Recent developments in modelling of industrial dryers Ian C Kemp Process Manual Product Manager, Hyprotech, AEA Technology plc, Harwell, UK

Transcription

1 Recent developments in modelling of industrial dryers Ian C Kemp Process Manual Product Manager, Hyprotech, AEA Technology plc, Harwell, UK Magdeburg - Slide 1

2 Scientific Approach to Dryer Design Overall System Model Equipment Model Material Model Material movement Gas flow patterns Heat transfer Drying Kinetics Drying Equilibria Product Quality Solids Handling Magdeburg - Slide 2

3 Design models Four levels of design: Heat and mass balance Scoping (approximate) design Psychrometric charts to give hot air flow rates Heat transfer area for contact dryers Scaling methods (integral model) Based on experimental batch drying curves Detailed (full) design Incremental models - stepwise integration Specialised models/techniques, e.g. CFD Magdeburg - Slide 3

4 Detailed calculations What do we want to use the models for? Three types of calculation: Design mode - design new dryer from basic spec, physical properties from databanks Performance mode - for existing dryer, find effect of changing operating conditions Scale-up - from laboratory-scale or pilot plant experimental data to new full size dryer Magdeburg - Slide 4

5 Heat and mass balance (continuous dryer) Heat losses Q wl Dry gas W G,Y I,T GI,I GI Wet solids W S,X I,T SI,I SI Dryer Wet gas W G,Y O,T GO,I GO Dry solids W S,X O,T SO,I SO Indirect heating Qin (Conduction, radiation, RF/MW) Mass balance: Heat balance: W ( Y Y ) = W ( X X ) G O I S I O WGI GI+ WSI SI+ Qin= WGI GO+ WSI SO+ Qwl Magdeburg - Slide 5

6 Scoping Calculations Design mode Throughput W S and inlet/outlet moistures X I, X O are known, need to calculate size of dryer required Select heat source temperature T, humidity Y I Continuous convective dryers: find outlet humidity, required air flow and hence cross-sectional area Continuous contact dryers: find evaporation rate and required heat transfer surface area, hence dimensions Batch dryers: find dimensions of dryer to physically contain batch, estimate required drying time Performance mode Size of existing dryer is known, deduce maximum drying duty (find W S or X I or X O given the other two) Magdeburg - Slide 6

7 Use of Psychrometric Chart Mollier Chart for Air/Water at kpa Enthalpy (kj/kg) Boiling Pt Triple Pt Sat. Line Rel Humid Adiabat Sat Spot Point Gas Temperature (C) Gas humidity (g/kg) Magdeburg - Slide 7

8 Drying Models Scoping design - initial approximate sizing Scaling/Integral Incremental CFD Local conditions Layer dryers Fluidised beds (simple models) Flash dryers Spray drying, complex flows Rotary dryers Magdeburg - Slide 8

9 Drying Kinetics Moisture Loss from Solid as a Function of Time Model Mass Transfer from Surface Internal Mass Transfer Characteristic Drying Curve Receding Evaporative Front Diffusion Measure Periodic Weighing Continuous Weighing Humidity Difference Magdeburg - Slide 9

10 Drying Kinetics Induction / Unhindered (Constant Rate) Periods Drying rate depends on external conditions Fairly easy to calculate and scale (by T or p) Falling Rate (Hindered Drying) Period Multi-phase moisture transport by: diffusion, convection, capillary action, adsorption etc. Drying rate depends on many parameters which cannot be measured easily, e.g. internal pore structure Hence difficult or impossible to calculate rigorously from first principles, though models exist e.g. Whitaker Should always be measured by experiment Magdeburg - Slide 10

11 Mass Transfer Drying Kinetics The Characteristic Drying Curve Concept Evaporation Wet Particle Drying rate per unit exposed surface = Mass Transfer Humidity f, Relative Coefficient x Difference x Drying Rate (Configuration (Driving (Material dependent) Force) dependent) Magdeburg - Slide 11

12 Drying rate curves for reference case f= N N cr Advanced model CDC φ = X X cr X X eq eq Magdeburg - Slide 12

13 Drying curves X (kg/kg) Advanced model CDC Time (s) Magdeburg - Slide 13

14 Processing of kinetics data Moisture - time curve Usually OK; can smooth using cubic spline Periodic sampling may give few, scattered points Humidity - time and drying rate - time curves Invariably more jagged than moisture curve Can be smoothed, but retain raw data Rate - moisture (Krischer) curve Also tends to have fluctuations, especially at low rate Great care needed if drying times are back-calculated Magdeburg - Slide 14

15 Unsmoothed drying curves Magdeburg - Slide 15

16 Integral (Scaling) Model Considers the dryer as a whole X Et () X() t dt 0 X O O = is the mean outlet moisture content E(t) is a residence time function; for batch dryers, all particles have same residence time τ X(t) is the drying curve function, found by scaling an experimental drying curve Magdeburg - Slide 16

17 Drying Curves for Well Mixed Fluid Bed Batch Drying Curve Moisture content, kg/kg Time, seconds Outlet moisture content, kg/kg Design Curve X I =0.3 X O = t 1 =220 Z =7.85 t 2 = Mean residence time, seconds IBBDC IITBDC Magdeburg - Slide 17

18 Scaling factors for X(t) Normalisation factor Z from pilot- to full-scale Factors involved in scale-up (modify time axis): Gas temperature T GI or bed temperature T B Gas velocity U G - as mass velocity (flux) G Bed depth z - as bed weight per unit area m B /A B e.g. Plug-flow and batch units: Type A: Type B: τ 2 Z = = τ 1 τ 2 Z = = τ 1 ( mb/ A) G TGI Tw b) m / A) G T T ) ( ( ( T T ( ( B 1 2 GI w b 2 GI GI T T w b w b ) ) 1 2 Magdeburg - Slide 18

19 Recent analysis of fluid bed scaling rules Original rules based on experimental results Why Types A and B, and how about transition? Recent rigorous derivation gives: τ 2 X τ X Type A; exponential term tends to zero Type B; (1-e -f.ntu.z )=fk Y φaz/g, so Z τ 2 = τ 1 = = f. NTU. z ( mb / A) G1( TGI Twb ) ( 1 e ) 2 1 f. NTU. z ( m / A) G ( T T ) ( 1 e ) ( mb / A) G1( T ) {( ) } 2 GI Twb m 1 B / A / G 1 1 ( m / A) G ( T T ) {( m / A) / G } B B GI GI wb wb 2 2 B = = τ 2 τ ( TGI Twb ) 1 ( T T ) 2 GI = wb Z Magdeburg - Slide 19 1

20 Effect of NTU Mollier Chart for Air/Water at kpa Enthalpy (kj/kg) Gas Temperature (C) Gas humidity (g/kg) Magdeburg - Slide 20

21 Drying curves for fluidised bed Moisture content E-04 Time Falling-rate drying But external conditions (G, z) affect drying strongly -Type A NTU>10, drying is controlled by heat content of inlet air except in very final stages 9.00E E E E E E E-04 Drying rate 5.00E E-04 Drying rate 5.00E E E E E E E E E Time 0.00E Moisture content Magdeburg - Slide 21

22 Temperatures in bed Temperature Gas - Bottom Layer 1 Gas - Lower Layer 2 Gas - Upper Layers 3/4 Particles Time Magdeburg - Slide 22

23 Integral (scaling) model - summary Basic concept: scaling an experimental batch drying curve to new conditions / throughput Similar to scoping method, but allows for falling-rate drying kinetics and heat transfer Implicitly assumes the characteristic drying curve concept (CDC) applies Effective for many layer and batch dryers Will have problems if heat transfer or kinetics change on scale-up or are not limiting factor Magdeburg - Slide 23

24 Incremental Model Stepwise integration along duct, drum or bed dq Wl WG, Y, TG, UG Gas WS, XT, S, US z dz Solids Magdeburg - Slide 24

25 Incremental Model Over a small increment of time, dt Equations involved: Gives: Heat transfer to particle surface Q P Mass transfer from particle dx/dt Mass balance on moisture X, Y Heat balance on particle T S Heat balance over increment T G Particle transport and velocity U P, z Local gas properties, e.g. density U G Magdeburg - Slide 25

26 Incremental Model Detailed Equations Heat transfer to solid Q = ha ( T T ) = m λ N Drying dx = N = function( X, Y, T, T, h, a ) = fn dt Function f obtained by drying kinetics test or model Heat balance on solid S S G S S ev cr S G S cr dx ha T T C C X dx dt S ( G S ) = ev + dt PS + PL + λ 2 dt S Magdeburg - Slide 26

27 Incremental Model Detailed Equations Material transport dz = U dt S Mass balance WdX= WdY S G Heat balance (ignoring second order terms) (( + ) + ) ( ) ( ) W C C X dt C T dx S PS PL S PL S ( ) ( ) + W C + C Y dt + C T + λ 0 dy + dq = 0 G PG PY G PY G Wl Magdeburg - Slide 27

28 Force Balance: m du P dt P Particle Motion in a Vertical Flash Dryer ρ 2 = C U U A m g m f U DS ( ) 2 2D G G P xs P P P P Acceleration Drag Weight Wall Force Force Term Friction f P = solids-wall friction factor K f = f P U P Acceleration for spherical particle/agglomerate: 2 du dt 2 3C ρ U = g 4d aρ P D G R P PW fu P 2D 2 P Magdeburg - Slide 28

29 Results before fitting to pilot plant data Magdeburg - Slide 29

30 Fitting Calculations Results after fitting Changes: d P(SM) x ins K f Magdeburg - Slide 30

31 Scale-up Method Obtain results from existing dryer Pilot plant or current operating unit Test against model in Fitting/Pilot mode Compare theory with actual results Adjust model parameters Fit model to experimental results Design new full-scale plant Using optimised model as above Magdeburg - Slide 31

32 Case Study: Effect of Gas and Solid Flowrate Magdeburg - Slide 32

33 Drying in layers Top Boundary Layer N Layer 2 Layer 1 Bottom Boundary One-dimensional vertical incremental model for layer dryers Gives temperature, humidity and moisture profiles through bed e.g. deep-layer grain dryer Can be extended to 2-D or 3-D grid by combining with horizontal increments e.g. thick-layer band dryer Magdeburg - Slide 33

34 Computational Fluid Dynamics (CFD) Rigorous three-dimensional model Solves Navier-Stokes equations Fine mesh grid in body-fitted coordinates Requires extensive computing e.g. CFX Tested on industrial dryers e.g. by SPS Verification by observations of flow patterns Small-scale and industrial spray dryers Gas flow patterns and particle tracking Feedpoint of pneumatic conveying dryers Still limitations and unknowns for internal transport Magdeburg - Slide 34

35 Model verification by LDA measurement Magdeburg - Slide 35

36 Spray dryer simulation with CFD Particle trajectories with no swirl - short particle residence times - high final moisture content Particle trajectories with swirl - longer residence times - more effective use of chamber - lower final moisture content Magdeburg - Slide 36

37 Conclusions Drying has often been a graveyard for pure theory Simple heat and mass balances are very useful Integral model is effective for fluidised beds and for most layer and contact dryers One-dimensional incremental model works well for pneumatic conveying and rotary dryers CFD useful in spray dryers and swirling flows Models are generally more reliable for scale-up than for design from published data only Experimental characterisation of a new material in lab or pilot-plant is essential Magdeburg - Slide 37

38 Typical solids process flowsheet FEED Solution Preparation e.g. Solvent Extraction Effluent Processing Wastewater treatment Gas Cleaning e.g. Bag Filter Solution Filtrate Particles, VOC s Particle Formation e.g. Crystallization Solid-Liquid Separation e.g. Filtration Solids Drying e.g. Fluidised Bed Slurry Wet cake Dry solids PRODUCT Slurry handling and Pumping Wet Solids Handling e.g. Screw Feeder Post-processing e.g. Comminution Magdeburg - Slide 38

39 Overall Process Concept Chemical Development Molecular deconstruction Select synthesis strategy NCE from discovery Isolation development Evaluate potential isolation routes Crystallization Establishes: Polymorph Purity A Yield CSD Habit Defines: SLS task SLS Establishes: Purity B May alter: Yield CSD Defines: Drying task Drying Removes: Solvent(s) May alter: CSD Agglomeration Lumping Breakage Defines: Micronisation Micronise Delumping CSD reduction Provides consistency Bulk product Magdeburg - Slide 39