The Simulation and Control of the Reactive Distillation Process for Dimethylcarbonate(DMC) Production

ICCAS2004 August 25-27, The Shangri-La Hotel, Bangkok, THAILAND

1. INTRODUCTION

Green chemistry focuses on a safer use of chemicals, devising alternative processes and products aimed at substituting the traditional ones, involving use of toxic materials or intermediates and/or by-production of wastes. For an industrial chemical company, the development of environmentally favorable processes and products, which will be at the same time economically compatible, in order to stand the competition with the existing ones, is increasingly a real challenge. Substituting toxic, dangerous, highly reactive chemicals with less reactive, less harmful, more selective building blocks. The scope of this paper is to highlight the DMC production in the substitution of phosgene in industrial productions. [1]

The issue is green chemistry in part of industry. Like this, reactive distillation is interested subject in aspect of industry.

Recently RD process that used for a long time focused again, because RD system has many benefits. From the foregoing examples, the benefits of RD can be summarized as follows [2]:

(a) Simplification or elimination of the separation system can lead to significant capital savings.

(b) Improved conversion of reactant approaching 100%. This increase in conversion gives a benefit in reduced recycle costs.

(c) Improved selectivity. Removing one of the products from the reaction mixture or maintaining a low concentration of one of the reagents can lead to reduction of the rates of side reactions and hence improved selectivity for the desired products.

(d) Significantly reduced catalyst requirement for the same degree of conversion.

The Simulation and Control of the Reactive Distillation Process for Dimethylcarbonate(DMC) Production

Yong Hee Jang and Dae Ryook Yang*

* Department of Chemical and Biological Engineering, Korea University (Tel : +82-2-3290-3298; E-mail: dryang@korea.ac.kr)

Abstract: Reactive distillation (RD) is a combination process where both separation and reaction are considered simultaneously in a single vessel. This kind of combination to enhance the overall performance is not a new attempt in the chemical engineering areas. The recovery of ammonia in the classic Solvay process for soda ash of the 1860s may be cited as probably the first commercial application of RD. The RD system has been used for a long time as a useful process and recently the importance of the RD is enlarged more and more. In addition to that, the application fields of RD are diversely diverged. To make the most of the characteristic of RD system, we must decide the best operating condition under which the process shows the most effective productivity and should decide the best control algorithm which satisfies an optimal operating condition.

Phosgene which is a highly reactive chemical is used for the production of isocyanates and polycarbonates. Because it has high reactivity and toxicity, its utilization is increasingly burdened by growing safety measures to be adopted during its production.

Dimethyl Carbonate (DMC) was proposed as a substitute of phosgene because it is non-toxic and environmentally benign chemical. In this study, RD is used for DMC production process and the transesterification is performed inside of column to produce DMC. In transesterification, the methanol and ethylene carbonate (EC) are used as the reactants. This process use homogeneous catalyst and the azeotrope exists between the reactant and product. Owing to azeotrope, we should use two distillation columns. For this DMC production process, we can suggest two configurations. One is EC excess process and the other is methanol excess process. From the comparison of steady state simulation results where the Naphtali-Sandholm algorithm is used, it showed the better performance to use the methanol excess process configuration than EC excess process. Then, the dynamic simulation was performed to be based on the steady state simulation results and the optimal control system was designed.

In addition to that, the optimal operating condition was suggested from previous results.

Keywords: Reactive Distillation, DMC, Dimethylcarbonate, Naphtali-Sandholm method

(e) Avoidance of azeotropes. RD is particularly advantageous when the reactor product is a mixture of species that can form several azeotropes with each other. RD conditions can allow the azeotropes to be reacted away a in a single vessel.

(f) Reduced by-product formation.

(g) Heat integration benefits. If the reaction is exothermic, the heat of reaction can be used to provide the heat of vaporization and reduce the reboiler duty.

(h) Avoidance of hot spots and runaways using liquid vaporization as thermal fly wheel

Though RD has many benefits and a long history (The recovery of ammonia in the classic Solvay process for soda ash of the 1860s may be cited as probably the first commercial application of RD) the reason why RD was not used is that optimal operating condition of process was not gained easily and stability of process was not good. The simulation of the reactive distillation for the DMC production was performed to solve these problems.

2. DMC PRODUCTION PROCESS

In non-phosgene DMC production process, various routes exist as shown Fig. 1. Among those routes, DMC production process through transesterification reaction has benefits that can produce DMC by modifying EG production process and that can simultaneously produce EG. However when DMC is produced through transesterification reaction, it forms an azeotrope with the reactant, methanol. For this reason, the purity of DMC can not be over an azeotrope composition, so using more than two distillation columns is needed to improve

is methanol excess process and the other is EC excess process.

Fig. 1 Various DMC synthesis routes

As shown Fig. 2, in the case of methanol excess process, there exist an azeotropic distillation column to separate methanol and DMC produces from reactive distillation column. This process can produce high purity DMC and recycle separated methanol to fee stream.

Fig. 2 Methanol excess process for DMC production

Fig. 3 shows EC excess process. AS the methanol entirely converts into product in ideal case, almost pure DMC is produces in the top column. On the other hand, unreacted EC, EG and catalyst in bottom shift another distillation column to be recycled.

Fig. 3 Methanol excess process for DMC production 3. REACTION KINETICS

Overall transesterification reaction of methanol with EC is written as follows: [3]

jo_Y jo_Y

v

v j v R Yjo_Zvo jo_Zv j v

vjo_Z jo_Y jo_Y

vo R vo

(1) This reaction is catalyzed by a homogeneous catalyst, sodium methylate, a strong alkaline soluble in both methanol and EG, and follows a nucleophilic substitution mechanism as given by:

jo_Y jo_Y

v

v j v R jo_Zv^T jo_Zv j v

vjo_Y jo_Yv^T OpP

OpP R jo_Zvo jo_Zv j v

vjo_Y jo_Yvo R jo_Zv^T

(2)

R jo_Zv^T jo_Zv j v

vjo_Z OppP

OppP R jo_Yvo jo_Yv^T

(3)

R jo_Zvo Ojo_YvoP_Y

OpppP R jo_Zv^T

(4) Among the above elementary steps, reaction (2) is the slowest one, and controls the total rate. As a result, the kinetics equation for the transesterification reaction is derived as follows: [3]

3

3

EC EC

EG DMC

EC CH OH

CH OH

r dC dt

C C

k C C k

   C

(5)

Table 1 display the rate constant in equation 5.

MeOH EC

DMC

EC+Cat.

EG

(Excess)

EC+EG+Cat.

MeOH EC

DMC+MeOH

EG+Cat.

MeOH

(Excess)

Azeotropic Distillation System

DMC

Table 1 Estimated value of rate constants [3]

Active energy

(kJ mol^-1) Rate constant Forward 13.06 k_ 1.3246e^{13060 /RT} Reverse 28.60 k_ 15022e^{28600 /RT}

4. STEADY STATE SIMULATION

4.1 Steady state simulation by Naphtali-Sandholm Algorithm

Both the MESH (Material balance, vapor-liquid Equilibria equation, mole fraction Summation and Heat balance) equations for systems in vapor-liquid equilibrium and the MERQ (Material balance, Energy balance, Rate equations for mass transfer, and phase eQuilibrium at the vapor-liquid interface) equations for mass transfer can be used to model distillation processes. The MERQ equations have been recommended by some researchers (Sundmacher and Goffman, 1995) and are gaining in popularity due to advances in computational power available. However, they are more complex, require the estimation of more empirical parameters, and appear to offer no improvement in accuracy for most systems. The MESH model was used here. The MESH equations was composed as follows : [4]

1) M equations: Material balance for each component (C equations for each stage)

, 1 , 1 1 , 1 ,

, ,

( ) ( )

i j j i j j i j j i j

j j i j j j i j i j

M L x V y F z

L U x V W y R

     

     _, 0

0

(6)

2) E equations: phase Equilibrium relation for each component (C equations for each stage)

, , , , 0

i j i j i j i j

E y K x (7)

3) S equations: mole fraction Summations (one for each stage)

, ,

1 1

( ) 1.0 0, ( ) 1.0 0

C C

y j i j x j i j

i i

S

¦

y  S

¦

x  ⁽⁸⁾

4) H equation: energy balance (one for each stage)

1 1

1 1

( ) ( )

j j j

j j

i j L j V j F

j j L j j V j

H L H V H F H

L U H V W H Q

 

   

     (9)

Figure 4 shows Naphtali-Sandholm algorithm for distillation process simulation based on MESH equations, this method has various alternatives as given specification.(Table 2) In this paper, two specifications are used: and an enthalpy balance equation instead of suggested alternatives.

,1

0

v

i

 D

¦

Fig. 4 Algorithm for Naphtali-Sandholm Simultaneous Correction (SC) method

Table 2 Alternative functions for H1

Specification Replacement for H1

Reflux or reboil (boilup) ratio,

(L/D) or (V/B)

¦

l_i^,1( /L D)

¦

v_i^,1 0 Stage temperature,

T_D or T_B T¹T_D 0 Product flow rate,

D or B

¦

v_i^,1D 0

Component flow rate in

product, d_i or b_i v_i^,1d_i 0 Component mole fraction

In product, y_iD or x_iB v_i^,1(

¦

v_i^,1)y_iD ⁰ Naphtali-Sandholm algorithm find quickly the results when give the initial condition that near the steady-state condition, but it diverge easily when give bad initial condition.

Therefore selection of initial condition is very important.

Table 3 Alternative functions for HN

Specification Replacement for HN

Reflux or reboil (boilup) ratio,

(L/D) or (V/B)

¦

v_{i N}^, ( / )V B

¦

l_{i N}^, 0 Stage temperature,

TD or TB

N B 0 T T Product flow rate,

D or B

¦

l_{i N}^, B 0 Component flow rate in

product, di or bi

, 0

i N i

l b Component mole fraction

in product, yiD or xiB

, ( , )

i N i N iB

l 

¦

l x ⁰

4.2 Assumption and simulation conditions

For the effective simulation suggested that some assumption as follows :

(1) Chemical equilibrium is attained on all reactive stages.

(2) By-products formation is negligible (3) Negligible heat losses

(4) Ideal vapor phase

The assumption of chemical equilibrium allows the reaction kinetics to be neglected, simplifying the modeling of the reactive stages of the column. Negligible by-product formation allows the number of components to be reduced.

Negligible heat losses allow enthalpy derivatives to be effectively excluded form the model, and finally, and ideal vapor phase allows fugacity coefficients to be neglected. [4]

Under these assumptions, equilibrium equations and thermodynamic properties are needed to simulate the steady state. As mentioned above modified Raoult’s law is applied and NRTL activity coefficient model is used in liquid phase.

NRTL activity coefficient model, equation 11, to calculate the activity coefficient needs the binary parameter. However there are not the binary data of DMC and EC in database. Therefore these values are calculated based on the experimental information of the vapor/liquid equilibrium data. [5]

1 1

1 m

ji ji j E m

j

i m

i li l

l

G x

g x

RT

G x

¦

W

¦ ¦

(10)

1 1

1

1 1 1

ln ( )

m m

ji ji j m r rj rj

j j ij r

i m m ij m

li l j lj l lj l

l l l

G x x G

x G

G x G x G x

W W

J  W 

¦ ¦

¦ ¦ ¦ ¦

(11)

Fig. 5 NRTL parameter estimation result (MeOH and DMC)

Table 4 Thermodynamic properties

MeOH EC DMC EG

Normal Boiling Temp.

(ଇ) 64.70 238.00 90.35 197.30

Critical Temp.

(ଇ) 239.43 516.85 274.85 446.55 Critical Press.

(kg/cm2) 82.555 69.035 45.887 87.420 Heat of

Formation

(cal/mol) -48004.92-121070.98-136166.05-91945.64

Besides the temperature dependent function, Cp and vapor pressure, used the polynomial that fitted data in database.

4.3. steady state simulation results

The steady state simulation results of suggested configurations are compared each other and as shown Fig. 5, the 28.77% top composition of DMC is produced in the case of methanol excess process. On the other hand, that of DMC is 38.18% in EC excess process.

However Table 5 shows methanol excess process has a higher DMC conversion and productivity. Therefore methanol excess configuration can be concluded a more effective process and this is considered as a good configuration for DMC production.

Fig.6 Steady state simulation result of methanol excess process

Fig.7 Steady state simulation results of EC excess process

Table 5 Comparison result of methanol excess process and EC excess process

Methanol excess EC excess Top product

rate(g/Hr) 666.89 28.74

Bottom product

rate(g/Hr) 145.00 783.15

Weight fraction of DMC in Top product

(%)

28.77% 38.18%

Conversion (%) 99.5%

(EC conversion)

47.31%

(MeOH conversion) DMC production rate

(g/Hr) 191.86 10.97

5. DYNAMIC MODELING AND CONTROL The dynamic simulation was composed of MESH equations whose derivative term is non zero. In dynamic simulation

results, composition is changed by top product rate change.

This data is based to compose the PI controller parameter. The controller parameters of PI controller were calculated by these results.

Fig.8 Open-loop response to step increase in distillate rate As shown figure 9, DMC composition in top product is controlled by well tuned PI controller. But dynamics of bottom methanol composition is very slow.

Fig.9 Regulation control results with step increase in EC feed rate

6. CONCLUSION

For this DMC production process, we can suggest two configurations. One is EC excess process and the other is methanol excess process. From the comparison of steady state simulation results where the Naphtali-Sandholm algorithm is used, it showed the better performance to use the methanol excess process configuration than EC excess process. Then, the dynamic simulation was performed to be based on the steady state simulation results and the PI control system was designed

RD system can be controlled by just PI controller. However because the dynamics of bottom is very slow and objective functions are satisfied, which are to control one composition

and to minimize energy consumption, the new control strategy for two column system is needed to improve the performance.

REFERENCES

[1] Franco Rivetti, “The role of Dimethylcarbonate in the replacement of hazardous,” International Journal of Control, Vol. 23, No. 4, pp. 123-145, 1989.

[2] Kai Sundmacher and Achim Kienle, “Reactive Distillation”, WILEY-VCH, 2003

[3] Yun-Jin Fang and Wen-De Xiao, “Experimental and modeling studies on a homogeneous reactive distillation system for dimethyl carbonate synthesis by transesterification,” Separation and purification technology, Vol. 34, pp255-263,2004

[4] Martin G. Sneeby, Moses O. Tade, Ravindra Datta, and Terence N. Smith, “ETBE Synthesis via Reactive Distillation. 1. Steady-State Simulation and Design Aspects”, Ind. Eng. Chem. Res., Vol. 36, pp 1855-1869, 1997

[5] Hu-Ping Luo, Wen-De Xiao and Kai-Hong Zhu,

“Isobaric vapor-liquid equilibria of alkyl carbonates with alcohols”, Fluid Phase Equilibria, Vol. 175, pp 91-105, 2000

[6] Bruce E. Poling, John M. Prausmitz and John P.

O’connnell, “The Properties of Gases and Liquid”, Mcgraw-Hill International Editions, Boston , 2001 [7] Martin G. Sneeby, Moses O. Tade, Ravindra Datta, and

Terence N. Smith, “ETBE Synthesis via Reactive Distillation. 2. Dynamic Simulation and Control Aspects”, Ind. Eng. Chem. Res., Vol. 36, pp 1870-1881, 1997

[8] R. Taylor, R. Krishna, “Modeling reactive distillation”, Chemical Engineering Science, Vol. 55, pp 5183-5229, 2000