Flash Devolatilization.pdf

Vol. 6

FLASH DEVOLATILIZATION

305

FLASH DEVOLATILIZATION

Introduction

Flash devolatilization is a method for removing solvents and unreacted monomers

from a polymer solution. The solution is heated and ﬂashed into a chamber oper-

ated at reduced pressure. Boiling occurs by homogeneous nucleation and generates

the surface area necessary for mass transfer of the volatile components from the

liquid phase to the vapor phase. The overhead product is sent to a solvent recov-

ery system. The bottom product, devolatilized polymer, is collected compressed,

and pumped away. Figure 1 shows a typical process. Flash devolatilization is com-

monly used in vinyl polymerizations and in some condensation polymerizations.

High volume, commercial examples include the removal of unreacted ethylene

from low density, high pressure polyethylene (LDPE), the removal of hexane in

the solution process for high density polyethylene (HDPE), and the removal of

styrene and ethylbenzene in the bulk polymerization of polystyrene (see E THYLENE

P OLYMERS , LDPE; E THYLENE P OLYMERS , HDPE; S TYRENE P OLYMERS ;B ULK AND

S OLUTION P OLYMERIZATIONS R EACTORS .

Important process variables are the preﬂash temperature T in and the

post-ﬂash pressure P out . The ﬂash illustrated in Figure 1 is rapid, adiabatic, and

approaches thermodynamic equilibrium. The response variables are the outlet

compositions of the vapor and polymer streams and their common outlet temper-

ature T out . The ﬁrst step in analyzing this process is the estimation of thermody-

namic equilibrium.

Equilibrium Calculations

A heat balance gives

T out =

T in − ν C p

(1)

306

FLASH DEVOLATILIZATION

Vol. 6

T in , P in

Polymer

solution

Shell and tube

heat exchanger

Back-pressure valve

Vapors

to vacuum

T out , P out

Gear or screw pump

Polymer

melt

Fig. 1. Flash devolatilization process.

is the

heat of vaporization evaluated at T out , and C p is the speciﬁc heat of the polymer

solution, prior to evaporation, evaluated as a mean value over the range from T in

to T out . An overall material balance gives

is the weight fraction of the entering stream that is vaporized,

w in = ν +

− ν

) out

(2)

where w is the weight fraction of solvent. The volume fraction of solvent deﬁned

φ = ρ mixture /ρ solvent

(3)

Assume that the partial pressure of the solvent equals the total pressure P out

in the ﬂash chamber. Then

P sat = φ out exp 1

−

− φ out )

+ χ

− φ out ) 2

(4)

is the FIory-Huggins interaction

coefﬁcient (typically 0.45 for a good solvent). The vapor pressure of the pure solvent

P sat is a function of T out and is obtained from correlations such as the Antoine or

Lee–Kesler equations. Equations 1–4 are solved simultaneously to determine the

outlet temperature and solvent concentration.

The above treatment assumes there is only one volatile component, but the

equations can be extended to multicomponent systems by writing a separate ver-

sion of equation (4) for each component.

Partial pressure of component n

Vapor pressure of pure component n =

(

φ out ) n exp 1

−

− φ out )

+ χ

− φ out ) 2

(5)

where

P out

where m is the degree of polymerization and

Vol. 6

FLASH DEVOLATILIZATION

307

φ out ) n is the volume fraction of the n th component. The various partial

pressures sum to P out . The exponential term in equation 5 is the activity coefﬁ-

cient for that component. As indicated above, the same value is assumed for each

solvent. It will have a value of about 4 for nearly complete devolatilization of a

high molecular weight polymer. Special cases arise when the mixed solvents are

highly nonideal or if the polymer crystallizes during the ﬂash.

Many polymers degrade at the temperatures needed for devolatilization. Un-

zipping to monomer is typical of stryenics. The resulting volatiles are removed in

the ﬂash but will be regenerated if there is a substantial inventory of polymer

in the process line after the ﬂash. The degradation kinetics can be measured by

isothermal weight loss in a thermal gravimetric analyzer. The degradation rates

can be ﬁt as a ﬁrst-order reaction with Arrhenius temperature dependence. Unzip-

ping to monomer imposes a practical limit of a few hundred ppm for polystyrene

(see D EGRADATION ).

Staged Flashes and Induced Foaming

A common application of ﬂash devolatilization is the removal of 10–15%

styrene-ethylbenzene from polystyrene. One commercial process uses T in =

240 ◦ C

and P out =

1.3 kPa (10 torr). Equilibrium calculations for an adiabatic ﬂash predict

500 ppm; actual operation gives about 1200 ppm.

The secret to approaching equilibrium in a ﬂash is the generation of sur-

face area through foaming. Vapor bubbles nucleate, grow, and coalesce into an

open-celled foam provided T out exceeds the glass-transition temperature of the

polymer. The small web thickness allows quick diffusion of the solvent from the

polymer ﬁlm, and the open-celled structure allows ready escape of the vapors.

Thus equilibrium is approached; but as suggested by the polystyrene example,

mass transfer limitations arise when w out is low. Market demands for lower

volatiles than achievable in a single ﬂash prompted the use of a second ﬂash.

One process decreases the severity of the ﬁrst ﬂash so that 1–2% volatiles remain

to cause natural foaming in the second ﬂash. The partially devolatilized polymer

is reheated and ﬂashed a second time, again with P out

220 ◦ C and w out =

1.3 kPa (10 torr) but

240 ◦ C because the adiabatic temperature drop in the second

ﬂash is small. In practice, this process gives w out 500–700 ppm. An alternative

process retains the severe ﬁrst ﬂash but injects a foaming agent prior to the sec-

ond ﬂash. The foaming agent should be innocuous and highly volatile. It need not

have signiﬁcant solubility in the polymer, eg, water has been used for polystyrene.

The primary purpose of the foaming agent is to generate an open-celled foam. A

secondary beneﬁt is the reduction in solvent partial pressure at a ﬁxed total pres-

sure in the ﬂash chamber. The foaming agent can be mixed with the polymer using

a short, melt-fed single-screw extruder. Pressure is applied so that the foaming

agent remains a liquid. Static mixers have also been used.

A strictly adiabatic ﬂash has disadvantages when polymerization or degrada-

tion reactions occur in the heat exchanger. Adiabatic operation forces high values

of T in , which can lead to by-products or undesired low molecular weight polymer.

When the inlet viscosity is high, ﬂashed polymer can be displaced from the heat

exchanger by the incoming polymer solution. This allows boiling in the heat ex-

changer, giving higher heat transfer coefﬁcients and a lower value for T in because

where (

T out =

now with T out

308

FLASH DEVOLATILIZATION

Vol. 6

some of the heat of vaporization is supplied directly rather than as sensible heat.

In theory, the equilibrium ﬂash calculations are easily modiﬁed to account for a

partially vaporized feed. In practice, the extent of vaporization occurring within

the tubes is difﬁcult to calculate. A conservative approach to designing the heat

exchanger is to assume that boiling does not occur. However, there is no need to

apply back-pressure to the heat exchanger when the entering solution is highly

viscous. For polystyrene, entering concentrations of 70% polymer are sufﬁcient

to allow boiling in the heat exchanger. Some designs install the heat exchanger

directly above the ﬂash chamber with no intermediate piping. Other designs use

elaborate nozzles and foam distributors, but there is little evidence that these

devices provide any practical advantage.

The generic process in Figure 1 can be applied to relatively dilute polymer so-

lutions, although two- or three-staged ﬂashes are needed for complete devolatiliza-

tion. A process for recovering 10% polyethylene from a xylene solution employs a

ﬁrst ﬂash with w in =

0.10, T in =

280 ◦ C, and P out =

55 kPa (8 psig). This gives a

0.85. The P in for this ﬂash is set at 1896 kPa (275 psig)

to avoid boiling and possible plugging of the heat exchanger. The second ﬂash

gives w out =

180 ◦ C and w out =

0.99 and provides the fed to a rear-vented, single-screw extruder to

complete the devolatilization.

Extruder and Wiped-Film Devolatilization

The surface area needed for mass transfer from the polymer phase to the vapor

phase can be generated by mechanical devices such as single- and twin-screw ex-

truders and wiped-ﬁlm evaporators, but strictly mechanical generation of surface

area scales up very poorly. The throughput of such devices scales as D 3 , but the

generation of surface area scales only as D 2 where D is the screw diameter. Surface

area created by foaming scales as D 3 so that foaming is a highly desirable comple-

ment to the mechanical generation of surface area. Foaming occurs naturally in

many extrusion processes so that some vendors suggest using D 2 . 5 as the scale-up

factor where there is no explicit ﬂash. Figure 2 shows a devolatilization scheme

that has been employed in extruders to attain maximum removal of solvents.

The polymer solution is preheated and then ﬂashed into the barrel of the

extruder at a point downstream from the normal inlet. The polymer moves from

left to right. Solvent vapors ﬂow countercurrently and are removed at the rear

end of the extruder. The partially devolatilized polymer is transported to the

right. It is compressed, reheated, and pumped to a milling zone where addi-

tional devolatilization is caused by the mechanical generation of surface area.

Figure 3 shows idealized results for devolatilization as a function of operating and

scale-up variables. The y -axis indicates the extent of devolatilization, measured as

( w out −

∗

)/( w in −

∗

) where w

∗

T out =

is its equilibrium concentration of volatiles corre-

sponding to the temperature and pressure in the extruder vent. The x -axis mea-

sures the collection of operating and scale-up variables that determine the extent

of devolatilization according to a conventional model. These variables are the

screw diameter D , the molecular diffusivity of the volatile component D the screw

speed N , and the polymer ﬂow rate Q (see E XTRUSION ).

For the case of devolatilization extrusion with back-venting, Figure 2 shows

a discontinuity at the origin that corresponds to a ﬂash step at the inlet of the

extruder. To a ﬁrst approximation, the decrease in solvent concentration for this

Vol. 6

FLASH DEVOLATILIZATION

309

Vapors

to vacuum

Vapors to vacuum

Milling section

Devolatilized

polymer melt

Heat

exchanger

Polymer

solution

Fig. 2. Devolatilizing extruder with back-venting.

0.1

Without back-venting

0.01

With back-venting

0.001

0.0001

D 2

√

N/Q

Q and thus occurs to the same extent in any

extruder, regardless of size. In the milling region, the solvent concentration de-

creases gradually, and t he devolatilization capacity of this region scales rather

uneconomically as D 2 √ D

Q . In practice, some foaming occurs in the milling

section of extruders and in wiped-ﬁlm evaporators. This foaming signiﬁcantly im-

prove s de volatilization. Some extruder vendors suggest using a scale-up factor of

D 2 . 5 √ D

Q to account for foaming.

Equipment Considerations

Flash devolatilization is a relatively mature technology for the removal of

volatile components from polymer melts and solutions. It was developed by major

Fig. 3. Scale-up of devolatilizing extruders.

ﬂash is independent of D 2 √ D

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: