Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

Application of COMSOL in Chemical Engineering

Suzana Yusup*, M Tazli Azizan Universiti Teknologi PETRONAS
*Chemical Engineering Department, Universiti Teknologi PETRONAS, Seri Iskandar, 31750 Perak, Malaysia [email protected]

The paper highlights the application of COMSOL in solving chemical engineering problems. Two case studies are highlighted including survey feedback from undergraduate students who used the software as part of their modeling and simulation lab work. The first case study is on the prediction of tube skin temperatures of catalytic reforming unit (CRU) heaters. The second case study covers the modeling and simulation of a plug flow reactor for liquid phase methanol synthesis. These case studies were carried out by final year students as part of their requirement before graduation. The student survey assess the students understanding on their ability to simulate case studies related to fixed bed reactor for catalytic hydrocarbon oxidation, non isothermal reactor with isothermal cooling jacket, porous reactor with injection needle, radial effects in tubular reactor: isothermal and adiabatic reactor and monolithic reactor.
Abstract: Keywords: Simulations, Tube furnace, Finite

model can give an indication of the heaters performances and shutdown periods for coke removal. For the past four decades, catalytic reforming of naphtha has evolved rapidly to become one of the most advanced processes in the refining industry. The main objective of the process is to convert low-octane naphtha to high-octane reformate to be used as high performance gasoline fuel. Alternatively, the naphtha reformer unit can also produce high yields of aromatics for petrochemical feed-stocks. Today, reforming technology, catalysts and processes continue to lead petroleum refining in innovation, safety, and reliability as well as environmental impact and profitability. Moreover, because of some environmental regulations on gasoline burning wastes, refiners and catalyst manufacturers are trying to find means for catalyst performance improvement and process design improvement to enhance the selectivity to aromatics and highoctane reformate. All these researches and studies aim at the following objectives: Higher reformate octane yields More efficient catalyst regeneration Longer catalyst life and enhanced surface stability Lower-pressure operation and less hydrogen recycle [1]

Element, Reactor 1. Case Study 1

Prediction of tube skin temperature of catalytic reforming unit heaters (CRU) In a heater, tube skin temperature is one of the most important parameters that need to be closely monitored since it dictates the overall performance of the heater. Moreover, it can also be an indicator of coke formation in the heaters tubes. A high tube skin temperature is a result of refinery fired heater heat-flux imbalances that can cause high coke-formation rates. The materials of the heaters tubes can only withstand a maximum allowable tube skin temperature. When this temperate is reached, either heater firing must be reduced or the heater must be shut down for coke removal. Thus, the developed

In industrial practice, the operating conditions of the reforming unit are usually described in terms of severity. The most widely used measure of reforming severity is the research octane number (RON). After years of operations, the catalyst deactivates; to compensate for the lower activity, the severity (reactor temperatures) of the process must be increased. This leads to the high firing rate of the heaters of reformer unit. In this way, the required RON can be maintained at the desired level. However, this high-severity operation is at the expense of short run-length of heaters due to high firing rate, resulting in higher coke formation rate in tubes and high tube skin temperatures. In a refinery when the catalytic reformer Unit (CRU) has been operating at

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

maximum capacity, it indicates a high severity of operation. This high throughput results in high recycle gas, higher firing rate of heater, more catalyst required in the rectors and an increased load on the columns. This severity has long been put under studies and one of the equipment under close monitoring are the heaters. Because of the high firing rate, the tube skin temperatures are closing to its limits [2]. Tube metal temperatures are often monitored with skin thermocouples. Their accuracy is greatly debated; nevertheless, skin couple readings are often used to determine heater endof-run. Once the tube metal temperature of the heater tube is reached, heater firing must be reduced or the heater must be shut down to remove coke. In the extreme case, long-term operation at high heater tube wall temperature can cause the tube to rupture [4]. Moreover, skin thermocouples are known to be vulnerable and often inaccurate, giving readings some 20 to 50C higher than the real tube skin temperature. The higher reading is caused by the fact that the skin thermocouple construction itself may act as a heat barrier, giving the thermocouple a higher temperature than the tube wall. Skin thermocouples in reformer heaters also tend to fail quite often due to the very high tube temperatures compared to those in other process heaters. Y. Suyadal (2006) [3], carried out experimental work on estimation of kinetic parameters for combustion process in a bench scale fluidized bed. For the modeling work they concluded that the surface reaction constant can be incorporated into a fluidized bed combustor model. Thus experimental results obtained from fluidized bed can establish the basis for the design of fluidized bed combustors. G.D Stefanidis et. al. (2006), [5] simulated a steam cracking furnaces using computational fluid dynamic approach. They found that a smaller flame volume has important effects on the predicted temperature distribution in the furnace as well as on other significant design parameters like the refractory wall and tube skin temperatures. In the present work, the simulation of tube skin temperatures of CRU heaters is carried out using COMSOL MULTIPHYSIC. The Plant Information data is used to verify the model

1.2 Theory The mathematical equation for heat transfer by conduction is as shown in equation (1)

T (kT ) = Q t

(1)

For steady state, equation (2) applies.

(kT ) = Q

(2)

To model heat conduction and convection through a fluid, the heat equation includes a convection term. In COMSOL, this formulation is represented in the Conduction and Convection application mode as in equation 3.

C
(3)

T + (kT + C P Tu ) = Q t

The heat flux vector is defined by the expression within the brackets in equation (3), which for is transported through conduction and convection yields equation (4)

q = kT + C P Tu
Assumptions

(4)

Following are the assumptions made to simplify the modeling: Only one side of tube wall exposed to direct firing is modeled since the tube is symmetrical and assuming the rate of firing is equally distributed to both sides. Only finite length of the tube is considered as the actual length of the tube is higher than the thickness of the tubes. Momentum and mass transfer are neglected. Constant thermal conductivity

1.3 Methodology

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

Figure 1 and Figure 2 show the flow chart followed in simulating the heater tube temperature using COMSOL MULTIPHYSIC and 3-D view of the heater tube respectively.
Model definition Model Navigator Geometry modeling Boundary conditions Subdomain conditions Mesh generation

Figure 2: 3-D view of the heater tube.

1.4 Result and Discussion

Solve Post processing and Visualization

Figure 1: Flowchart in GUI of FEMLAB

The simulation results of tube skin temperatures of the four CRU heaters are shown in Figure 3. The temperature distributions for all four heaters exhibit the same trend, as shown in Figure 3. The tubes wall temperature that was exposed to firing has the highest temperature. The values of the tube skin temperatures for all four heaters are summarized in the following Table 1. For all the cases, the temperatures were found to decrease

Process Fluid Side Firing side

Boundary conditions Inner Process side temp=549C Outer heat flux, q=25948W/m2

Process Fluid Side Firing side

Boundary conditions Inner Process fluid temp=549C Outer heat flux, q=38415 W/m2

Process Fluid Side

Process Fluid Side

Boundary conditions Inner Process fluid temp=549C Outer heat flux, q=24306 W/m2

Firing side

Firing side
Boundary conditions Inner Process fluid temp=549C Outer heat flux, q=20218 W/m2

Figure 3: The front view for tube skin temperarature of Heater1, Heater2, Heater3, Heater4 [clockwise]

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

across the tube to the fluid side. Table 1 shows the comparison between the tube skin temperatures of actual heaters obtained after turnaround with that predicted through FEMLAB simulation. For all the four heaters, simulation results give slightly higher values of the tube skin temperature.
HEATER 1 Base case FEMLAB (15/02/2005) design case 612.04 614 HEATER 2 Operating Base case FEMLAB parameters (30/04/2005) design case Tskinmax (oC) 593.78 601 HEATER 3 Operating Base case FEMLAB parameters (09/02/2005) design case Tskinmax (oC) 597.28 600 HEATER 4 Operating Base case FEMLAB parameters (09/04/2005) design case Tskinmax (oC) 578.15 582 Table 1: Tube skin temperature of Heaters Operating parameters Tskinmax (oC)

Figure 4: Effect of varying heat flux of Heater 1.

When the inlet temperature of Heater No.1 tube increases to 500oC and the process fluid side temperatures increase to 575 oC, the tube skin temperature was found to increase from 614oC to 630oC as shown in Figure 5. Thus, the developed model, can be used to predict the trend of the tube skin temperature of the heater when subjected to variation of operating condition such as variation of the heat flux, and the inlet and outlet temperature of the process fluids.

The difference in the maximum tube skin temperature between the two cases is 7.22oC with % difference of 1.2%, for Heater 2. The other heaters exhibit smaller differences approximately, 2 to 4 oC. Thus, based on the results shown in Table 1, a recommendation can be made as to further increase the firing rate since the maximum allowable tube skin temperature of 650 oC is not reached. By increasing the firing rate in a controllable manner, the required research octane number (RON) can be maintained at the desired level and coke formation within the tube can be avoided. Variations of operating variables The change in the tube skin temperature values, under influence of variation of heat flux, heater tubes inlet and outlet temperatures are also studied. As the heat flux is increased by 10% of its current value to 28542 W/m2, the tube skin temperature was found to increase from 614 oC to 620 oC as shown in Figure 4.

Figure 5: Effects of varying inlet and process fluid temperatures of Heater 1.

1.5 Conclusion It is deduced that the tube skin temperature depends on the heat flux and the inlet and outlet process fluid side temperatures. In addition the developed model and simulation can be used as a monitoring tool to predict the trend of tube skin temperature within the metal tube of CRU heaters for a refinery. The higher the temperature differences between the base case and design case, acts as indicator of amount of coke deposition in the actual tubes. With the

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

simulation ability to predict the values of the maximum temperature that the heater tube can withstand, metal fatigue and fracture can be avoided. The simulation can further be extended to incorporate the effect of coke deposition and variation of temperature profiles within the catalytic reformer unit in future.

operating in temperature range of 313 -323K. Concentration profiles of both reactants and products are simulated using Method of Lines. The approaches combine a numerical method for Initial Value Problem (IPV) of Ordinary Differential Equation (ODE) and a numerical method for Boundary Value Problem (BVP).

2. Case Study 2
Modeling and simulation of a plug flow reactor for liquid phase methanol synthesis In this work a new route to synthesize methanol at milder conditions is explored. The simulation work is carried out to predict the concentration and conversion profiles of reactants and products along the reactor. In addition the variation of operating parameters such as the effects of varying initial flowrates and initial concentrations of reactants on their conversion were also studied. Methanol is a fuel alcohol with similar chemical and physical properties as similar as ethanol. Chemically, it is methane with a hydroxyl radical (OH) replacing one hydrogen molecule [5], methanol is used as a fuel additive for gasoline in the form of methyl tertiary butyl ether (MTBE), an oxygenate that reduces ground level ozone emissions. Methanol is also used in variety of industrial processes, both as a fuel and as a chemical catalyst. Methanol is primarily produced from natural gas, although it can be produced from other nonpetroleum, sources such as coal and biomass, (albeit at a higher cost). Researchers, are currently focusing on how to lower the cost of methanol production from these renewable alternative sources. Methanol can also be produced by steam reforming of natural gas to create a synthesis gas, (syngas), which is a combination of carbon monoxide (CO) and hydrogen (H2). This gas is then reacted with a catalyst to produce methanol and water vapor. Limited studies have been done to simulate and synthesize the saponification process of methyl acetate and sodium hydroxide to produce liquid phase methanol. Weifand Yu et al. [6] studied the hydrolysis reaction of methyl acetate catalyzed by Amberlyst 15 in packed bed reactor

2.1.

Saponification process

The hydrolysis of an ester in a base is called saponification process [7]. The reaction to produce methanol is as following: NaOH + CH3COCH3 CH3CONa + CH3OH Ester hydrolysis in base is called sapofinication as shown in the expression below. RCOR + HO- RCO- + ROH

2.2

Design equation

The schtoichiometric equation for an irreversible Bimolecular-Type Second Order Reaction is as shown below [8]. A+B Products With corresponding differential equation as in equation (5): FAOdXA = (-rA)dV (5)

Insulation

CB CC
Figure 6: Schematic diagram of a tubular reactor.

CD CE

2.3

Methodology

The transient temperature profiles and concentration profiles inside the plug flow reactor are initially solved using explicit finite difference method. Comsol Multiphysics program is utilized to obtain the insights of the problem. The main advantage of Consol is that the equations can be expressed algebraically in terms of finite difference equations. The flow

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

diagram for this function program is as shown in Figure 7 below:

Diffusion mode Boundary Condition Boundary Condition Equation

Start Define Geometry

Inward Flux Impervious or Symmetry Concentration Zero Concentration

n.( Dc) + qc = g n.( Dc) = 0

c = c0 c=0

Input Parameter Values

Choose Application Mode

Table 2: Boundary Conditions Available as post processing variables are the concentration c, the concentration gradient, the diffusive flux and the normal component of diffusive flux. Also available is the source Q.

Define Boundary Settings and Sub-domain Settings for each Mode

Refine each mode

2.4 Concentration Profile

Display Graphical Results

Display Concentration or Temperature Profile

End

Figure 7 shows the variation of concentration reactant along the length of the reactor. Sodium hydroxide and methyl acetate react along plug flow reactor to form sodium acetate and methanol. The reactant concentration starts to decrease from 100 moles/l to around 5 moles/l and, the products starts increases from 0 moles/l to 95 moles/l.

Figure 7: Procedure Identification

One of the important advantages of using COMSOL MULTIPHYSICS in engineering application is the availability of the built in function in each application mode. In this work, the Convection-Diffusion Mode applied in solving the diffusion of reactants. Diffusion processes are governed by the same equation as heat transfer. Thus, the generic diffusion equation has the same structure as the heat equation as shown in equation (6). The diffusion process can be anisotropic, in which case D is a matrix. The diffusion equation is modeled in the Diffusion application mode. The boundary conditions for diffusion mode are shown as in Table 2.
Figure 8: Solve Mode for Adiabatic Plug Flow Reactor

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

2.5 Effects of Various Initial Concentration of Methyl Acetate on Conversion of Reactants Various concentrations of Methyl Acetate and Sodium Hydroxide were used to study the effect of different concentrations to conversion of reactants along the length of the plug flow reactor. The initial values of some constant that were used for the modeling is shown in Table 3.

From Figure 9, it is observed that as initial concentration of reactant increases, the conversion of the reactant increases, the conversion of methyl acetate to methanol also found to increase up to approximately 95%.

2.6 Effects of Various Flow Rates of Methyl Acetate on Conversion of Reactants Various flow rates of Methyl Acetate and Sodium Hydroxide were used to study the effect of different flow rate to conversion of methyl acetate along the length of plug flow reactor as shown.

Constants Parameters Inlet Temperature, To (K) Total Volumetric Flow Rate , Vo (m /s) Heat of Reaction, Hrx (J/mol.K) Thermal Conductivity, k (W/m.K) Radius, R (m) Density, (kg/m ) Special Heat ( J/mol.K) Reaction Rate Constant, k ( mol.s/m3)-1 Maximum Velocity (m/s) Length, L
3 3

Value 298 0.0005 -80929 0.649 0.1 1510 152 2.9E-3 Vo/(3.1R2) 1.4

Table 3: Values for Constants Parameters

When the initial concentration of reactants increased or decreases conversion profile of reactant to products along the length of plug flow reactor would also be affected. Figure 8 shows the effect of varying the initial concentration of reactants on conversion profile.

Figure 10: Effect of flow rate on the conversion of Methyl Acetate and Sodium Hydroxide

Figure 10 shows that changes in initial flow rate will affect the conversion of reactants to the products. Increased flowrate results in increase value of conversion up to approximately 95%. The effect of varying flowrate was apparent up to reactor length of 0.4m. Beyond this; varying flowrate does not affect the conversion obtained.

Figure 9: Effect of Initial Concentration of Reactants to Conversion

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

Conversion Vs Length Bt Modeling and Experiment

1

0.9

0.8

whether the students have gained and benefited in solving and understanding complex equations concepts related to multiple reactions. Five case studies were explored as listed in the followings;
Modeling Experiment

0.7

0.6 Conversion

Fixed bed reactor hydrocarbon oxidation

for

catalytic

0.5

0.4

0 0.2 0.4 0.6 0.8 Length (m) 1 1.2 1.4 1.6

0.3

Non isothermal reactor with isothermal cooling jacket Porous reactor with injection needle Radial effects in tubular reactor : Isothermal and Adiabatic reactor Monolithic reactor

0.2

0.1

Figure 11: Comparison of the conversion of Methyl Acetate and Sodium Hydroxide between modeling and empirical method

Figure 11 shows the comparison between modeling and experiment data. Similar profiles of methyl acetate conversion were observed between experimental and modeling method. Both approaches give 95% of reaction final conversion. Hence it can be concluded that finite element method using FEMLAB can produce the conversion and temperature profiles along the plug flow reactor effectively.

The survey covered three main aspects of questions as listed in the following paragraphs. Rating of 1 to 7 was allocated in which 1 demonstrates strong disagreement, 4 indicates neutral and 7 shows strongly agree. 3.1 Reaction engineering understanding with COMSOL The first section of the survey required the students to answer three questions related to their understanding of reaction engineering course by using COMSOL. The survey results are displayed in Figure 12. 69% of the students rated 5 to 7 indicating that they got better understanding of reaction engineering course by using COMSOL. 71% agreed that the COMSOL case studies can provide them a better link between reaction engineering theories and practical. 59% believed that they were able to cope with their reaction engineering course with the used of COMSOL. This survey strongly suggested that the COMSOL software with its case studies is an efficient tool to be used in reaction engineering course. The students can learn more from the simulation work and relate the results to the theory that they have learned during the lectures. This in turn is a good learning experience for the students to quickly see and predict the complex reaction behaviour from the mathematical model developed throughout the reaction engineering course.

The optimum conversion of methyl acetate to produce liquid phase methanol is found the be 95% at a temperature of 331K with reaction condition of RCOR:OH = 1:1 and pressure of 1 atm using a laboratory scale for both modeling and experiment values. From this project it can be concluded that COMSOL MULTIPHYSICS can effectively predict both the concentration and conversion profile of reactant along plug flow reactor for both adiabatic and non-adiabatic operating conditions. In addition, visualization of concentration profile of reactants and product within the reactor is possible using COMSOL MULTIPHYSICS.

3. Case Study 3
Survey on students view regarding Comsol Multiphysics applicability A survey was conducted recently for 117 students enrolled in CAB 2074 Reaction Engineering course semester July 2007. The students were given assignments on simulation and modeling using the software as part of their laboratory work. The survey intended to seek

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

engineering course. The results can be referred from Figure 13. 79% of the students rated 5 to 7, agreed that the kinetics of reaction engineering can be easily solved throughout the simulation path provided by the software. 77% of the students further agreed that they can easily interpret the results that they obtained from the simulation. 82% of the students demonstrated it was easy for them to construct or define the parameters required for the simulation. All of these clearly indicated that COMSOL is userfriendly software, easy to be learned and it provides a practical approach towards understanding the reaction engineering course. However, not all students can solve the case study by themselves in which only 39% can solve the case study without or minimum supervision by the tutors. 25% of them were unsure (rated 4), and the remainders did require assistants from the tutors. It is expected that if there are continuous exercises on this software, the student can easily solve any simulation work related to the reaction engineering course.

Figure 12: Results of students survey on the understanding of reaction engineering with the used of COMSOL

3.2 User-friendly aspect of COMSOL The second section of the survey consisted of four questions related to the user-friendly element of COMSOL software for reaction

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

Figure 13: Students view on the user-friendly element of COMSOL in reaction engineering course

3.3 Motivation for future application of COMSOL in other chemical engineering courses The final section of the survey posed three questions related to the students motivation for future use of this software in chemical engineering courses. 84 % of the students rated from 5 to 7 showing their interest to learn more on COMSOL for reaction engineering course. 78% rated 5 to 7 demonstrating their interest to use COMSOL in other courses and 69% already had an awareness that this software can be extended for other applications. It is clear that from these survey results, this software is an interesting tool to educate the students to gain a deep understanding of the subject matter. The students are very motivated to see what else can be explored from COMSOL. The results are plotted as in Figure 14.

Figure 14: Students Motivation for future application of COMSOL in other chemical engineering courses

4. Conclusions
The three case studies presented here indicated that there were an extensive works that have

Excerpt from the Proceedings of the COMSOL Users Conference 2007 Kuala Lumpur

been carried out in the area of reaction engineering within the Chemical Engineering department, Universiti Teknologi PETRONAS. However, there are yet so many areas in chemical engineering that can be explored and it has been proven that COMSOL software is an effective simulation tool to assist the researchers and students in their learning.

[11] R. Bryon Bird, Warren E. Stewart, Edwin N., Lightfoot, Transport Phenomena, 2nd Edition, John Wiley & Sons, Inc, 2002.

6. Acknowledgements
The author would like to acknowledge Universiti Teknologi PETRONAS for providing the computing and experimental facilities that enable the work to be carried out.

5. References
[1] Gary R. Martin, Heat-flux imbalances in fired heaters cause operating problems, Hydrocarbon Processing, 1998, pp. 1-7. [2] Nguyen Duy Vinh, Industrial Internship Final Report June 2005. [3] Y.Suyadal, Gas temperature profiles at different flow rates and heating rates suffice to estimate kinetic parameters for fluidized bed combustion, Experimental Thermal and Fluid Science, 2006, [4] George J. Antos, Abdullah M. Aitani, Jose M. Parera,, Catalytic Naphtha Reforming, Science and Technology, Marcel Dekker, Inc. 1995, pp. 231-236. [5] G.D.Stefanidis,B.Merci,G.J.,Heynderickx, G.B. Marin, CFD simulations of steam cracking furnaces using detailed combustion mechanisms., Computers and Chemical Engineering, 30, 2006, pp. 635-649. [6] Francis A. Carey, Organic Chemistry, 4th Edition, McGraw Hill, 1990. [7] Weifang Yu, K. Hidajat and Ajay K.Ray, Determination of adsorption and kinetic parameters for methyl acetate esterification and hydrolysis reactiojn catalyzed by Amberlyst 15, Applied Catalyst A, pg 1-22. [8] Robert H. Perry, Don W. Green, Perry Chemical Engineerings Handbook, 7th Edition, McGraw Hill. [9] Octave Levenspiel, Chemical Reaction Engineering, 3rd edition, John Wiley & Sons, Inc., 1990. [10] Fogler, H. S., Elements of Chemical Reaction Engineering, Prentice Hall, 1999.

7. Appendix
Symbol Q proc Tout Tin qav Do Di lamda Tbulkmax Description Process side heater duty, [kW] Process outlet temperature of heater, [oC] Process inlet temperature of heater, [oC] Average heat flux to heater tubes, [W/ m2 ] Outside tube diameter, [ m ] Inside tube diameter, [ m ] Thermal conductivity tube metal, [W/ m.K] Bulk temperature at the location of the maximum tube skin temperature, [oC] Maximum tube skin temperature in the heater, [oC] Velocity field [ms-1] Table 4: Nomenclatures

Tskinmax u