Advanced technical tools for the solution of high capacity adsorption separation

Adsorption and its Applications in Industry and Environmental Protection Studies in Surface Science and Catalysis, Vol. 120 A. Dabrowski (Editor) 9 1998 Elsevier Science B.V. All rights reserved.

275

A d v a n c e d t e c h n i c a l tools for t h e s o l u t i o n of h i g h c a p a c i t y a d s o r p t i o n separation G. Horvfith a and M. Suzuki b aDepartment of Chemical Engineering, University of Veszpr~m, P.O.Box 158, 8201 Veszpr~m, Hungary bInstitute of Industrial Science, University of Tokyo, 7-221 Roppongi, Minato-ku, Tokyo 106, Japan

1. INTRODUCTION The surface and transport properties of adsorbents have been widely used in different technological fields. The development of compact and large capacity separation processes is one of the most urgent problems in the field of environmental protection. The oxygen production, the purification of stack gases and other polluting off gases demand at least partly new technical solutions. Among the new geometrical constructions new methods are taken into consideration, like ultra rapid pressure swing adsorption (URPSA), piston driven PSA, parallel passage adsorbers and so on. The main disadvantages of the traditional packed bed columns are the relatively high pressure drop and the thermal resistance of adsorbent materials leading to a significant power cost. In order to realise a high capacity adsorptive separator, some fundamental engineering studies have been done during the past years. 2.

HYDRODYNAMIC ASPECTS

The large-scale "gas-separation" processes may be conveniently divided into two classes, the cryogenic separation systems and the adsorption separation processes. In this chapter we are dealing with adsorption processes only. Figure 1 shows schematically the pressure swing, thermal swing and combined adsorption operations. In order to estimate the practical adsorption capacities necessary to know the equilibrium relations. The equilibrium relation depending on temperatures between the specific adsorbed amount a (g/g) and the pressure p (bar) is plotted with continous line. In industrial operation this capacity can be as maximum considered and naturally it cannot be fully utilized. All of

276 separation technologies demands energy introduction. From this point of view the form of separation work classifies the technologies. Thermal swing solution in ideal case (ideal Tswing) shows a difference between isotherms of T1 and T2 at constant pressure. In the reality the ideal differece between the two equilibrium states cannot be reached because the adsorbent warms during the adsorption step and it cools during the desorption step. Thus the dotted isotherms are the virtual equilibrium lines (T,' .and T~ ). The mass transfer resistances have to be also taken into consideration, but here we do not deal with them.

a

[g/g] ideal T s w i n . . .

.... ads.

.

,,~ "k~"

.

.

.

.

.

.

~ , "4"

~ -

T'I

real T s w i n g ..~?, ....

des.

__

- -

_

_

_

T2>TI

T'2

/

J

Pz

', P'2

P, P',

P [bar]

Figure 1. Schematic adsorption swing operations.

As for the ideal pressure swing (ideal Pswing) adsorption the situation is similar. Usually the main reasons of deviation are the pressure drop in the colums and the adsorbent warming, so the dotted isoterms symbolizes the differences from the ideal operation. Having based on the same logic, the ideal and real combined adsorption processes face to the same limitations. One of the most important problem of large apparatuses is the solution of the gas transport. The produced gas amount could hardly reach the 1 metric ton/hour. This capacity usually is too small for environmental applications.

277 2.1.

Method

of Ergun-equation

The hydrodynamic of flow through packed beds has been widely investigated, except the closed end column, which is i m p o r t a n t from the point of adsorption view. To evaluate the pressurization time from experiments the model of isothermal gas flow in porous materials has been used. The model is based on the following assumptions: [1,4]: the gas t e m p e r a t u r e is constant, - radial gradients in the pressure and in the macroscopic velocity are negligible (the plug flow condition), - the inertia forces are negligible, - the friction forces obey the Ergun equation, - an ideal gas is assumed, - the gravitational forces are negligible, - the dynamic viscosity does not depend on pressure. In the range of our experimental data the inertia forces and the t e m p e r a t u r e effects are negligible usually. The mathematical model consists of a mass balance

~C

~--+

:-/~uc}

/)x"

"

: o

(1)

and a m o m e n t u m balance which contains only two terms expressing pressure and friction forces. -

gdp

u2

(2)

For the friction coefficient )~, the Ergun equation has been used

Re

+ B

(3a)

where

dp~;pu

R e = ~ (1 - ~)~

(3b)

Ergun [1] recommends A equal to 150 and B equal to 1.75 based on a correlation of experimental data. The state equation for an ideal gas connects the molar concentration c and pressure p, c = p/RT. The density of-a gas is the product of the molar concentration, c, and the molecular weight, p = Mc. Instead of the velocity, u, we introduced the mass flux density in the form M = uc.. With the constant t e m p e r a t u r e assumption, the Eqs. (1) and (2) can be r e a r r a n g e d to the form

278

1 ap+ 0M, = 0

RT & 0p

=

(4)

Ox

_AI.tRT

k, adp)

M p-B

1- ~ M2 MRT ~dp p

(5)

There are two dependent variables, p and M, and two independent variables, x and t. Under the assumptions mentioned above, all other quantities are constant. In addition to Eqs. (4) and (5), the two initial conditions p(x,0)=pi

(6)

M(x,O)=O

(7)

and two boundary conditions are required. The model describes the pressurization of a fixed bed closed at one end. The pressure at the other end is suddenly fixed at a different pressure. At the closed end of the column, the boundary condition M(x=H,t)=0

t>0

(8)

is applied, and at the open end p(x=0,t)=pF

t>0

(9)

Equations (4) and (5) were transformed into dimensionless form by means of the following dimensionless variables * P-PI p =~ PF - P l

(10)

M*= M RT(1-c] 2 p--~, ~:dp------) HAIa

(11)

x

rl = - -

(12)

z = t ( e d . ) 2 P, \ 1 - ~J H2AIa

(13)

H

and parameters

279 Ap* = PF~ -- P l Pl B

(14)

p2M

(adp/3

5 =-~\1-

s)

Hp2RT

(15)

The transformed Eqs.(4), (5) become: Ap

*

0:)

Ap* ~o~* 0rl

aM + ~ =0

= -

M* ( 1 + 6M* ) 1+ Ap*p*

(16)

(17)

with the initial conditions P* (n, x=0)=0

(18/a)

M* (q,x=0)=0

(18/b)

and the boundary conditions M* (n=l, z)=0

(19)

p* (vl=0,x)=l

(20)

Equations (16) to (20) have been solved numerically. The Figure 2 shows a traditional industrial adsorber.This type of adsorber has a capacity as high as 10 - 50 m3/hr depending on the raw materials. The calculated pressure drop of the former traditional adsorber in a steady state regime can be see on the Figure 3. The parameters are the temperature (T) and the volumetric standard gas velocity (Y). A master sheet for the determination of pressurisation time and dependence of the pressurisation time on the complex parameter 5 can be seen in Figure 4 [20]. For 5 approaching 0, the flow approaches Darcy's-region, while for 5 >> 1 it is in the high velocity range [2-5]. From the results of these studies it can be concluded that the Ergun type equation can be used for the simulation of slower pressurisation processes (i.e. in case of small particles) provided the constants of the equation are determined preliminarily. The very high velocity range like ultra rapid PSA should be studied with more detail.

280

/

/

508x6 -i

560x3

li I1

!

Figure 2. A traditional industrial adsorber.

281

l

35000 30000 25000

I~

20000

dP(Pa)

~

15000

,~

~,/~

10000

! I I// 200 sm31h 75sm3/h 50 sm3/h

~

30 sm3/h 20 sm3/h

5000

10 sm3/h 50 100 200 300 400 500 600 700 800 900

T (~

Figure 3. Pressure drop, temperature dependency in steady state. fl_

2.0.

5 _o =

~Rp=0.96 1.5'

~:Rp= 1.03

1.0-

~ap=9 2 I / 8 " ~

0.5-0.0 -0.5. -1.0,

__,0

i ii:'//

~20

-1.5 -2.0 -6

I

I

I

I

I

I

I

I

I

-5

-4

-3

-2

-1

0

+1

+2

+3

+4

log 5

Figure 4. The pressurization time vs. 8 for different inlet pressure differences. (Reprinted with permission from ref. [20]).

y

282 2.2. T h e g e o m e t r y ( w i t h the permission of L'AIR L I Q U I D E ) Several industrial firms began to construct adsorbers mainly for air separation. So numerous alternatives can be found on the market, but the specific energy in the case of classical column constructions used to be high (0,92,1 kWh/Nm 3 O2). Therefore the low temperature separation as for the energetic point of wiev is more economic. The L'AIR LIQUIDE opened a new direction some years ago, and developed the so-called COMPACT VSA family. They could decrease the energy demand into the 0.3-0.45 kWh/Nm302 intervall . The gas production lies in a range of 5 metric tons/day to 160 metric tons/day.

i

I

I .'." I v ".:.: I I ": I

I I I

!', 1 ALUMINA BED

, :::,

I

I I

2 ZEOLITE BED

I

'.'.: ":

I I

i ":':

I I

".:: "'"

I

.... :::,

I I

I

I C.:.: I i ,'.:.:.: I

I I

b OXYGEN PIPE

I I

i I I

a AIR/WASTE GAS PIPE

,

i

I

I :.':, I '~':':'

I I

I I

a

I I

, I':':'." I ~ :.:.:.: i , ::::..:: '

2

I,,,

I I

/

, ,

,

,

I I I

I I I

b

Figure 5. Radial adsorber of L'AIR LIQUIDE.

Among others the secret of this sucsess is the radial adsorber (Figure 5). The air enters into the under part of column, flows through an alumina adsorber, then passes radially the zeolite bed. Evacuation is used for regeneration. Supposing that the adsorber is adiabatic, the temperature change reaches an order of magnitude 50~ Such types of constructions open new possibilities in the fled of technological applications.

283

3. FUNDAMENTALLY NEW CONSTRUCTIONS 3.1. Ultra rapid PSA The development of processes is one of the most urgent requirements in such fields as carbon dioxide recovery from stack gases and oxygen enrichment for more efficient engine performance of automobiles. The pressure swing adsorption (PSA) process is a good candidate for achieving these purposes, but the production capacity of PSA must be improved for this purpose. This can be done by the development of more selective adsorbents, operation at low temperature where more adsorption capacity is expected, use of short cycle times where more throughput ratio is expected, etc. Among these approaches, the most significant improvement of capacity is expected to be achieved by rapid cycle operation. However, short cycle operation by conventional PSA processes is difficult bacause the lifetime of the ordinary valves applied is limited, affecting the continuous operation of the process. The piston-driven ultra rapid pressure swing adsorption (URPSA) equipment was developed and oxygen enrichment from air was examined as an example [6-19]. The adsorbent bed is directly connected to the cylinder where a piston moves at high frequency. Thus pressurization and depressurization in the bed are driven by mechanical piston motion, which can achieve far more rapid cycles compared with the conventional pressure swing operation using valves. The cycle time is usually on the order of seconds or sub seconds. Oxygen enrichment from air up to about 60% or higher of oxygen concentration was achieved by smallscale equipment using zeolite 5A with a oxygen production capacity of 100 Nm 3product gas/ma-zeolite/h. This specific value is about ten times larger than those of commercialized PSAs for the same purpose. A method to realize rapid operation of PSA, a piston-cylinder module has been applied in a conventional reciprocal engine equipment as the pressurizing and depressurizing mechanism combined with an adsorption unit. Using this equipment, ultra-rapid operation was achieved and oxygen enrichment from air was experimentally investigated. A simplified numerical model has been proposed to describe ultra-rapid pressure swing adsorption (URPSA) operation [21]. Having used this model, the basic characteristics of this operation can be characterised. 3.1.1. Basic steps of piston-driven ultra rapid PSA The URPSA equipment (Figure 6) consists of a piston, a cylinder, valves and an adsorption bed. The piston is moved by an arm connected to a rotating disc driven by an electric motor. The basic operational steps are suction, adsorption and production, desorption, and exhaust step, as shown in Figure 7. The operation in each step is as follows.

284

Nv Zeolite Packed Bed

Dehumidifier

Compressor

Product

~ ...........

(

Gasbag

1

i

] @D

-~ C~gstater

lAir

lyli!der

Valve timing controler

D'<]Electromagnetic valve N.V. !><3 Needle valve

@

Pressure gauge

Figure 6. Schematic diagram of experimental apparatus. (Reprinted with permission from ref. [21].

1)

B

2)

3) /~I'

f /

4)

5)

i start c::> A

suction

adsorption

air 9 ~ suction 9 valve B

0 2 rich gas " product valve

desorption

exhaust

~,~ N2 9 rich gas C "exhaust valve

zeolite 9 bed

Figure 7. Basic steps of piston-driven ultra rapid pressure swing adsoption. (Reprinted with permission from ref. [21]).

285 Step 1: Raw gas is introduced into the cylinder while the piston goes down from the top position to the bottom position. Step 2: While the piston goes up, gas in the cylinder is compressed and introduced into the adsorbent bed, where adsorptive separation takes place and a less-adsorbable component is enriched in the product gas which is released from the top of the adsorbent bed. Step 3: The piston goes down with all the valves closed. During this step, the depressurization and evacuation of the adsorbent bed take place. Desorption occurs and the cylinder is filled with the desorbed gas. Step 4: In this step, the piston goes up again and the desorbed gas in the cylinder is removed as exhaust through the valve at the cylinder wall. One adsorption-desorption cycle is completed with two up-and-down strokes of the piston. The cycle time in this sequential operation is controlled by the rotation speed of an electric motor and can be adjusted to seconds or sub seconds. Oxygen enrichment by using zeolite as an adsorbent was tried as an example of application.

3.1.2. Apparatus Details of the adsorption unit are shown in Figure 8. Product gas

T Pressure gauge Thermocouples Filter Zeolite

Rubber Filter

Cooling water

Air

Pressure gauge

Piston

Figure 8. Details of adsorption unit. (Reprinted with permission from ref. [21]).

286 The packed bed of zeolite in this case was 35 mm in height and 60 mm in diameter. The zeolite bed was fixed at the top and bottom by a sheet of filter paper (Advantech Toyo Filter Paper, No. SA) and a stainless steel screen of 400 mesh size. Three copper-constantan thermocouples of 1.0 mm in sheath diameter were inserted into the center of the bed at the inlet, middle and outlet sections. The applied adsorbent was zeolite 5A (Union Carbide Corporation) which was crushed and sieved into 48 to 80, 80 to 150 and 200 to 325 mesh size groups. As a preliminary t r e a t m e n t the zeolite was heated at 653 K for over 3 hours under evacuated conditions. The pre-treated zeolite was carefully packed in the bed to avoid humidity uptake. The rotation speed of the motor was set to 80, 120, 160, 200 and 240 rpm corresponding, respectively, to cycle times of 1.5,1.0, 0.75, 0.6 and 0.5 seconds. The oxygen concentration in the product gas collected in the gas bag was determined by an oxygen analyzer using zirconia (Toray Corporation, LC-100). The flow rates of the feed air and the product gas were measured with an integration type flowmeter (Shinagawa Manufacturing Co., Dp-2A-1). The temperature of the cooling water was keep constant at 278 K.

3.1.3. A simplified m a t h e m a t i c a l model The assumptions employed in the modeling of URPSA were as follows (Kowler and Kadlec [11], Jones and Keller [8, 9] and Guan and Ye [6]): all gases are ideal, air is a mixture of nitrogen and oxygen only, and each gas is independent in terms of the adsorption equilibrium and rate, the adsorption equilibrium is linear, the adsorption rate is expressed by the linear driving force (LDF) approximation with a correction factor, due to rapid cyclic adsorption and desorption (Nakao and Suzuki, [13]), the flow in the packed bed is the plug flow.J6]. The shape of the zeolite particle is spherical, the pressure drop in the packed bed is given by Ergun-equation, the temperature of the entire system is constant at 303 K. The model is a box model consisting of the cylinder, the zeolite bed divided into 10 compartments and the head space of the packed bed. The mass balance of each box is written as follows. The mass balance in the cylinder is, dPcyI dVcyI dN cyl RT dt Vcyi + Pcyl d-----~= d-----~ where

(21)

287 dN cyl

dt

= Ffeed + Fin + Fexhaust

Vcz! = Sczl + x pi Xpi =

x pi max - x pi min {cos(wt + n)+ 1} 2

(22) (23) (24)

Pcyl,Vcyl, R and T present the total pressure in the cylinder, the volume, a gas constant and the temperature, respectively. Ncyl is the amount of gas present in the cylinder Ffeed, Fin and Fexhaust the flows of the feed gas, to the packed bed and of the exhaust gas, respectively. Say1 and Xpi represent the cross sectional area of the piston and the length between the piston head and the ceiling of the cylinder. Xpi max and Xpi min are maximum and minimum of Xpi. w is the angular velocity. The mass balance of component i in the j-th box of the zeolite bed is given as: dNbedi,j dt

= (Fbed)i j-I + (Fq)i, + (vbed)i, ' J J

(25)

Nbed i,j, is the amount of gas present in the bed. (Fq)i,j and (Fbed)i,j a r e the fluxes

due to adsorption and desorption, and to the adjacent box, respectively. Adsorption rate and equilibrium relation are written as dqi,j (Pbed)i,j- (P;ed)i,j RT 3' dt =( KFa v )i,j

(KFav)i, j =

qi,j

~i,j(Dp)i,j(1- ~:) R2 P (Pbed)i,j aT

(26)

(27)

(28)

(29)

where KFav, Dp, ~, Qst and (~0) are the overall mass transfer coefficient, the pore diffusion coefficient, bed void fraction, the heat of adsorption and pre-exponent

288 constant, respectively.

(') Pbed i,j is the pressure

in equilibrium with the present

amount adsorbed, qi,,j, ~2i,,j is a coefficient depending o n

(l:c)i, j (Nakao and

Suzuki, [13]) given by

(1;c)i,j --

Dp)i,jtc 2 Rp

(30)

where t c is the adsorption time and Rp is the radius of adsorbent particle. Dp is evaluated as follows. ~p (Dp)i,j = -k2-(D0)i,J

(31)

1

1

1

( D0)i,j

(D m)i,j

( DK )i

2 a 18RT (DK) i =-~r Mi

(32)

(33)

where Dm and DK are the molecular diffusion coefficient and the Knudsen diffusion coefficient, respectively. ~ is the particle void fraction, k 2 is the tortuosity factor, ra is a the macropore radius and Mi is the molecular weight of the i-th component. The mass balance in the head space is given as dNdtup = (Fbed)i, j max + Fproduct

(34)

where Nup is the amount of gas present in the head space over the zeolite bed and Fproduct is the flow of the product gas. The calculation method is as follows. - The pressure change in the cylinder by the piston motion at time, t, is calculated. - The amount of gas flowing from the cylinder into the first box of the zeolite bed is obtained. - The pressure in the first box is calculated. - The pressure decrease by adsorption is calculated.

289 - The flow to the next box, the second box, is evaluated according to pressure gradient between the first box and second box. - The same procedure can be repeated for the following boxes. Pressure changes at the inlet and outlet of the zeolite bed with cycle time are shown in Figures 9-11. for those cases where particles of 48 - 80, 80 - 150 and 200 - 325 mesh size were employed. The first quarter of the cycle time is the suction step when the piston descends from the upper dead point. The solid and dotted lines respectively represent the pressure swings at the inlet and outlet of the zeolite bed. These were calculated using the model described above. When smaller particles were used, smaller pressure changes at the outlet of the zeolite bed and larger at the inlet were found. These are caused by the different pressure wave propagations due to different conductance in the beds of different particle sizes. This empirical fact suggests that the separation performance depends greatly on particle size because the system pressure changes effective for adsorption and desorption are controlled by the pressure drop performances in the packing bed. Having compared with Rapid PSA (Turnock and Kadlec [19]), larger pressure swing over and under the atmospheric pressure is achieved in a shorter cycle time.

2.0

1.5 E 1.0 r

r~

0.5

pressure

exp.

inlet outlet ,0

i

0.0

i

0.2

i

i

0.4

|

i

0.6

i

i

0.8

i

1.0

Time/Cycle Time [-]

Figure 9. Pressure swing in adsorption bed ( dp:# 48-80, cycle time: 1.0 s.) (Reprinted with permission from ref. [21]).

calc.

290

2.0

1.5 E ..) L..

~, ~

. \ .... .

._/

1.0

if) L.. n

0.5

pressure inlet outlet

.0

,

,

0.0

,

,

0.2

,

,

0.4

,

exp.

calc.

9

,

0.6

0.8

1.0

Time/Cycle Time [-]

Figure 10. Pressure swing in adsorption bed ( dp" # 80-150, cycle time: 1.0 s.) (Reprinted with permission from ref. [21]). 2.0

1.5

E .m. 1.0

....... ~

if) L.

a.. 0.5

pressure inlet outlet

.0

i

0.0

o

0.2

o

u

0.4

1

a

0.6

o

o

0.8

exp.

9

o

1.0

Time/Cycle Time [-]

Figure 11. Pressure swing in adsorption bed ( dp: # 200-325, cycle time: 1.0 s ) (Reprinted with permission from ref. [21]).

calc.

291

100

dp # 48- 80 # 80-150 #200-325

80 tO

60

o o

t--

0

0 tO 0 tO

exp. 9

"

I

calc.

!

I

o

o t

~

o

~

t

40

: ...... .~.....

X

o

........... ~

20

O0

\

9 #".

, 10

-1

10

-0

10

+1

Cycle Time [sec]

Figure 12. Dependency of oxygen concentration on cycle time. (Reprinted with permission from ref. [21]).

The pressure swings obtained from experiments and calculations in spite of the simple model are in good agreement. The dependence of the oxygen concentration in the product gas on cycle time is showed in Figure 12. The oxygen concentration depends on the cycle time, since the pressure swing profile is controlled by the cycle time as mentioned before. The oxygen concentration also varies considerably with particle size of zeolite. The highest oxygen concentration was achieved with the particles of 80 to 150 mesh size. Usually, the production capacity of PSA is defined as: Production Capacity [Nm 3 product gas/m 3 bed~r] -

Since the oxygen production capacity 10 Nm~ oxygen m -3 bed hr -1, the oxygen URPSA can be estimated to 10 times more highest oxygen production capacity can be 150 mesh size were used (Figure 13).

G product

Vbed

(35)

of normal PSA is usually below production capacity of the proposed than in the case of normal PSA. The achieved, when the particles of 80 to

292 §

10

c-

E

10 ``2

.1

E

Z =....., >,, .m

t~

9

.

9

t~ O r

.

9

10 +1

"

o

9

.O n

9

O

..

0-

. "

O L.. f:l. r

9

(D X

o

dp

10 ~

exp.

# 48- 80 # 80-150

#200-325 1

0 -1

9

10

-1

calc.

9

,~

'

10

-0

10

+1

Cycle Time [sec]

Figure 13. Dependency of oxygen production on cycle time. (Reprinted with permission from ref. [21]).

Oxygen yield is defined in conventional form as follows" (Oxygen output) G productC prouct • Oxygen yield [%] - /Oxygen input) xl00= G feedC feed

(36)

The cycle time dependence of the oxygen yield is shown in Figure 14. The oxygen yield of the URPSA is less t h a n 5%. It has become clear t h a t the oxygen enrichment by this URPSA gives extremely high oxygen production capacity and low oxygen yield. The piston-driven ultra rapid pressure swing adsorption (URPSA) is a new technical solution for oxygen enrichment from air using zeolite. Nearly 60% oxygen-enriched gas was produced with an oxygen production capacity of about 100 Nm 3 m -3 bed hr ~ which is one order of magnitude higher t h a n t h a t obtained by commercialized oxygen-enrichment PSAs. This method has been shown as a

293

variation of the improvement and optimization of cyclic operations and feasibility studies for application of this novel PSA.

dp

exp

cal¢"

# 48- 80 # 80-150 #200-325

~x x\ \

\

o~

"O

>.m

"...

~ ...

ko

,./

•

o

•

• °

o

$ -1

lO

• -0

lO

+1

lO

Cycle Time [sec]

Figure 14. Dependency of oxygen yield on cycle time. (Reprinted with permission from ref. [21]).

3.2. P a r a l l e l p a s s a g e c r o s s - f l o w a d s o r b e r To measure fast pressurization transients, an experimental apparatus has been constructed [5]. The schematic picture can be seen in Figure 15. The experimental results of dynamic pressurization show, that this method is suitable for the determination not only the measurement of pressurization velocity, but the real adsorption velocity too. That will bridge over the difficulties originated from the differences among industrial applications and the equilibrium or very slow scientific measurements. With the help of these equipments the RPSA processes can be carefully examined. This system is new in this field in the sense of high frequency as well as the easy data acquition. The maximum velocities of different mass flows show the boundary values of URPSA technologies ( Table 1).

294 Table 1 The different mass flows for zeolite 5A Mass flow Gas

Inlet (mol/s)

Pressurization (moVs)

Adsorption (moYs)

Specific adsorption (mol/s/dm 3)

He He CH4 N2 Ar 02 CO2

0.29 0.24 0.107 0.088 0.078 0.08 0.061

0.216 0.24 0.021 0.024 0.036 0.036 0.0008

0.074 0 0.086 0.064 0.042 0.042 0.053

0.056 0 0.065 0.059 0.032 0.032 0.04

Vacuum pump

Gases:

(~

piezosenzitive pressure

gauge (range: 0-10 & 0-100 bar)

Helium Hydrogen

......

~ _ . _ _ _ ~

._L,

CT 101~----

•[

Nitroge

IBM PC

Methane 007 .__J

I

CT 101 TA 102: signal processing and conditioningunits

Figure 15. Apparatus for the investigation of fast adsorption processes. (Reprinted with permission from ref. [5]).

3.2.1. Method of building cross-flow adsorbers The schematic picture of a cross-flow parallel passage adsorber-desorber can be seen on the Figure 16. The structure of the adsorber-desorber was produced by alternate crosswise stacking of square shaped pieces of sheets provided by adsorbent layers on both sides. The technique of manufacturing consists of two parts. One is the method of adsorbent layer production, and the other the manufacturing method of structured metal frame. Because of the necessary high heat conductivity, usually the copper has been chosen as frame material. Taking into consideration, the thin

295 layers, a thermocompression or the so called diffusion bounding can be the suitable technical solution.

Figure 16. Schematic picture of a cross-flow parallel passage adsorber-desorber.

This process uses both heat and pressure to make a bond. Knowing the equilibrium diagram of a copper-oxygen alloys the bounding process takes place without chemical changes only in inert atmosphere or high vacuum. For the bounding process necessary to decide the pressure, t e m p e r a t u r e and time. The usual values for the t e m p e r a t u r e T b = ( 0 . 7 ~ 0.8)Tmp, for the pressure P = 3-100 MPa, the m a x i m a l bounding time 1-240 minutes. On the other h a n d could be hardly find material into which copper does not diffuse. The graphite seems to be the best. It can bear the relatively high pressure too. Among our circumstances first we pressed the sheets and t h a n they were heated. The applied bonding pressure was 27 MPa. As for the t e m p e r a t u r e 1100 K and practically 0 time gave good result. The two 0.1 m m thick sheets were bounded. The 0 time needs a small explanation. It means t h a t the h e a t i n g period took 3 hours, following directly the cooling period of 10 hours. An adsorbent layer ( NORIT activated carbon powder d-0.01 mm) in the suspension form with starch was smeared. The layer thickness - after a drying at 1500 C t e m p e r a t u r e - rises as highly as 0.1 - 0.2 mm.

3.2.2. The adsorption heat and heat transfer Let us consider how large is the h e a t transfer capacity of the former adsorberdesorber construction. The heat conductivity of a thin copper layer is larger at least with two m a g n i t u d e s t h a n the carbon or other adsorbents layers ones. Thus the governing transfer properties are the conductivities of adsorbent layers. The t h e r m a l conductivity of a carbon p o w d e r - 0 . 6 kJ/m hr K. The average thickness is - 0.1-0.2 mm. Using these data the technically possible h e a t transfer velocity

296 - 800 J/m 2 s K. Having based on the proposed construction the transfer surface is 1600 m2/m 3. Taking into consideration that the adsorbent amount is 70-100g/m 2, and the heat of adsorption (isosteric) in average 28 kJ/mol, the necessary cooling velocity 800-1000 J/m2s. This heat flux seems to be enough for the desorption at a lower pressure. The former experiments concerning to the adsorption velocities allow an estimated cycle time in the magnitude of 10 s. In this case for a recovery process (600 MW Boiler using liquid natural gas [LNG]) for 1700 000 Nm3/hr an adsorber-desorber system with TCRPSA (thermally coupled rapid PSA) needs -500 m 3 volume of adsorber-desorber apparatus. Comparing with the supposed recovery process (Figure 17), the necessary volume is only a small fraction of them. The proposed technological process can be seen on the Figure 18. The possible adsorption capacity is -3000 m 3 flue gas/hr/m 3 bed.

Flue Gas Concentration CO; 9 % 02: 2.1% N,: 71.4 % H,O: 17.5 %

Example: 600 MW Boiler using LNG

§

iNo,

j

ls,A0sor0e IH2~ [[

I

Air Blower

^~ ~ 2 n d ,-),.,,~Ads. Compressor

Heat Exc.

1st Vac.Pump H~O 2nd Vac.Pup.

Flue Gas Generated: 1,718,000 Nm3/h CO2 to be Recovered: 131,430 Nm'/h Example of Adsorber Towers: 10 mDiaX15 mHX16(1st), 8 mDiaXl0 mHX8(2nd)

Figure 17. Supposed recovery process of CO2 by PSA (lst adsorbers for CO2, H20, the 2nd for CO2 drying).

297 Example: 600 MW Boiler using LNG

NH,

t :bil sorber APsA I

NO.

Flue Gas Concentration CO,: 9 % O,: 2.1% N,: 71.4 % H,O: 17.5 %

,l

1st Blower

removal

Air Blower

-lwr I r

Vacuum

Pump

Compressor

Heat E x c . ~

Heat Exc.

Flue Gas Generated: 1,718,000 Nm?h CO, to be Recovered: 131,430 Nm'/h

Figure 18. Supposed recovery process of CO2 by TCRPSA.(Necessary adsorber tower l pc: 10mDiaX 15mH).

CONCLUSION Results of different experimental and theoretical studies suggest that the traditional mass transfer devices for adsorption p r o c e s s e s - in the spite of the fact that the packed beds are now in fashion and they are cheap - cannot answer the new challenges arising from the environmental protection technologies. The main disadvantages of the traditional packed bed are the high pressure drop leading to significant power costs and the low specific capacity resulting telatively large apparatures. The applications of adsorptive separation on same important fields like the t r e a t m e n t of industrial offgases or large amount of CO2 emmisions were the new technical solutions. In this chapter we introduced some solutions for the decreasing of energy demand originating from the hydrodynamic properties and the application of inner heat coupling, demonstrated with new constructional solutions like parallel passage adsorbers. The velocity barriers for the application of piston driven adsorption methods have been discussed. These basic elements

298

are to be expected for the benefit of the constructors involved in the large capacity adsorber desing.

NOMENCLATURE A,B c

C Do DK Dm

Dp dp dp F G H k2 KFav M M(x,t) M* (~l,t) Mi N p P p(x,t) p*('rl ,t) p*i (t)

pF PI Ap* q Qst

R ra Re

Rp S T Tb Tmp

constants of the Ergun-type equation gas concentration, kmol/m~ molar fraction constant defined by Eq. (32), m2/s Knudsen diffusion coefficient, m2/s molecular diffusion coefficient, m2/s pore diffusion coefficient, m2/s particle diameter, m particle size, mesh flow rate, mol/s flow rate, m3/hr bed height, m tortuosity factor overall mass transfer coefficient,1/s molar weight kg/kmol mass flux kmol/m2/s dimensionless mass flux molecular weight, kg/mol mass, mol partial pressure, Pa total pressure, Pa pressure, Pa dimensionless pressure dimensionless pressure at the ith gauge feed pressure, Pa initial pressure, Pa dimensionless pressure difference amount adsorbed, mol/kg heat of adsorption, J/mol gas constant, kJ/kmol/K macropore radius, m Reynolds number particle radius of adsorbent, m cross sectional area, m 2 temperature, K bounding temperature, K melting point, K

299 t tc U X

V

time, s adsorption time, s gas velocity, m/s axial co-ordinate, m volume, m 3

Greek letters adsorption capacity coefficient, m3/kg ~0 0 pre-exponent constant in Eq. (9) ~c 7 5 E Ep

P T Tc,TPR OJ

packing density, kg/m 3 ratio of lin. and non-lin, friction forces void fraction particle void fraction dimensionless co-ordinate drag coefficient gas density, kg/m a dimensionless time dimensionless times angular velocity, rad/s constant given by ~c,

ACKNOWLEDGEMENT We thank the New Energy and Industrial Technology Development Organization and Hungarian National Scientific Fundation for supporting this collaborative work and personally to B. Ferencz and J. Argyelhn for drawing figures.

REFERENCES

1. S. Ergun, Chem. Eng. Prog., 48 (1952) 89. 2. G. Horvhth, T. Szhnya, P. Szolcshnyi, A. L~szl6, L. Bilitzky and J. Hfittner, Method for the separation of helium-neon mixtures, Hungarian patent No. HU 195 690 (1990). 3. L. Arhnyi, J. Hay, G. Horvhth, Z. Szab6, T. Szhnya and P. Szolcshnyi, Magy. K~m. Lapja, 36 (1980) 39. 4. K. Kutics, Hydrodynamic Study of Packed Beds, M.Eng. Thesis, University of Veszpr~m, 1981. 5. G. Horvhth, K. Kutics and M. Suzuki, NATO ASI Series. vol. 491,511 Kluwer, 1996. 6. J. Guan and Z. Ye, Optimization of the Operation Parameters for Rapid Pressure Swing Adsorption, Fundamentals of Adsorption Proceedings of Fourth International Conference on Fundamentals of Adsorption, (1992) 243.

300

.

8. .

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

J.Hart and W. J. Thomas, Gas Sep. Purif., 5 (1991) 125. R. L. Jones, G. E. Keller and R. C. Wells, Rapid Pressure Swing Adsorption Process with High Enrichment Factor, US Patent No. 4 194 892 (1980). R.L. Jones and G. E. Keller, J. Separ. Proc Technol., 2 ( 3 ) (1981) 17. K. Kawazoe, I. Sugiyama and Y. Fukuda, Kagaku kogaku, 30 (1966) 1007. D.E Kowler and R. H. Kadlec, AIChE Journal,18 ( 6 ) (1972) 1207. G.W Miller, K. S. Knaebel and K. G. Ikels, AIChE Journal, 33 (2) (1987) 194. S.Nakao and M. Suzuki, J. Chem. Eng. Japan, 16(2) (1983) 114. C.L. Pritchard and G. K. Simpson, Chem. Eng. Res. Des., 63 (1986) 467. R. Rota and P C. Wankat, AIChE Journal, 36 (9) (1990) 1299. A. Sakoda and M. Suzuki, Sep. Tech., 1 (1991) 73. C.W. Skarstrom, Recent Developments in Separation Science, N. N. Li (ed.), CRC Press, Ohio, vol. 2, 95 1975. M. Suzuki, Adsorption Engineering, Kodansha, Tokyo, 1990. P.H. Turnock and R. H. Kadlec, AIChE Journal, 17 (2) (1971) 335. L. Rousar, P. Ditl, L. Kotsis and K. Kutics, Chem. Eng. Comm., 112 (1992) 67. M. Suzuki, G. Horvath, D.M. Ruthven and S. Farooq, Research Report, NEDO, Tokyo, 1996.