Drop and Liquid Holdup Correlations MARE (%)* L P / L Iliuta & Larachi (1999) 18 27 Ellman et al. (1988, 1990) 23 54 Saez et al. (1985) 22 41 Al-Dahhan & Dudukovic (‘95, ‘96) 17 32 Larachi et al. (1991) 73 22 *Mean Absolute Relative Error Carbonell, O&G Sci & Tech, vol 55 (4) (2000) 37 Key Transport Resistances • Gaseous reactant resistances 1 - Gas-to-liquid resistance 2 - Liquid-to-solid resistance 3 - Intraparticle diffusion and kinetic resistances • Liquid reactant resistances 1 - Liquid-to-solid resistance 2 - Intraparticle diffusion and kinetic resistances • Heat transfer resistances 1 - Bulk gas-to-particle 2 - Bulk liquid-to-particle 3 - Intraparticle 38 Transport Parameter Correlations kLaB - Gas to liquid ( liquid - side ) volumetric mass transfer coefficient kSL - Liquid to actively wetted solid mass transfer coefficient kSg - Gas to dry solid mass transfer coefficient h - Overall heat transfer coefficient e - Effective conductivity of particles 39 Interphase Mass Transfer Correlations Summary Liquid side of gas-to-liquid mass transfer • Separate correlations for low and high interaction regimes • Wild et al. (1992); Larachi (1991); Cassanello et al. (1996) Gas side of gas-to-liquid mass transfer • For most situations negligible resistance • Gotto et al. (1977); Fukushima & Kusaka (1978) Liquid-to-solid mass transfer • Some have separate correlations for low and high interaction regimes • Goto & Smith (1975), Satterfield et al. (1978), Specchia et al. (1978) 40 Liquid - Solid Contacting in TBR’s • Incomplete liquid - solid contacting can occur due to: 1. Reactor- scale (gross liquid maldistribution) 2. Particle - scale (local catalyst incomplete wetting) • Internal particle incomplete contacting is unlikely in the absence of highly exothermic reactions • External particle incomplete contacting is likely in the trickle - flow regime when Lm < 5 kg / m2 - s 41 External Contacting Efficiency Low Gas-Liquid Interaction Regime CE = 1.617 ReL0.146 GaL 0.9711 or L < 5 kg / m2 s 0.224 CE = 1.021 D CE = 1.0 L > 5 kg / m2 s where: D = Dynamic liquid saturation D = LD / B 42 Liquid-Solid Contacting - Summary Combining flow pattern deviations from ideal liquid plug flow, and incomplete catalyst wetting: • Liquid not in plug flow and there is no radial mixing, but all catalyst is wetted • Liquid not in plug flow and extensive radial mixing, and all catalyst is wetted • Partial external wetting of catalyst • Partial internal wetting of catalyst Correlations for liquid-solid contacting: • Ruecker & Agkerman (1987), Ring & Missen (1991), Al-Dahhan & Dudukovic (1995) 43 Intraparticle Diffusion Resistance Conventional Thiele-modulus/effectiveness factor approach needs to be modified to account for partial external and intraparticle wetting: • Mills & Dudukovic (1980) solved the diffusion-reaction equations for partial external wetting for slab, cylinder and sphere-shaped particles • The numerical solution can be approximated by weighted average of effectiveness factor of totally wetted and totally dry particles, the weighting factor being the contacting efficiency TB = CE W + (1- CE) NW • Internal wetting effects have been largely ignored 44 Catalyst Effectiveness Factor for a Differential TBR • Assume: (1) Gas-limiting or volatile liquid-limiting reactant (2) First-order reaction (3) Incomplete external wetting, complete internal wetting • Approximate solution only possible for large modulus p 45 Overall Effectiveness Factor for a Trickle-Bed Reactor (limiting reactant in Gas phase), O o = CE P BiW 2 + P tanh P te rm due to inter nal pore diffus ion + exte rnal m ass tr ans fer CE x through the active ly w e tte d surface + of the catalyst pe lle t + (1 - CE ) P 2 P + BiD tanh P te rm due to inter nal pore diffus ion + exte rnal m ass tr ansfe r (1- ) x through the inactively w e tte d surf ace CE of the catalys t pe llet Increasing ηCE decreases conversion ! LHSV based scale-up alone is not suitable ! CE = external liquid-solid contacting efficiency CE < 1 for cocurrent downflow; CE = 1 for upflow 46 Overall Effectiveness Factor for a Trickle-Bed Reactor, (limiting reactant in liquid phase) O o = CE P P + BiW tanh ( P / CE ) 2 1 p k1 P Ls De H k L Bi w A gs s De Bi D k gls Ls Bi w k s Ls De 1 1/2 Increasing ηCE increases conversion ! LHSV based scale-up is suitable ! De 47 Trickle-Bed Reactor Catalyst Effectiveness Factors Overall effectiveness factor, O • Both external and internal transport resistances are included VP r V dV O = 0 VP r Vs observed average rate in the pellet = rate obtained if all the pellet exists at the bulk conditions adjacent to the pellet but outside the boundary layer around it 48 Comparison of Effectiveness Factors Calculated From Previous approximate Solution and Actual Numerical Simulation 49 Rigorous Multicomponent Diffusion Modeling - Gas Liquid Interphase Function Vector N1GL ctl {[ L ][k Lo ][]}1, j ( x Ij x Lj ) x1 (q) / x j o I L N GL c 2 tl {[ L ][k L ][ ]}2 , j ( x j x j ) x2 ( q ) / x j . . . N GL c o I L {[ L ][k L ][]}nc 1, j ( x j x j ) xnc 1 (q) / x nc 1 tl j GL o V I N c 1 tg {[ G ][k G ]}1, j ( y j y j ) y1 ( q ) / y j GL o V I N 2 ctg {[ G ][kG ]}2, j ( y j y j ) y2 (q ) / y j . FGL ( z , t ) . . o V I N GL c {[ G ][kG ]}nc 1, j ( y j y j ) ync 1 (q ) / y nc 1 tg j I I y1 K1 x1 I y2I K 2 x21 . yncI K nc xncI I I I I x1 x2 x3 ...xnc 1 y1I y2I y3I ...yncI 1 n GL G L qG qL N i ( H i H i ) i 1 3nc 1 Khadilkar et al., 1998 50 CREL General Geometry • Discuss MFS use here • See muthana. Eusebio paper 51 Level III TBR Model -Catalyst Scale EquationsExternally Half Wetted, Partially Liquid Filled Pellet Liquid Filled Zone d N iLC i Ri ,liq dx NiLC CiLC C nc 1 d C NtL {[ BL ]1[] CiL }i , j C CtL dx j 1 d N iLC H iLC d2 keL 2 TLC 0 dx dx | Wet Zone | Dry Zone | Lc Gas Filled Zone d C N iG i Ri , gas dx C NiG keG C nc CiG d C C N {[ BG ]1 CiG }i , j tG C CtG dx j 1 d 2 C d NiG H iG TG 0 dx2 dx C Khadilkar et al., 1998 C Intra-catalyst G-L Interface Continuity of temperature, mass and energy fluxes, and equilibrium relations for all species Liquid imbibition velocity v=NCtL/CCtL=(RP2/8 L)(P(x=Lc)-P(x=)+2cos/RP) 52 CREL Methods of Determining Contacting Efficiency • Tracer Method • Chemical Reaction Method 53 Prediction of TBR Multiplicity Effects • Hysteresis Effects Predicted 1.2 • Two Distinct Rate Branches Predicted (as Observed by Hanika, 1975) 1 Conversion 0.8 • Branch Continuation, Ignition and Extinction Points 0.6 Wet Branch(Exp t) 0.4 • Wet Branch Conversion (~30 %) Dry Branch (Exp t) Wet Branch (L-II) Dry Branch (L-II) 0.2 • Dry Branch Conversion (> 95 %) Wet Branch, (L-III) Dry Branch (L-III) 0 0 5 10 Hydrogen Feed Ratio, N 15 • Continuation of the dry branch • Thermal conductivity - L II model • Intracatalyst interface location-LIII model System: Cyclohexene hydrogenation 54 CREL Three Types of Catalyst for Highly Exothermic Reactions 55 Nonvolatile Liquid-Limiting Reactant Completely Wetted Catalyst ( ce = F i =1 ) Reaction : A (gas) + B (liquid) = P (liquid) • Kinetic rate : kVBS ( mol / m3 catalyst - s ) ( per unit catalyst volume ) • Rate in catalyst : kvPBS ( 1- B ) ( mol / m3 reactor - s ) ( per unit reactor volume ) • Transport rate : kLS ap BL - BS ) ( mol / m3 reactor - s ) ( per unit reactor volume ) 56 Overall or Apparent Reaction Rate Liquid-Limiting Reactant ( mol / m3 reactor - s ) P kV (1 - B ) BL = k app BL = BL 1 k LS aP + 1 P kV (1 - B ) 57 Plug-Flow Model for Scale - Up Nonvolatile liquid, 1st order reaction XB 3, 600 (1 - B ) k app = 1 - exp LHSV 3,600 uL LR where: LHSV = Using : • Same catalyst activity • Same size particles • Same packing procedure ( B ) • Same feed • Same Temperature 58 Gaseous-Limiting Reactant Completely Wetted Catalyst ( ce = F i =1 ) Reaction : A (gas) + B (liquid) = P (liquid) • Kinetic rate : kVAS ( mol / m3 cat - s ) ( per unit catalyst volume ) • Rate in catalyst : kv ( 1- B )PAS ( mol / m3 reactor - s ) ( per unit reactor volume ) • Transport rate : ( mol / m3 reactor - s ) ( per unit reactor volume ) 1. Gas - liquid KLaB ( AG /HA - AL) 2. Liquid-solid kLS aP AL - AS ) 59 Overall or Apparent Reaction Rate Gas Limiting Reactant (mol / m3 reactor - s ) A A P kV (1 - B ) G = k app G = H A H A AG H A 1 KL aB + 1 k LS aP + 1 P kV (1 - B ) 60 Reactor Performance for a Gas-Limiting Reaction with First-order Reaction A (gas) + B (liquid) XB = P (liquid) A * (1 - B ) 3600 k app BLO LHSV where: k app = CE 2 BiWG + tanh (1 - CE ) tanh + An increase in CE may decrease kapp so that equal LHSV for scale-up may not work, i.e., if kapp decreases as uL increases. 61 Scale-up Methodology for a Gas-Limiting Reaction • Keep same liquid hourly space velocity (LHSV) • Keep same ratio of liquid to gas mass velocities (L / G ) • Keep same packed-bed length (i.e., same L (uL) ) These criteria are often impractical to implement. Hence, a fundamental reactor model that captures the key phenomena is needed for scale-up or scale-down. 62 TBR Scale - Up for Aldehyde Hydrogenation Scale - up done based on equal LHSV with disastrous results Data Plant Laboratory Height (m) 19.4 0.235 Diameter (m) 0.455 0.0341 LHSV (h-1) 1.3 1.3 UL (LHSV) (mh-1) 26 0.26 H2 flow (STD) (m3h-1) 1000 - GHSV - 312 Pressure (bar) 65 - 80 70 Temperature (oC) 110 110 Bed porosity 0.425 0.425 Catalyst tablets 3 / 16 “ x 1 /8 “ (Vp / SP = 0.31 cm ) Conversion (XB) 0.40 0.90 63 Scale-up & Scale-down from Pilot Plant to Commercial Reactor • Catalyst orientation (flat surface preferred) • Addition of fines in pilot plant to simulate good liquid distribution and absence of wall effects • • Reactor internals - Inlet distribution - Quench zones with redistribution - Outlet collector geometry 64