STUDY OF GAS-LIQUID MASS TRANSFER IN A GRID STIRRED TANK

Transcription

1 STUDY OF GAS-LIQUID MASS TRANSFER IN A GRID STIRRED TANK T.Lacassagne M. ElHajem F.Morge S.Simoëns J.Y Champagne Laboratoire de Mécanique des Fluides et d Acoustique UMR CNRS 5509 Ecole Centrale de Lyon Université Claude Bernard Lyon 1 INSA de Lyon tom.lacassagne@insa-lyon.fr

2 Outline I. Introduction II. Experimental techniques 1. Oscillating grid turbulence 2. Concentration measurements 3. Velocity measurements III. Results and discussion 1. BPLIF 2. SPIV 3. Simultaneous measurements IV. Conclusion 2

3 Introduction Objectives: Dissolution and mixing at gas-liquid interfaces Understand the influence of turbulence Model its impact on mixing Predict mass transfer velocities Applications: Energy and environment (atmospheric gas dissolution into oceans, lakes ) Chemical engineering (Dissolution into non Newtonian fluids) Pharmacy and health (vaccines production ) Studied case: Flat interface Bottom shear turbulence Gaseous CO 2 dissolution into water Principle of the experiment Presented results: ff = 1HHHH, SS = 4,9cccc, PP CCOO2 = 1aaaaaa, HH gggggggg ssssssssssssss = 45cccc 3

4 II. Experimental techniques 1. Oscillating grid turbulence 2. Concentration measurements 3. Velocity measurements 4

5 Setup SPIV BPLIF Coupling Spatial resolution mm mm mm Size of observation field 20x20 mm 20x20-20x200 mm 20 mm Max. acquisition frequency 4 Hz 10 Hz 4 Hz 5

6 Oscillating grid turbulence Oscillating Grid Turbulence (OGT) Thompson & Turner 1975, Hopfinger & Toly 1976 Janzen et al 2010 Principle: Oscillating motion of a grid with square section bars Interaction of wakes and jets Turbulence diffusion towards the interface Advantages Low mean velocity Nearly homogeneous turbulence in an horizontal plane Grid parameters: MM = 35mmmm, dd = 7mmmm Solidity: σσ = dd MM 2 dd MM = 0,36 < σσ mmmmmmmmmm = 0,4 6

7 Oscillating grid turbulence Hopfinger and Toly s equations, 1976 (No gas-liquid interface) 1 uu RRRRRR = AA. zz 1 ; ww RRRRRR = CC 2 HHHH uuu AA = CC 1HHHH ffss 1,5 MM 0,5 ; LL = ααzz 1 RMS velocity profiles in the tank obtained by PIV 7

8 Concentration measurements Blocked Planar Laser Induced Fluorescence (BPLIF) Homogeneous fluid seeding with a fluorescent molecule Quenching/Inhibition/Blocking of local fluorescence intensity by a «quencher» Application to dissolved CCOO 22 tracking CO 2 dissolution decrease in ph decrease in ph decrease in fluorescence intensity («quenching») Example of Blocked fluorescence 8

9 Concentration measurements: II pppp [CCOO 22 ] II CCCCCCCCCCCCCCCCCCCCCC pppp: global behaviour of type XX XX = II oooo εε pppp [HH + ] + solving of chemical equilibria [CCOO 22 ] XX rrrrrr ~ tttt(aa. pppp + BB) ; avec Intensity ph relationship 9

10 Velocity measurements: PIV Particle Image Velocimetry (PIV) Classic: 2 Dimensions, 2 Components Stereoscopic (SPIV): 2 Dimensions, 3 Components SPIV principle for a single particle e Laser sheet thickness dd = dd xx, dd yy, dd zz xxxxxx Real displacement in absolute space (3D) dd ii = dd ii xxxx, dd iizzzz xx ii yy ii Displacement measured on sensor i in space i (2D) α Stereo inclination angle SPIV objectives Quantify horizontal components of mass flux Get velocity access to horizontal structures foreseen by horizontal BPLIF images Scalar structure in an horizontal plane underneath the interface 10

11 Results and discussion 1. BPLIF 2. SPIV 3. Simultaneaous measurements 11

12 Results and discussion BPLIF: Observations Subsurface C at ff ggrrrrrr = 1HHHH, ff aaaaaa = 4HHHH, ff vvvvvvvvvv = 8HHHH CC = CC CC bbbbbbbb CC ssssss CC bbbbbbbb CC ssssss = cccccc CC bbbbbbbb = ff(tt) 12

13 Results and discussion BPLIF: Concentration profile Scalar sublayer Taking turbulence into account (mean profile) Used in many mass transfer models (eg: KK LL = DD δδ cc = 3, mm. ss 1 ) Mean concentration profile and scalar sublayer 13

14 Results and discussion SPIV: Instantaneous velocity fields Evidences of 3D structures 3D XZ Plan Scalar structure in an horizontal plane underneath the interface VV yy 14

15 Results and discussion Simultaneous measurements: instantaneous fields 2D structures (XZ plan) Simultaneous concentration and velocity measurements for ff gggggggg = 1HHHH, ff aaaaaa = 4HHHH, 2D vue 11 cccc. ss 11 15

16 Results and discussion Simultaneous measurements: instantaneous fields 3D structures (3rd component influence) Simultaneous concentration and velocity measurements for ff gggggggg = 1HHHH, ff aaaaaa = 4HHHH, 3D vue 11 cccc. ss 11 16

17 Results and discussion Simultaneous measurements: Flux Coupled fields turbulent transport phenomena: Concentration flux: jj zz = DD CC + cc www (1D) or jj ii = DD CC xx ii + cc uu ii (3D) 2 variables probability density functions (Variano & Cowen 2013) Quadrants identification 4 variables conditional analysis and comparison with mean flow profiles (Vinçont et. al 2000) Impact of each events type and comparison with mean gradients profiles Example of each type of event s probability as a function of depth 4 variables, var > 0 or var < = 16 Types 17

18 Conclusion Assessment on experimental techniques Good use of OGT for this type of study Laser techniques coupling should lead to a better understanding of mass transfer phenomena (2D-3C approach) First results on CCOO 22 mass transfer study Conclusions Thin sublayer rule mass transfer Injections and renewal events complex and linked Upcoming Conditional analysis and identification of event types contributing the most to mass transfer To Non-Newtonian fluids (viscoelasticity shear thinning) New turbulence characteristics close to the interface? Influence on velocity and scalar sublayers? Impact on renewal and injections events? 18

19 Extra slides A few scales and numbers Experimental setup OGT additional information Turbulence characteristics Injection and surface renewal Simultaneous measurements Options Binning About Fluorescein-Sodium General information Impact on chemical equilibria I-pH calibration 19

20 Results and discussion SPIV: Trubulence characterstics close to the interface Notion of «surface influenced layer» : uu zz < uu xx, uu yy defined by δδ LL, (LL integral length scale) Concept of a «viscous sublayer» successfully checked: RMS peak at zz 1 δδ vv Profiles of the 3 RMS velocity components scaled by their values at zz = δδ vv δδ vv = LL RRee TT = 33, 88 mmmm Brumley & Jirka (1987) 20

21 Introduction A few scales and numbers Flow: Grid Stroke: Oscillations frequency: SS 0,25.. 0,49 mmmm ff HHHH Grid-based Reynolds number: RRee HHHH = ffss1,5 MM 0, Brumley & Jirka (1987) Reynolds number: 20.νν RRee TT = uu.2.ll νν Taylor-based Reynolds number: RRee λλ [0.. 30] [0.. 80] Dissolution CCOO 2 partial pressure: PP CCOO2 0, aaaaaa Corresponding phs in distilled water: phh eeee [5,65.. 3,95] Schmidt number: SSSS 527 nn 1 kkcc CCOO 2 Hatta number: HHHH = δδ cc 3 DD CCOO 2 Sublayers Surface influenced layer: Viscous sublayer: Scalar sublayer: δδ ~30 mmmm δδ vv = δδ RRee TT 0,5 ~3 mmmm δδ cc ~0,5 mmmm 21

22 Experimental setup 22

23 Oscillating grid turbulence Hopfinger and Toly s equations, 1976 (No gas-liquid interface) 1 uu RRRRRR = AA. zz 1 ; ww RRRRRR = CC 2 HHHH uuu AA = CC 1HHHH ffss 1,5 MM 0,5 LL = ααzz 1 Based on grid axis Successfully checked by every experiment in literature since then Avec uu RRRRRR, ww RRRRRR : RMS of horizontal and vertical velocity ff: Grid oscillations frequency S: Stroke ZZ GG : Distance from grid average position LL : Turbulence integral length scale CC 11HHHH, CC 22HHHH,αα: H&T constants 23

24 SPIV in a liquid medium Refraction phenomenon Air/water interface beam refractions: influence of the angle of incidence Correction using a liquid prism (Prasad & Jensen 1995) Scheimpflung angle Refraction: Usual Scheimpflug condition is invalid. (Prasad 2000) Need of an adjusted Scheimpflung angle to optimize the depth of focus Liquid prism and Scheimpflung condition from Prasad & Jensen (1995) 24

25 Binning Light filtering Fluorescence for BPLIF Bandpass filter centred around 515 nm (S)PIV if needed high pass filter (λλ cc < 500nnnn) or low pass filter (λλ cc > 520nnnn) depending on laser wavelength Adjustment of spatial resolution: binning BPLIF resolved up to pixel size (S)PIV resolved up to PIV window size Binning: Reduction of BPLIF information to PIV resolution (Mean value, centre value ) 25

26 About Fluorescein-Sodium: General informations Fluorescein Sodium: Studied by Martin & Lindqvist (1975) Dissociates into several aqueous forms, fluorescent or non fluorescent Acid base equilibria yield a change in Absorption Fluorescence intensity Absorption peak at λλ = Fluorescence peak at λλ = Aqueous forms of fluorescein according to Martin & Lindqvist 1975 Fluorescence spectrum according to Martin & Lindqvist

27 About Fluorescein-Sodium: impact on chemical equilibria Acid-base reactions of CCOO 22 into water should not be modified by those of fluorescein sodium Solving of full system of acid-base equations with a given fluorescein concentration 27

28 II pppp [CCOO 22 ] pppp [HH + ] + solving of theoretical chemical equilibria [CCOO 22 ] Exponential behaviour Strong dependence on initial ph and water composition Notion of total dissolved concentration (DIC) and aqueous CO 2 Experimental validation: ph-meter and CCOO 2 aaaa probe ph-co2 theoretical curves for two initial phs 28

29 About Fluorescein-Sodium: impact on chemical equilibria Acid-base reactions of CCOO 22 into water should not be modified by those of fluorescein sodium Impact of total fluorescein concentration for different initial phs 29

30 About Fluorescein-Sodium: impact on chemical equilibria Acid-base reactions of CCOO 22 into water should not be modified by those of fluorescein sodium Impact of total fluorescein concentration for different initial phs 30

31 About Fluorescein-Sodium: impact on chemical equilibria Acid-base reactions of CCOO 22 into water should not be modified by those of fluorescein sodium Impact of total fluorescein concentration for different initial phs 31

32 II = ff pppp - BPLIF Calibration Option 1: Single calibration curve Hypothesis: Mapping of excitation intensity I0 does not depend on ph Consequence: Single curve II II 0 = ff(pppp) for every pixel Option 2: Pixel by pixel calibration accounting for variations of extinction coefficient variations (Valiorgue et. al 2013, Souzy 2014) Principle: Recording of several homogeneous ph pictures for a constant laser exposition II = ff pppp curves fitted for each pixel Hypothesis: Beer-Lambert law for optically thick systems where extinction coefficient ε depends on ph II ff,pppppp II ff,rrrrrr,pppppp = φφφφ φφ rrrrrr εε rrrrrr ee cc εε rrrrrr εε ddrr LLpppp pphh00 Case a: Reactive zone LL pppp pphh00 small (Valiorgue et. al 2014) II ff Fluoresced intensity φφ Fluorescence quantum yield εε Extinction coefficient cc Fluorescein concentration II ff,pppppp II ff,rrrrrr,pppppp φφφφ φφ rrrrrr εε rrrrrr = ff(pppp) 32

33 II = ff pppp - BPLIF Calibration Case b: Wide reaction zone fully included in observation area (Souzy 2014) Progressive ph reconstruction by columns of pixels Knowlege of ph in column i Computation of ee cc εε rrrrrr εε ddrr LLpppp pphh 0 Between i and i+1 Computation of φφφφ = II ff,pppppp ee cc εε εε rrrrrr ddrr LLpppp pphh 0 φφ rrrrrr εε rrrrrr II ff,rrrrrr,pppppp And ph transformation through calibration Case c: Reaction zone wider than observation area Wider observation area needed to visualize full laser path before interest zone 33