Sie sind auf Seite 1von 10

Verbindungen zeigten unter den gewalten Bedingungen entweder keine oder aber eine nur schwach ausgepragte antioxidative Wirkung, so daB praktisch nur wenig Aus- sicht besteht, durch diese Vorstufen der Maillard-Reaktion, wie sie im Rahmen des Trocknungsprozessesin Lebensmit- teln gebildet werden konnen, eine antioxidative Wirkung zum Schutze von Lipiden zu erzielen.

Literatur

1 W Grosch, Z. Lebensmittel-Unters. u. -Forsch. 157,

70 [1975].

2 P. Schieberleu. W Grosch,J. Amer. Oil Chemists' SOC.58, 602

[1981].

3 K. E. Peers et al.,J. Sci. Food Agric. 35, 813 [1984].

4 C. M. Houlihan,J. Amer. Oil Chemists' SOC.61, 1036 [1984]. .5 H. P. KauJinann,Die Ernihrungsindustrie 120 [1973]. 6 A. Seher u. D. Loschner, Fette Seifen Anstrichmittel 88, 1

[19861.

7 R. Yamauchi et al., Agric. Biol. Chem. 46, 2997 [1982].

8 R. S.Farag et al., Can. Inst. Food Sci. Technol.J. 15,174 [1982]. 9 K. Eichner, Prog. Fd. Nutr. Sci. 5, 441 [1981].

10 H. Lingnert, SIK Rapport No. 458, 45 [1979].

11 L. B. Rockland, Analyt. Chem. 32, 1375 [1960].

12

J. E. Hoke, Agr. Food Chem. 1, 928 [1953].

13

K. H

ns, G.Miiller u. H. Pauhen, Liebigs Ann. Chem. 703,202

[196$

l4 W Butte,J. Chromatogr. 261, 261 [1983].

15 E. SchuZte u. K. Weber, Fat Sci. Technol. 91, 181 [1989].

16

W Grosch, Z. Lebensmittel-Unters. u. -Forsch. 160, 371 [1976].

17

S.S.

Chng et al., J. Food Sci. 42, 1102 [1977].

18 J. C. Uhl, Dissertation Universit;it Miinster, 1989.

19

A. Golan-Goldhirsch,Z. Lebensmittel-Unters. u. -Forsch.182,29

[1986].

20

M. Tokati u. M. Morita,Agric. Biol. Chem. 49, 3545 [1985].

D anks agung

Die Untersuchungen wurden im Rahmen der industriellen Gemeinschaftsforschung aus Mitteln des Bundesministers fiir Wirtschaft (BMWi)gefordert. Hierfiir sei an dieser Stelle verbind- lich gedankt.

Eingegangen am 10. Mai 1990.

Contribution to Design and Scale-Up of Packed Columns*

By R Billet ** Institutefor Tharmo- and Fluid Dynamics, Ruhr-Universiq, Bochum, Federal Republic of Germany

The demand for high accuracy in packed column design for countercur- rent thermal separation processes, such as rectification, absorption and desorption, has been stepped up research in scaling of packed column efficiency in response to the growing application of packings for various systems in the chemical and allied industries. The investigations confmed the suitability of relationships for describing the performance, the validity of which was proved with differingsystems and packings. This equations takes due account of geometric,hydrodynamicand system-relatedparameters that govern mass transfer and the fluiddynamics in packed columns.

In response to the growing application of packings for various systems in the chemical and allied industries the demand for high accuracy in packed column design for countercurrent thermal separation processes, such as rec- tification, absorption and desorption, has been stepped up research in scaling of packed columns on the basis of fundamental studies of the fluid dynamics and mass trans- fer. These investigations confirmed the suitability of theo- retically developed relationships, the validity of which was proved with differing systems and packings. The equations take due account of geometric, hydrodynamic and system- related parameters that govern the fluidity behaviour and mass transfer.

Fundamental relationships The correlation, defining the specific column volume v,

* Summary of papers presented at 38'h Canadian Chem. Eng. Confer., Edmonto, AlbertaXanada, 1988; at XIIP Inter-

american Congress of Chem. Eng. PACHEC, Acapulco/Mex- ico, 1988; at Intern. Confer., on Chem. Eng. and Petroleum

Technology, New Delhi, India,

Dr.-Ing. R Billet, Ruhr-University

Bochum, Institute for Thermo- and Fluid Dynamics, Depart- ment of Thermal Separation Processes, P. 0. Box 10 21 48, D-4630 Bochum.

1989.

** Author's address: Prof.

Fat Sci. Technol 92. Jahrgang Nr. 9 1990

Berechnung und Beitrag zur Auslegung von Fiillktirperkolonnen Die Forderung nach hoher Genauigkeit bei der Auslegung von Fullkor- perkolonnen fiir Gegenstrom-Trennprozessewie Rektifiation, Absorption und Desorption fiihrte zu einer verstirkten Forschung auf dem Gebiete der Beschreibungder WirksamkeitdieserApparate bei ihrem steigendenEinsatz fir Prozesse der chemischen und verwandten Industrien.Es werden Berech- nungsmodelle zur Beschreibung der Leistungsfihigkeit von Fiillkorperko- lonnen vorgestellh die gleichermal3en konstruktiven, hydrodynamischen und stomichen Parametem Rechnung tragen, und ihre Gultigkeit und Eignung fiir die praktische Anwendung wird durch Ergebnisse experimen- teller Untersuchungen mit verschiedenen Stoffsystemen bestiitigt.

per unit efficiency and per unit gas throughput according to Eqn. (11,

with nJH being the number of theoretical stages per unit packing height and u, the vapour or gas load, is a criterion for the investment costs of the packings applied in the column. The relationship expressed by Eqn. (2) between nJH and the height equivalent to an overall transfer unit in the gas or vapour phase HTUOv is valid for the separation of mixtures with mass transfer predominantly in the gas or vapour phase. The stripping factor I. is fixed by the vapour/ liquid ratio U Vand the slope myxof the equilibrium curve according to Eqn. (3).

"I;-.-1

H

HTU,,

1IV

x-1

Inh

(2I

(31

In the light of the HTU/NTU model, Eqn. (4) applies for the height HTUov of an overall transfer unit on the gas

side.

I41

Its evaluation requires the knowledge of the volumetric

361

mass transfer coefficients pv . a,,h in gas or vapour phase and PL * a+ in liquid phase which are principally dependent on gas load uv respectively on liquid load uL.Therefore the efficiency of a packed column and consequently its specific column volume are closely related to the fluid dynamics under operating conditions. Based on a theoretical model, recently developed by the author, a description of the fluiddynamics and mass transfer takes into account the great diversity in the geometry, structure and materials for packings in industrial columns, which normally entails differences in the fluiddynamics under operation conditions and thus in the packing

performance. The model correlations were checked against the results of comprehensive and systematic experimental studies performed in the Chair for Thermal Separation Processes in the Ruhr-University and figures taken from the literature. The number of measurements performed in the fluiddynamic tests was 500; in the tests to determine mass transfer on the liquid side, 1050; and in the tests to deter- mine mass transfer on the gas side 1100.The measurements were made on about 60 packings of different designs and dimensions, some of which are shown in Fig. 1 as examples for dumped packings and in Fig. 2 as representatives for regular packings.

Packing main size 50 mm

metal

ceramic

plastic

 

-

CMR No. 1.5

Pall ring

Pall ring

Haket te

No. 1.5 Pall ring Pall ring H a k e t t e N = 19350
No. 1.5 Pall ring Pall ring H a k e t t e N = 19350

N = 19350 a= 120

N = 6400 a= 120

E = 0.978

E

= 0.78

Hiflow ring

Hiflow ring

E = 0.978 E = 0.78 Hiflow ring Hiflow ring N = 5000 a = 97.3
E = 0.978 E = 0.78 Hiflow ring Hiflow ring N = 5000 a = 97.3

N = 5000 a = 97.3

N = 4950 a = 86.7

E = 0.973

E

= 0.815

VSP ring

Intalox saddle

= 0.973 E = 0.815 VSP ring Intalox saddle N = 6700 N = 12400 a=
= 0.973 E = 0.815 VSP ring Intalox saddle N = 6700 N = 12400 a=

N = 6700

N = 12400

a= 110

a= 135

6 = 0.92

E = 0.93

Hiflow ring

DINPAC

6 = 0.92 E = 0.93 Hiflow ring D I N P A C N =
6 = 0.92 E = 0.93 Hiflow ring D I N P A C N =

N

= 6400

N = 29000

a= 112

a= 135

E

= 0.93

E =0.92

~

~~

Norpac ring

ENVI PAC

= 0.93 E =0.92 ~ ~~ Norpac ring E N V I P A C N

N = 7800 a= 104 E = 0.98

ring E N V I P A C N = 7800 a= 104 E = 0.98

N = 9300 a= 120 E = 0.77

= 7800 a= 104 E = 0.98 N = 9300 a= 120 E = 0.77 N

N = 7300 a = 90

E = 0.952

9300 a= 120 E = 0.77 N = 7300 a = 90 E = 0.952 N

N = 6800

a = 98

E = 0.961

Dimensions: N [ 1/m3 I, a [ m2/m3 I, E Im3/m3 1

~

~~~

Fig. 1. Examples for fillings of dumped packed columns, investigated in this work

362

Fat Sci. Technol.

92. Jahrgang

Nr. 9

1990

Mellapak 250Y

Mellapak 250Y a=250, ~=0.96 Montz packing B1 - 300 Montz C1-300 a=300, &=0.90 Sulzer BX”’ a=300,

a=250, ~=0.96

Montz packing B1 - 300

Mellapak 250Y a=250, ~=0.96 Montz packing B1 - 300 Montz C1-300 a=300, &=0.90 Sulzer BX”’ a=300,

Montz C1-300

250Y a=250, ~=0.96 Montz packing B1 - 300 Montz C1-300 a=300, &=0.90 Sulzer BX”’ a=300, ~=0.93

a=300, &=0.90

Sulzer BX”’

B1 - 300 Montz C1-300 a=300, &=0.90 Sulzer BX”’ a=300, ~=0.93 a=500, &=0.90 Ralupak 250Y

a=300, ~=0.93

a=500, &=0.90

Ralupak 250Y

Gempak A2T

~=0.93 a=500, &=0.90 Ralupak 250Y Gempak A2T a=250, ~=0.963 a=202, E = 0.978 Dimensions a [m2

a=250, ~=0.963

&=0.90 Ralupak 250Y Gempak A2T a=250, ~=0.963 a=202, E = 0.978 Dimensions a [m2 /m3 1,

a=202, E = 0.978

Dimensions a [m2/m31, E [m3/m31

-

Fig. 2. Examples for regular packings, investigated in this work

pack-

ings The void space in the bed can be simplified regarded as a network of numerous channels, having the void fraction E and the specific area a of the packing, through which the liquid passes downwards in the form of a film or in lamell flow. The gas flows upwards through the bed, countercur- rent to the liquid. Under stationary conditions, the force of gravity and the shear forces in the liquid film of the density

Modelling

of

countercurrent

flow in

Fat Sci. Technol. 92. Jahrgang XI.$1 1990

pL and the dynamic viscosity qL are assumed to be in equilibrium at a point representing any given thickness s in a coaxial layer within the liquid film, and the fnctional force exerted by the vapour of the density pv acts at the surface of the film. Operating magnitudes are the liquid load uL and the gas velocity u,, which effects also the liquid hold- up h,. The solution of the differential equation describing this flow model gives rise to Eqn. (5)where the index n and the flow factor 4 depend on the flow pattern.

363

dynamic viscosities qLand qv of the phases, and a factor C,

151 that characterizes the geometry and surface area of the packing.

191

Eqn. (5) can be quahtatively described in accordance with the experimental investigations by the diagrams of Fig. 3, which is selfexplanatory and which shows the conditions, under which loading and flooding occur.

Vapour velocily uv

r-

a

s

 

I

I

 

Hold-uph,

r

2

z

--

v

-

h,= f(UY)u,.cmrl

-c- loading line

--llmding line

"1 =f(hl)u"mst

o loading point

a flwding paint

Flooding condilions :

%*o

dh,

-du,

dhl

-

uV.FI

=0 - U1.J

- hl.FI

All that is required for its application is a knowledge of the packing constants C,, which has to be determined by experiment. The index n, is a constant for the system.

Below the phase inversion point, corresponding to (Wq.

it is given by n, = -0.326. Above the phase

inversion point, it has a value of n, = -0.723. If the gas velocity is higher than u., = uv,s, the liquid hold- up increases rapidly above the value at the loading point, i. e. hL > hL, until the flood point is reached, i. e. h, = hL,n. At this point, all the liquid is theoretically forced upwards by the stream of gas in the bed. According to the model the gas velocity at the flood point is given by Eqn.

(pV/p3O.5 < 0.4,

(10)

and the liquid hold-up at the flood point is defined by Eqn. (11), which mathematically satisfies the boundary condi- tions in Fig. 3.

(11I

a8

2

B

>

Hold-UP hl

uv = f(~L'Ul,,,",t

0

0

loading point

floodingpoint

3. Qualitative describing of the fluiddynamic in packed

co umns, operated in countercurrentphase flow of gas or vapour-

Fif

liquid systems

Loading can be defined as the load at which the frictional force exerted by the gas prevents downflow of the inner coaxial layer of liquid in a flow channel, i.e. the load at which the liquid velocity becomes zero. Thus the gas velocity uv,s at the loading point is an important design criterion. It can be derived from the equations in the mathematical model resulting in Eqn. (6)

and in Eqn. (7)

161

(71

which describes the liquid hold-up h,, at the loading point. If the flow of the liquid film is laminar and vertically downwards the following values apply to C, and n: C, = 12" and n = 1/3. Finally Eqn. (8)is derived by combining Eqn. (6) and Eqn. (7).

The flow factor tsin Eqn. (8)depends accordingto Eqn. (9) on the phase ratio We the densities pL and pv and the

These equations are theoretically derived correlations, which can be used for practical estimation of the flooding

conditions.

The application demands a knowledge of the flow factor in, which .is related to the main parameters by Eqn. (12).

Below (Wq . (pv/pJJ.5

S 0.4, nn has an average value of

nR = -0.194; and above this point, nF1= -0.708.

This flow model was the subject of comprehensive experi- mental studies to determine whether is could be applied in general to all kinds of packed columns with countercurrent liquid/* or vapour systems. The correlation between all the measurements and the theoretically derived relation- ships for the loading and flood points was very good. The liquid loads varied between uL= 4.88 and 144m3/m2h; the densities of the gases or vapours, between p, = 0.3 and 1.37kg/m3; the dynamic viscosities, between qv = 7.14 . and 18.19 * 10-6kg/m . s; the liquid densities, between pL = 750 and 1026kg/m3; and the dynamic viscosities between qL = 0.36 . 10-3 and 92.6 10-3 kg/m * s. Detailed figures on the evaluation of the model parameters are dealt with in Tables 1 to 2. They allow application of Eqns. (6) to

(12).

The Reynolds number for film flow in the liquid phase is

defined by Eqn. (13). The evaluation

results has also revealed that if this Rqrnolds number is

higher than 10, Eqn. (14) will apply for the relationship between the index n in Eqn. (7)and the number of packings N, their area a per m3 and their main dimension d. For regular packings n was found to be n = 213.

of the experimental

364

Fat Sci. Technol.

92 Nr.

Jahrgdng

9

1990

Table 1

Constantsfor hold-up, loading and flooding conditions

Dumped Packings

Material

Size d

Ch

CS

C,,,

Metal

50 rnm

51.70

2.725

1.580

Pall ring

Plastic

50 rnm

(i7.95

2.816

1.757

Ceramic

50 mrn

70.29

2.84(j

l.!)l:3

Ralu ring

Plastic

50 mm

2.843

1.812

Metal

50 rnrn

81.71

2.702

1.62ti

Hiflow ring

Plastic

50 rnrn

73.03

2.8!/4

1,871

Ceramic

50 mm

2.81!)

l.6!)4

Hiflow ring; Super

Plastic

50 mm

2.86ti

1.7W

NOR PAC ring

Plastic

50 mm

143.96

2.!).5!)

1.786

Raschig ring Ceramic

50 rnm

2.482

1.574

VSP ring, No.

2

Metal

50 rnm

94.17

2.806

1.M!)

Glitsch, CMR ring

Metal

38 mm

121.M

2.6!)7

1.841

Envipac ring, No. 2

Plastic

60 mm

115.50

2.!)87

1.864

Dinpac ring, No. 1

Plastic

4.5 mm

2.!U!)

1.w1

Intalox saddle

Plastic

50 mrn

2.382

1.548

Intalox saddle; grid

Plastic

50 mm

2.675

1.657

Hackette

Plastic

4.5

mm

59.!) 1

 

Table 2

 

Constantsfor hold-up, loading and flooding conditions

 

Regular Packings

Material

Size

Ch

CS

c I1

Ralu pack YC

Metal

250 mz/rn:{

3.178

2.558

 

B1-100 mz//m.'

97.0.5

3.08!)

1.911

Montz packing

 

Metal

B1-200

rnVm'

52.3Y

3.1 16

2.33!)

 

B1-300 mVm1

34.80

3.098

2.464

 

Plastic

CZ-1100 m2/m+

2.653

1.973

Mellapak Y

Plastic

250 m2/m,'

43.!)7

3.1Fi7

2.464

Gempak A2 T-304

Metal

2012 rn2/m''

(j0.40

2.!)8(i

2.0!)!)

 

tL =

h .I .-

Lr

1

Modelling of mass transfer in packings A relatively new model not only describes the systems investigated quite accurately but is also just as valid for conventional as for modem packing. This model thus allows mass transfer to be described in terms of the packing geometry and the physical magnitudes that influence the gas-liquid system. The relationships derived from it can be applied to all countercurrent-flow columns, regardless whether the beds of packing have been dumped at random or arranged in a geometric pattern. As a rule, the direction in which the two phases flow through the channels in a bed of packing changes continuously. Thus the path followed by the one phase can be imagined to be split up into zones of mass transfer and mixing zones. In other words, after the phases have flowed through a mass transfer zone, the areas of contact are renewed. The average duration of contact of the phases T~ respectively 'cV in the individual zones of mass transfer in an element of length 1,is comparatively short in packed columns. It is governed by the effective void fraction e of the packing and the liquid hold-up h, under operating conditions. At a liquid load uL, tL can be described by Eqn. (15) and if the vapour load is u, tv is given by Eqn. (16).

Fat SCI.Technol.

92.Jahrgang

Nr. 9

1990

"1

Hence it assumed that mass transfer follows the law of instationary diffusion valid for brief periods of contact between the phases and described by the Eqn. (17),formu-

lated by Higbie,

P:&.(<

(17)

i.e. the mass transfer coefficient p as function of the diffusion coefficient D and the theoretical duration t required for renewal of the area of contact.

and the terms of Eqns. (15) and (7)

are substituted, the volumetric mass transfer coefficient pL

in the liquid phase can be derivated with 5h being the zase contact area, Eqn. (18).

If Eqn. (17) is applied

Likewise, the volumetric mass transfer coefficient p,. aphin the gas phase was found to be given by Eqn. (19).

In the derivation of Eqn. (19)was taken into consideration that the volumetric mass transfer coefficient on gas side

365

could be determined most accurately if the gas flow were described by the Reynolds number instead of the gas veloc- ity and the influence of the system by the Schmidt number instead of the diffusion coefficient only, according to the terms in the brackets having the exponents 3/4 and Y3. The magnitudes that affect the phase contact area in the

uv < u,, qL, density

oL and the geometric

range are the liquid load uL, dynamic viscosity

pL and surface tension

area a and the length of the path of contact 1,. If these magnitudes are subjected to a dimensional analysis, Eqn.

aph/a of the phase contact

area aPhto the geometric area a,

(20) will be obtained for ratio

in which the characteristic length of the path of contact 1, were to be described in terms of the hydraulic diameter dh, defined by Eqn. (21).

The model parameters CL and C,for application of Eqns. (18) and (19) are listed in Tables 3 to 4. They were determined from experimental studies carried out with 28 gashquid systems with different mass transfer resistances and directions.

Table 3

Constants for mms transfer in liquid and gas

Dumped Packing

Material

Size

C,

C,

Metal

50 mm

1.1%

0.410

Pall ring

Plastic

50 mm

1.239

0.368

Ceramic

50 mm

1.227

0.415

Plastic

50 mm

1.520

0.303

Ralu ring

Metal

50 mm

1.168

0.408

Hiflow ring

Plastic

50 mm

1.487

0.345

Ceramic

50 mm

1.377

0.37!)

Hiflow

ring, Super

Plastic

50

mm

1.2 l!)

0.341

NOR PAC ring

Plastic

.50 mm

1.080

0.322

Raschig ring

Ceramic

50 mm

1.116

0.210

VSP ring, No. 2

Metal

50 mm

1.222

0.328

Envipac ring, No. 2

Plastic

fi0 mm

1.522

0.296

Table 4

 
 

Constantsfor mas trager in liquid and gas

 

Regular Packings

Material

Size

 

c,

C"

Ralu pack YC

Metal

250 m'?/mi

1.334

0.385

Metal

2.50 mVmi

 

0.983

0.270

 

55 m2/m.+

0.039

Impulse packing

Ceramic

100 mz/m'+

1.317

112 mZ/m'+

1.170

 

250 m2/m"

 

BI-200

m2/d

0.971

0.390

Montz packing

Metal

BI-300 my/m:+

I. 1fi5

0.422

C1-200 m2/m5

1.006

0.412

 

'Iastic

C2-200 mZ/m+

0.739

Conversion of mass transfer When consideringEqn. (4),(19)and (20)for systemswith mass transfer resistance mainly but not exclusively located in gas side the simplified version of Eqn. (22) according to the correlation of Shemood can also be used for describing the mass transfer in terms of HTU,,,

366

HIU,,

= q Re;.Sc:''

(22)

which is interconnected with the equivalent number of stages nJH by Eqn. (2).In this case the Reynolds number is defined by Eqn. (23)

(231

and the Schmidt number in the usual form by Eqn. (24).

scv ' 3-

Ov. Pv

The product of the geometrical area a of the packing and the specific column volume v,, defined by Eqn. (l),

0'. 0.vy

(25)

can now be considered as a measure for the geometric efficiency a' of the packing, which is defined as the geomet- rical area per unit volumetric gas flow uv and per unit efficiency nJH accordingto Eqn. (2).Thus, a dimensionless correlation can be obtained from the equations above mentioned. It can be derived from the main factors that affect the efficiency,i. e. those connected with the system to be separated, the load and the properties of the gas. It assumes the form expressed by Eqn. (26),in which Re, is the mohfied Reynolds number for the gas as defined by Eqn. (23)

for

packing with a relative void fraction E and specific surice a

as expressed by Eqn. (27).

and expressed in terms of the particle diameter d

dp:

6.-

I-€

0

I271

The dimensionless I in Eqn. (26)is a paclung performance rating that connects specific magnitudes for the design, efficiency, load and system in the form of Eqn. (28)

for investigationsof the efficiency in terms of nJH and Eqn. (29) for those in terms of HTU,,

(291

For constant ratio i/Vbetween liquid and gas or vapour, the factor q is a magnitude that is characteristic for the packing. It is practically constant for an individual packing at loads less than 70% of the flooding load point. The exponental index m, however, is more or less independent on the packing and system. This can be shown by evaluat- ing the results of experiments on rectification and absorp- tion systems with resistance to mass transfer predominantly in the vapour or gas phase. When correlating on the basis of

Eqn. (30)

130)

as shown for two rectification systems in Fig. 4 for a dumped packing, consisting of CMR 1.5 Turbo rings, straight correlation lines are received in the double log- arithmic plot. They thus allow the determination of the quantities for q and m in Eqn. (26),on the basis of which the

Fat Sci. Technol.

92.Jahrgang

Nr. 9

1990

s%>

U

v?

3.0

31s 2.0

I

L

W

n

- 0

b

a

v)v1

2

0

.-

v)

1.0

0.8

0.6

0.4

0.3

g

0.2

a1

2

3

4

6

810

Vapor Reynolds number Re;lO'

Fig. 4.Dimensionless correlating of packing efficiencies

efficiencies of the test systems can be converted into those of other systems. Furthermore the investigationsconfirmed that Eqn. (26) can also be applied for absorption systems with resistance mainly on gas side. Examples are dealt with in Fig. 5 indicating that the efficienciesfor vacuum rectifica-

8

50mmHillowRing. PVOF. hydroph. dS:0.ZZm.H*l.47m.N= 7163 l/m3 ChlorobenzenelElhylbenzene p, ~50mmHg.LIV=I

0

50mmHiflow Ring.PP d, d3m. H1137m. N.6997 IIm3

l NH,-AirIti$. p=Ibor,1:293K
L

=

*:-

G

ZIP

&.

Lw

5

2

s

.5

-

0

&

-5

.s

b-

a6

0.4

0.3

a2

0.15

0.1

am

z

3

6

a10

20

Vopor Reyndds number Rev.103

Fig. 5. Dimensionless correlating of packing efficiencies

tion and those for normal pressure absorption can be described by one and the same correlation line, provided the liquid-gas ratio WVis in both operations the same. It also follows from this diagram that only in the range of Reynolds numbers Re, < Re,,, which correspond to gas loads uv <uv, the correlationlines for constantliquid-gas ratio W Vdiverse from those for constantliquid load up For Reynolds numbers, correspondingto gas loads uv, < uv < uv, F1 however, i. e. in the loading zone, both operating conditions can be correlated by more or less one common line. The knowledge of the interrelationship expressed by Eqn. (30) thus provides also the basis for scale up the evaluating of the efficiencies obtained in a pilot plant to a commercial scale packed column. For this purpose a dia- gram, according to the qualitative presentation of Fig. 6 can be developed for the individual packing considered. The

Fat Sci. Technol. 92. Jahrgang Nr. 9 1990

(?
N

95u

Correlollon line of

 

pilot plont lest system

c i

.

I

\

(

;

\

plot

sc5

 

I

 

Gos Reynolds number Re,

Fig. 6. Qualitative diagram for correlating the dimensionless efficiency

dotted lines indicate qualitatively the conditions of the operating point where loading of the packings is to be expected. It allows the estimation of the packing efficiency for the separation of a system, the gas viscosity qv of which

is known and the gas load ,u corresponding to

number Rev to be applied is determined. With the knowl- edge of the dependency of Fig. 6 expressed by Eqn. (30), which must be found by experiment in a pilot plant with'a proper defined test system, the efficiency can thus be determined with the equations indicated in the diagram.

the Reynolds

Conversion of pressure drop Experimental data for the pressure drop per unit of efficiency were correlated in the literature in the form of

Eqn. (31)

which takes into account the interrelationship of Eqn. (22) and that of Eqn. (24). It represents a modified specific pressure drop Ap per transfer unit NTUov for a system, the resistance to mass transfer of which is predominantly in the gas or vapour phase, i. e. the Schmidt number Sc, represents the only means of reflecting the properties of the system. Since the expression of Eqn. (31) is independent of the system,it is ideally suited for comparing column packing. It was thus revealed that a common curve can be drawn to represent the results obtained on various mixtures rectified in test runs under total reflux or under constant liquid- vapour ratio WVand those obtained in the absorption, for example of ammonia from air under constant liquid load uL by means of water. However this corrqlation does not apply generally under both conditions, i. e. WV= const. and uL= const., and must be treated with reserve concerning the exponent m. In this case a modified expression is taken for the pressure drop

per unit of efficiency.It is described by

Ep (32) and allows

for the difference in the phase ratios WVin the various separation processes.

(321

In

rectification experiments under total reflux, dtis equal

to

1; in rectification processes with a finite reflux ratio r, WY

367

is given by Eqn. (33) in the enrichment section,

_--- ir

Y r+l

(331

and in absorption experiments with constant liquid flow rate uLand variable gas capacity factors F, or vice versa, by

Eqn. (34).

Thus the modified pressure drop per unit efficiency in absorption experiments in which the liquid flow rate uLis constant but the gas capacity factor F, is variable can be correlated by Eqn. (35).

(351

The curves reproduced in Fig. 7 confirm the expediency of this method. They were plotted from the results of experi- ments on rectification with total reflux, i. e. UV= 1, and on absorption with a constant liquid flow rate uI.= 10 m3/m2h and a variable gas capacity factor F,

0.6 0.8 1

capacity factor F,

2

34

[m''2.s-!kg"z]

Fig. 7. Modified specific pressure drop as function of load in rectification and absorption test

Hence the results of rectification and absorption experi- ments cannot be generally correlated by means of a com- mon function in form of Eqn. (31) valid for both, unless Eqn. (36)

ii=m.n

(361

this case the correlations described by Eqns. (31),(32) and (35) are identical. If the conditions stated above does not apply, as shown in Fig. 7, Eqns. (32)and (35)must apply for correlating the results of rectification respectively absorp- tion experiments, which thus can be applied for scaling up. In Fig. 7 values measured in rectification are compared with fieres for absorption calculated from Eqn. (32)for the case WV= 1. In this example there is a very good agree- ment between the values for the system-independent mod- ified specific pressure drop. The specificpressure drop Ap/NTU,, is interconnected with the pressure drop Ap/H per unit height of packing by Eqn. (37).

HTUov =- AplNTUo, AplH

137)

In line with a recent proposal the pressure drop Ap/H of a wetted packing can be described in analogy to that of a dry packing by Eqn. (38)

138)

in the derivation of which the hydraulic diameter dh of the gas flow channels of the wetted packing with the effective void fraction (E - hJ is expressed by Eqn. (39)

dh'4.TE-hL

1391

Equ. (38) becomes identical with the well known correla- tion for a dry packing, i. e. for which h, = 0. The wall factor K takes into account the influence of the column diameter d, according to Eqn. (40).

From Eqns. (38) to (40), whereby F,

vapour capacity factor according to Eqn. (41)

defines the gas or

Fv=u,.q%

(41)

can now be derived the correlation for the related pressure drop per unit height Ap/H of the wetted packing resulting in Eqn. (42).

It contains the resistance coefficient sLof the gas flow in the wetted packing channels which is related to the flow of liquid according to the Eqns. (43) and (44), which were found by testing of various packings, such as in Figs. 1 and 2. Values for the packing constant C, are listed together with the characteristic data on various types of packing in Tables 5 to 6.

is

satisfied. However this is mathematically impossible until

Thus the Eqns. (32) and (42) allow conversion and scaling

n

is negligibily small, i. e. zero in the limiting case; only in

of pressure drops, but also of efficiencies, because in

368

Fat Sci. Techno].

92.Jdhrgang

Nr. 9

1990

Table 5

Geometrical data and pressure drop constantfor dumped packmgs

Dumped Packmgs

Material

Size d

N

a

E

CP

 

l/m'

mL/m?

m{/m'

 

Metal

50 mm

fi242

112.6

0.95 I

0.7fi3

Pall ring

Plastic

50

mm

fi7fi.i

111.1

0.!1 I!)

O.fi98

Ceramic

50

mm

(ill15

116.5

0.783

O.ti62

Ralu ring

Plastic

50 mm

5770

!I52

0.1138

0.468

Metal

50

mm

5000

!12.3

0.977

0.42 I

Hiflow ring

Plastic

50

mm

ti815

117.1

0.!125

0.327

Ceramic

50

mm

5 I20

89.7

0.809

0.538

Hiflow ring, Super

Plastic

50

mm

6050

82.0

().!I42

0.4 14

NOR PAC ring

Plastic

50

mm

7330

86.8

0.947

0.350

VSP ring, No. 2

Metal

50

mm

784 1

104.6

0.!)80

0.773

Envipac ring, No. 2

Plastic

ti0 mm

6800

!)8.4

().!)ti

1

().338

Top-pak

Alu

50

mm

(i!J47

1O6.fi

0.956

0.604

Bialecki ring

Metal

50

mm

6278

121.0

0.!M

0.7I!)

Italox saddle

Plastic

50

mm

Rtiiiti

122.1

0.!)08

0.758

Ceramic

50

mm

8882

114.6

0.7fi1

0.747

Hiflow saddle

Plastic

50

mm

9939

86.4

0.938

0.454

Hackette

Plastic

45 mm

12252

133.4

0.!)31

0.39!)

 

Table fj

 

Geometrical data and pressure drop constantfor regularpacking.

 

Regular Packings

Material

Size

a

E

CP

 

m2/m.?

m?/m"

Ralu pack

Metal

YC-250

250.0

0.!)45

0.I!)1

Impulse packing

-

Ceramic

I00

Y(i.7

0.828

0.417

Metal

B1-200

200.0

0.!)7!1

0.355

Montz packing

B1-300

300.0

0.930

0.295

Plastic

c 1-200

200.0

0.!I5 4

(1.453

 

c2-2 00

 

200.0

0.!100

0.48 1

connection with Eqn. (37) they form Eqn. (45), which is valid for the main loading range.

Final Considerations It is general knowledge that columns with randomly oriented or geometrically arranged beds of packing have much lower pressure drops per theoretical stage than plate columns and thus permit lower reflux ratios, particularly in vacuum rectification.Thus column internals with low pres- sure drops per theoretical stage are an advantage in realiz- ing all energy-saving measures in thermal separation pro- cesses ranging from product vapour compression to multiple-column operation. Furthermore in significant industrial applications, such as the vacuum rectification of ethylbenzene/styrene mixtures or the recovery of fatty acids, avoidance of thermal decomposition is an important factor in favour of columns with low pressure drops. Therefore most of the packings in common use today have been the subjects of systematic studies. The results obtained permit a comprehensive review of the main relationships that exist between the degree of mass transfer and the column hydrodynamics, which themselves depend on the operating conditions, the system, and the packing's geome- try and texture. Based on the numerous results a new correlating of these relationships was developed for process

Fat Sci. Iechnol. 92. Jahrgang Nr. 9 lYY0

calculation and design, but the results of laboratory or pilot plant experiments can be adopted direct for scale-up, if resistance to mass transfer is predominantly in gas or vapour phase, e. g. in most rectification and many absorp- tion processes. As far as scaling effect of height and diameter of packed columns is concerned the common literature gives quite useful1 information. In this connection, it is worth noting that latticework packing permits hardly any maldistribu-

tion, as was determined some time ago

in systematic studies

on Nor Pac packing in columns of 0.3 to 1.5 diameter. It is also worth mentioning, that the critical liquid distribution density, above which no further improvement can be expected in efficiency of a bed of packing, is 400 liquid distribution points per square meter of column cross- sectional area. All the liquid distribution points of the feed

and top distributor must function uniformly to ensure utmost accuracy and thus a constant W Vin all coaxial flow paths in order to maintain the column efficiency. If the distributor causes an initial liquid maldistribution of M [O/o], the packed column efficiency may be reduced by AE, the amount of which depends on the ratio d,/d between the column diameter d, and the main size d of the individ- ual-packing element and of the liquid-vapour or -gas ration L/y Fig. 8.

This work is mainly based on own studies and experi- mental research carried out in the laboratories of the author. The list of references therefore contains only some of his own publications.

50

c

E u40

a

g

u

5

ga

P

za a

0

10

is

is

22

m

x1

34

38

P

6

M

Ratio Column OiomeierPockingElement Size dS/d

Fig. 8. Decrease of packed column efficienc

caused by mal-

distribution according to Huber and Ifriltbrunner

Greek

Letters

E

m3/m3

Void fraction

tl

kg/ms

Dynamic viscosity

a

Stripping factor = my,.

E

kg/kmol

Molar mass Resistance coefficient

P

kg/m3

Density

Subscripts

Fl

Flooding point

L

Liquid

S

Loading point

V

Vapor or gas

W

Water

hi

Nomenclature

a Geometrical area of packing, mVm3

a' Geometrical area per unit volumetric vapor or gas flow rate per theoretical stage, m2dm3

aph Interfacial area for mass transfer, mVm3 Constant

C

Cm Packing factor for flooding capacity

C,

CL Packing constant for liquid phase mass transfer

Cp Packing factor for pressure drop

Cs Packing factor for loading capacity

C, Packing constant for vapor phase mass transfer

d

d,

dp Ciaracteristic packing dimension, m

d, Column diameter, m

D, Diffusion coefficient for vapor, mVs

DL Diffusion coefficient for liquid, mVs

Fr Froude number

Literature

Packing constant for holdup

1

R Billet, Fluiddynamisches Verhalten und Wirksamkeit von Gegenstromapparaten fiir Gas-Fliissig-Systeme bei fliissig- keitsseitigem Stoffiibergangswiderstand,Contributionin Fest- schrift, Lennings, Fakultiit fiirMaschinenbau der Ruhr-Univer- sit5t Bochum, 1983.

2

R Billet, Operation of Thermal Separation Plants under the Aspect of Environmental Protection, Congress Book 4th Intern. Symp. on Environmental Techniques Development, Dong-A University, Pusa.n/Korea, pp. 163-182, 1984.

3

R Billet, Modelling of Fluiddynamics in Packed Columns, I. Chem. E. Symp. Ser. No. 104, A171 [1987].

R Bill&, Performance of Low Pressure Drop Packings, Chem. Eng. Comm. 54, 93 [1987]. 5 R. Billet, Relationship between Residence Time, Fluid- dynamics and Efficiency in Countercurrent Flow Equipment, Chem. Eng. Technol. 11, 139 [1988]. 6 R Billet, J. H. Kim and A. Qaghouri, High Performance Absorption Columns, Proceedings: The First Korea Japan Symposiumon SeparationTechnology, KyongjdKorea, 1987.

4

Mackowiak, Neuartige Fiillkorper aus

Kunststoffen fiir thermische Stofftrennverfahren, Chem.

Techn. 5, 219 [1980].

R Billet and M.Schultex Determination of Liauid Hold-uD in Gas-Li uid Two-Phase 'Countercurrent Mas: Transfer hl-

umns, ?.Chem. E. S

R Billet and M. Schg Capacity Studies of Gas-Liquid Two- Phase Countercurrent-FlowColumns, I. Chem. E. Symp. Ser.

R.

Billet

u.

J.

p. Ser. No. 104, A159 [1987].

Nominal packing size, mm or m

H draulic diameter of a packing, m

F,

F-factor for vapor or gas = qp,1/2, m-%-lkg1/2

Packed height, m Height of an overall transfer unit, m

g Gravitational constant, 9.8m/s2

hL Liquid holdup, m3/rn3 packing volume

H

HTU,,

I

Dimensionlessseparation magnitude

K

Wall effect factor

L

Liquid throughput, kg moles/s or kg/s

7

my.

Slope of equilibrium curve

m

Exponent

iii

Exponent

8

n

Exponent

n,

Number of theoretical stages

N

Number of packings per unit volume, l/m3

9

NTU,,

Number of overall transfer units in vapor or gas

pT

r

Pressure at top of packed bed, mbar

No. 104, B255 [1987].

Reflux ratio

I0 R Billet u. M. Schultes,Vorausberechnungdes Stoffaustausches

Re

Reynolds number

bei Gas-Fliissigsystemen mit gasseitigem Stoffiibergangs-

Sc

Schmidt number

widerstand, Paper presented at the Symposium of T.U.

uL

Liquid load, mVm2s or rnVrn2h

Wroclaw/Poland, 1988.

uv

Superficial vapor velocity, m/s

11

R BiUet and M. Schultes, Mass Transfer in Gas/Liquid Systems

\I,

Specific column volume, s

with resistancein the Liquid Phase, Paper presented at AIChE

V

Vapor throughput, kg moles/s or kg/s

Spring National Meeting, New Orleans, MSA, 1988.

W

Wettingfunction

12

R Billet and M. Schultcs, Modelling pressure drop in packed

WG

Water column

columns, Paper presented at AIChE SpringNational Meeting,

X

Mole fraction in liquid

Houston, TWSA, 1989.

Y

Mole fraction in vapor

Z

Number of liquid pour points

Received 14th December 1989.

3 70

Fat Sci. Technol.

92.Jahrgang

Nr. 9

1990