DispersionLecture 13, 13.12.2017, Dr. K. Wegner
Mass Transfer
Mass Transfer – Dispersion 13-2
to disperse = to spread widely
Dispersion
1. Dispersion is the act or process of dispersing, of distributing one substance (small volume fraction) in another (larger volume fraction).
2. Dispersion is the state of being dispersed.
Examples: Paints: pigments are dispersed in water or solvent Fog, clouds: water droplets are dispersed in air Sooting candle: soot particles in air. Pollution of rivers, air,... Spreading of diseases, hamsters,...
Mass Transfer – Dispersion 13-3
Dust Clouds from Africa
Examples
Global travel of dust carries microbes across oceans and continents.
Mass Transfer – Dispersion 13-4
Pollen Clouds
In 2006, birch pollen from Denmark traveled across the North Sea to England.
Water droplets dispersed in air appear as fog/clouds (and they create rainbows)
Atmospheric Clouds
Mass Transfer – Dispersion 13-5
Suspensions and Emulsions
Paint, for instance, is a dispersion of pigments in a solvent.
Homogeneous dispersion of the pigments in the fluid as well as long-term stability of the dispersion (shelf-life) are important.
Mass Transfer – Dispersion 13-6
Soot, smoke
Condensation plumes for 500 ft and 250 ft stacks in Salem (MA), showing the complex thermal structure in the lower atmosphere.
Mass Transfer – Dispersion 13-7
Dispersion on the microscopic level involves diffusion (of droplets, particles, molecules,...). On the macroscopic level dispersion is governed by fluid dynamics (laminar, turbulent flow, eddy formation).
The mechanisms causing the mixing and dilution e.g. of the plumes are hydrodynamic. Moreover, the governing equations are very complex and an exact solution is impossible, especially if the variability, e.g. of the weather is included.
Goal: To understand how e.g. plumes emitted by industrial facilities and ships behave.
Mass Transfer – Dispersion 13-8
The dispersion equation
It is impossible to calculate the actual concentration at a certain point for a certain time but we try to calculate the average concentration.
This has been done before for molecules diffusing in a medium.
We try to apply the mathematics applied to diffusion to dispersion.
Let's start with a rather simple problem: A point source only active at time t = 0, e.g. an emission pulse of smoke from a stack.
Mass Transfer – Dispersion 13-9
The concentrated solute originally located at z = 0 diffuses as the Gaussian profile shown.B.C.:t = 0:
Decay of a pulse (see Chapter 2, Fick's 2nd law)
)z(AMc1 δ=
δ(z): Dirac fct.
t > 0: z = 0:
z → ∞ : c1 = 0
01 =∂∂
zc
z → ∞ : c1 = 0
Mass Transfer – Dispersion 13-10
21
21
zcD
tc
∂∂
=∂∂
Fick's 2nd law:
The concentration profile is (Chap. 2): Dt4z
1
2
eDt4
1AMc
−
π=
where M is the mass of the solute and A the cross-sectional area over which diffusion (dispersion) is occurring.
For particles with concentration N0 (#/m3) released at t = 0 and z = 0:
2
2
p znD
tn
∂∂
=∂∂
and Dt4z
p0
2
etD4
1Nn−
π=
with the particle diffusivity Dp, see "Diffusion coefficients"
Mass Transfer – Dispersion 13-11
However, measurements show that after about 1h a plume has spread 1 km! Use the dispersion coefficient E instead of the diffusion coefficient D! The dispersion coefficient must be determined experimentally and typically has different values in different directions.
As we have seen in Chapter 2, the Fourier number for mass transfer can indicate for an unsteady state problem how far (how long) mass transfer has occurred.
1L
tDFo 2 ≈⋅
=Assume
According to mass transfer by diffusion in gases with D = 10-5 m2/s, the width of the peak roughly should be after 1 hour:
m 2.0s3600sm10tDL
25 ≈⋅=⋅= −
Mass Transfer – Dispersion 13-12
The source of the plume is at (0,0) and wind is blowing at speed v0
in the x-direction. We want to know the average pollutant concentration at (x, y).
Dispersion in one or more dimensions
( )tE4
y
0y
1y
2
ev/xE4
1c ⋅−
⋅⋅π∝
Based on this general equation for one-dimensional dispersion in y direction, a variety of concentration profiles have been derived.
Mass Transfer – Dispersion 13-13
Concentration profiles for free dispersion
Mass Transfer – Dispersion 13-14
Forest fire in southern France (Le Boulou) near the Spanish border in July 2017. Vacation home of the lecturer was x = 3 km away from the point source. v0 ≈ 60 km/h.Photo: L'Indépendant (J. Gallardo)
Smoke plume of a forest fire follows the dispersion pattern of this pulse decay
Mass Transfer – Dispersion 13-15
Dispersion coefficients ...
- are rather independent of chemistry (no material property)- are strongly dependent on flow (velocity, lam. / turb.)- depend on position- have different values in different directions- typically must be determined experimentally
Mass Transfer – Dispersion 13-16
In 1905, five muskrats ("Bisamratte") escaped in Bohemia. The animals quickly spread over Europe, as shown in the Figure.
Define a dispersion coefficient similar to the diffusion coefficient(A. Einstein):
Skellam, J.G. (1951) Biometrika 38, 196.
Et2r2 =
Mass Transfer – Dispersion 13-17
Example 1: Chemical spillA container with chemicals breaks, accidently releasing its content into a river flowing at v = 0.6 km/h. When experts of the environmental protection agency arrive to take water samples, the maximum concentration of 860 ppm is located 2 km downstream the release point. 50 m from the maximum, the concentration is 410 ppm.
a) How large is the dispersion coefficient? b) What will the maximum concentration be 15 km downstream?
t=0initial spill
v = 0.6 km/h
t1, s = 2 km: c1max= 860 ppm
Mass Transfer – Dispersion 13-18
Solution:Start with a mass balance for a section of river dz around the maximum concentration, moving with velocity v:
11 j
zdtdc
∂∂
−= withzcEj 1
1 ∂∂
−= 21
21
zcE
tc
∂∂
=∂∂
→
Boundary conditions:
t > 0: z = 0 : z → ∞ : c1 = 001 =∂∂
zc
z → ∞ : c1 = 0)z(AMc :0t 1 δ==
Mass Transfer – Dispersion 13-19
Concentration profile: Decay of a pulse
Et4z
ionconcentrat maximum measured
1
2
e Et4
1AMc
−
π=
a) So, 50 m away from the maximum: ( )
−
−
⋅== hs3600
hkm6.0km2E4
50m
1
2
eppm 608 ppm 410c
scm700E
2
= (compare diffusion coefficients: 10-5 cm2/s)
b) Note: t
1~c max,1 , thus ( ) ppm314km15km2ppm860km15c1 =⋅=
Mass Transfer – Dispersion 13-20
Example 2: Turbulent flow in gas pipelineA 3 km long pipeline with 10 cm diameter is used to transport different gases from a storage area to a chemical reactor. The gas velocity is 5 m/s.After switching from gas A to B, how much will the gases mix?
First check the condition of flow:
000'50s Pa 1010
m 1.0sm 5
mkg 1dvRe 6-
3=
⋅
⋅⋅≈
η⋅⋅ρ
=
Turbulent flow! We'll treat laminar pipe flow afterwards!
Mass Transfer – Dispersion 13-21
Plug flow profile
Assumption: Well mixed radially but concentration changes in axial direction.
ABz
r
v = 5m/s
Mass balance for point at the interface moving with v = 5 m/s:
21
21
zcE
tc
∂∂
=∂∂ B.C.: t = 0, z > 0: c1 = c1∞
t > 0, z = 0: c1 = c10
t > 0, z = ∞: c1 = c1∞
where c10 is the average concentration of the two gases. ∞= 110 c21c
Mass Transfer – Dispersion 13-22
The solution is similar to the general semi-infinite slab solution:
Et4zerf
cccc
101
101 =−−
∞
The dispersion coefficient for turbulent dispersion in pipelines is approximately:
v2dE ⋅≈ ,thus in this case E ≈ 2500 cm2/s
Let us consider a concentration change significant for Et4z =and closer. erf(1) ≈ 0.84, thus c1 = 0.84 c1∞+ 0.16 c10 = 0.92 c1∞
This is the case for m 24
sm5.0
m 3000s
m 25.04z2
≈⋅⋅=
Mass Transfer – Dispersion 13-23
Example 3: Laminar flow – Taylor dispersionA typical example for dispersion is the spread of a solute pulse in steady laminar flow.
The solution is dilute The flow is always laminar, no friction (no axial change in velocity) Mass transport is by axial convection and radial diffusion only
We can do an analysis using the following assumptions:
Injection of solute pulse
vzR0
Mass Transfer – Dispersion 13-24
Goal: Prediction of the dispersion coefficientLet us look at the solute mass balance of a ring-shaped region with dimensions Δr and Δ z, while the radius of the ring is r:
( ) 1z11
r11 r
znn
r1rn
rr1
tc
+∂∂
−θ∂
∂−
∂∂
−=∂∂ θ (1)
This simplifies to:
( )zvcrj
rr1
tc z1
r11
∂∂
−∂∂
−=∂∂
Velocity profile for laminar tube flow:
( )
−=
2
0
0z R
r1v2rv
(2)
(3)average velocity max0 v
21v =
Mass Transfer – Dispersion 13-25
Eq. (2) can be transformed by substituting Fick’s law and eq. (3) into:
zc
Rr1v2
rcr
rrD
tc 1
2
0
011∂∂
−−
∂∂
∂∂
=∂∂ (4)
Boundary conditions:
t = 0, all z: )z(RMc 2
01 δ⋅
π
=
t > 0, r = 0 : 0rc1 =∂∂
r = R0: 0rc1 =∂∂
Mass Transfer – Dispersion 13-26
One solution to this equation is:
−⋅
ζ∂
∂+= ∗∗
==
∗
∗
42
0r
100
0r11 r21rc
DRv
41cc (6)
Let us “sit” on a reference frame moving with velocity v0 to observe the action happening on the z-axis:
0
o
Rtvz −
=ζ0R
rr =∗Substitute and
So eq. (4) becomes:
ζ∂∂
−=
∂∂
∂∂ ∗
∗∗
∗∗12
001 cr
21Rv2
rcr
rrD
(5)
Mass Transfer – Dispersion 13-27
New approach with further simplification:Assuming that the radial variations in c1 are small relative to the axial ones (esp. for fast diffusion), we can work with the radial average:
∫ ∫ ∗∗=⋅⋅ππ
=0R
0
1
0112
01 drcr2dr)z,r(cr2
R1c
Now we think of this concentration as a solute diffusing in the moving reference frame and rewrite the mass balance:
( )0
11Rj
tc
ζ∂∂
−=∂∂
Here 1j (the average flux in flow direction) is equal to:
∫ =∗−−⋅π
π= 0R
0 0r110
z20
1 dr)cc)(vv(r2R1j
(8)
(9)
where tvzR 00 ⋅−=⋅ζ
(7)
Mass Transfer – Dispersion 13-28
Eq. (8) can be written as:
( )( )
⋅⋅
−
ζ∂∂
−=ζ∂
∂−=
τ∂∂
=∂
∂ ∗∗∗∫ drcr21r4vjc
Rtvc
1
1
0
011
00
1 2
Inserting eq. (6) and using eq. (7) we have thus:
21
20
01 c
D48Rvc
ζ∂∂
=
τ∂∂
The boundary conditions for the new equation are similar to the ones for the "decaying pulse" problem, namely:
(10)
( )ζδ⋅π
=ζ∀=τ
=ζ∂
∂=ζ>τ=±∞→ζ>τ
30
1
11
RMc : ,0
0c :0 ,0 ;0c : ,0
Mass Transfer – Dispersion 13-29
The solution to this problem is a Gaussian curve:
tE4)tvz(
1
20
11
20
etE4)R(Mc
−−
ππ
= (11)
The dispersion coefficient is given by (Taylor, 1953):
=
D48RvRvE 0
0
00
1 (12)
Note that E depends inversely on the diffusion coefficient! Rapid (radial) diffusion leads to small (axial) dispersion.
Taylor, Sir G. (1953), "Dispersion of soluble matter in solvent flowing slowly through a tube", Proc. Royal Soc. A 219, 186-203
Mass Transfer – Dispersion 13-30
Rapid (radial) diffusion leads to small (axial) dispersion.
=
D48RvRvE 0
0
00
1
radial diffusion
Mass Transfer – Dispersion 13-31
Correction for axial diffusion: Taylor-Aris Dispersion
∂∂
+
∂∂
∂∂
=∂
∂+
∂∂
21
21
011
zc
rcr
rr1D
zvc
tc
axial diffusion
tE4)tvz(
1
20
11
20
etE4)R(Mc
−−
ππ
=also yields but now
+=
D48RvDE 0
0
1
Aris, R (1958), "On the dispersion of a solute in a fluid flowing through a tube", Proc. Royal Soc. A 235, 67-77.
Mass Transfer – Dispersion 13-32
UV Absorbance by particlesin the sample zone as fct of time
MovingsampleZone
PressureGradient
Particle-freecarrier solution
Belongia B.M. and Baygents, J.C. (1997): "Measurements on the Diffusion Coefficient of Colloidal Particles by Taylor-Aris Dispersion"
Belongia B.M., Baygents, J.C. (1997), J. Colloid Interface Sci. 195, 19-31.