SIMULATION OF BATCH- DISTILLATION OF METHANOLWATER SYSTEM

Simulating the behavior of batch distillation column in the separation of methanol water mixture, using ANN and ChemCAD.

Presented by:
Satadru Chakrabarty (09uch013) Sayantan Roy Choudhury (09uch014) Tapas Saha (09uch016) 2012-13

NAME OF PROJECT:
Batch distillation of methanol-water system
and modeling and simulation of distillation column using ARTIFICIAL NEURAL NETWORK and ChemCAD

Table of contents:
Preface3 Acknowledgement.4 Literature review.5 Brief description about the project7 Batch distillation
o o o o o o o o o o o Introduction9 Theory.10 Calibration curve.12 Description of apparatus13 Utilities13 Experimental procedure.14 Observation & calculation.15 Introduction 17 What is Artificial Neural Network? ........................18 Use of a.n.n in the project19 Procedure..22

Artificial Neural Network

ChemCAD
o Overview of batch distillation..23 o Output from chemCAD...27

Conclusion.31

PREFACE
Technology is any technique, instrument or device that makes human life on this planet easier. This gives me immense pleasure to be working on a project that deals with a very important phenomena or unit operation of the industry. This project Batch distillation of methanol-water system and simulation of working of distillation column using ARTIFICIAL NEURAL NETWORK and ChemCADis a humble step in trying to understand and design a fully operational batch distillation column, with the help of software. The mixture chosen for the simulation is basically methanol- water mixture, which is a very common mixture found in industries and is usually separated by a distillation column. In this project there would be wide scale use of Artificial Neural Network (ANN), basically for the purpose of data prediction and data validation. Finally the simulation and design of the distillation column would be completed using ChemCAD, with the data acquired from the initial experiments and data prediction from ANN. Hope this humble effort would manage to earn the appreciation of all the readers.
Sayantan Roy Choudhury Satadru Chakrabarty Tapas Saha

ACKNOWLEDGEMENT
We would like to express my sincerest gratitude to our Director Prof. Dr. P.K.Bose, who has been kind enough to provide us with the infrastructural facilities that we have in our labs. Then we would like to give our sincere regards to Mrs. Soma Nag (H.O.D, Dept. of Chemical Engg.), without whose help and support this project would not have been possible.

Next we would like to give regards to our project guide Prof. Bibhab Lodh (Asst. professor, dept. of chemical engg.) who has been so patient with us, and always giving his advice for the betterment of the project and correcting our errors whenever necessary.

We are also grateful to all the teachers of the dept. of chemical engg. for their valuable advice and insights. We are also grateful to each and every person who has helped us in the project with their valuable supports and advices.

LITERATURE REVIEW:

Barton, P., and E. C. Roche, Jr., Batch Distillation, Section 1.3, in Schweitzer, P. A. (Ed.), Handbook of Separation Techniques for Chemical Engineers, 3rd ed., McGraw-Hill, New York, 1997. Carey, J. S., Distillation, in J. H. Perry (Ed.), Chemical Engineers Handbook, 3rd ed., McGraw-Hill, New York, 1950, pp. 582-585. Davies, J. T., Chemical Engineering: How Did It Begin and Develop? in W. F. Furter (Ed.), History of Chemical Engineering, Washington, D.C., American Chemical Society, Adv. Chem. Ser, 190, 1543 (1980). Diwekar, U. W., Batch Distillation: Simulation, Optimal Design and Control, Taylor and Francis, Washington, D.C., 1995. Ellerbe, R. W., Batch Distillation, in P. A. Schweitzer (Ed.), Handbook of Separation Techniques for Chemical Engineers, McGraw-Hill, New York, 1979, p. 1.147. Gentilcore, M. J., Reduce Solvent Usage in Batch Distillation, Chem. Engr. Progress, 98(1), 56 (Jan. 2002). Luyben, W. L., Some Practical Aspects of Optimal Batch Distillation, Ind. Eng. Chem. Process Des. Develop. 10, 54 (1971). Pratt, H. R. C., Countercurrent Separation Processes, Elsevier, New York, 1967. Rayleigh, Lord (J. Strutt), On the Distillation of Binary Mixtures, Phil. Mag. [vi], 4(23), 521 (1902). Robinson, C. S., and E. R. Gilliland, Elements of Fractional Distillation, 4th ed., McGraw-Hill, New York, 1950, Chaps. 6 and 16. R T, Distillation through the Ages, Chem. Heritage, 25 (2), 40 (Summer 2007).

Treybal, R. E., Mass-Transfer Operations, 3rd ed., McGraw-Hill, New York, 1980. Treybal, R. E., A Simple Method for Batch Distillation, Chem. Eng., 77(21), 95 (Oct. 5, 1970).

BRIEF DESCRIPTION ABOUT THE PROJECT

The aim of our project is to simulate the behavior of a batch distillation column using ARTIFICIAL NEURAL NETWORK and ChemCAD. So the project would be based on the 3 main parts: Distillation column Artificial Neural Network ChemCAD The first stage of the project includes conduction of series of experiments in the batch distillation column of the mass transfer lab to get the initial readings for our experiments. The behavior of the column will give us an initial idea about the behavior of a real column and all the constraints relating to its operation. After getting the initial data from the experiments, we move to the second phase of our operation. Here we predict some extra sets of data with the help of Artificial Neural Network (ANN). The initial data which we got from the experiment are used to train the ANN software. After that the software is used to predict the data for unknown variables. Here the prediction of data comes with a certain degree of error. So, care must be taken to keep the error minimum. Hence data validation is very important.

After that we come to the third and final stage of our project. In this phase we use all the experimental data and validation data from ANN and input it into the ChemCAD simulator of distillation column. In this software we are able to completely design the column and also estimate its efficiency. The huge amount of predicted data helps us to make a regression analysis of the data and predict the true relation between the temperature and mole fraction of the components

BATCH DISTILLATION

INTRODUCTION: Batch Distillation is often preferred to continuous distillation in cases where relatively small quantities of material are to be handled at irregularly scheduled periods. The simplest case of batch distillation is one in which the material to be separated is charged to a heated kettle fitted with a total condenser and product receiver. The material is distilled without reflux until a definite quantity of one of the components of the mixture has been recovered or until a definite change in composition of the still contents has been effected. In all types of batch distillation, a quantity of feed is charged to a still pot, or kettle, and heat is applied to it. The vapor which is usually passed through a fractionating column is then condensed giving the overhead product while a less volatile residue remains in the kettle at the end of the distillation. Continuous distillation is a steady state process because once equilibrium has been attained, conditions at any given point remains constant whereas batch distillation is an unsteady state process the concentration of the more volatile component decreasing continually so that the temperature and composition of the mixture at a point in the system must alter as the distillation proceeds.

THEORY:
Batch distillation with only a single still does not give a good separation unless the relative volatility is very high. To obtain product with a narrow composition range, a rectifying batch still is used that consists of a reboiler, a rectifying column, a condenser, some means of splitting of a portion of condensed vapor or distillate as reflux and the receiver. The operation of a batch still and column can be analysed using the same operating line equation as for the rectifying section of the continuous distillation.

Yn+1=

n+

XD (1)

For the binary system: y2= 1- y1x2= 1- x1

Since the slope of the operating line is R/(R+1), the slope increases as the reflux increases, until when reflux is infinite. Under total reflux slope is 1. The operating line then coincides with the diagonal. The number of plate is minimum at the total reflux. Minimum number of plates required can be calculated from the terminal concentration of xb and xd based on the relative volatility of the components , which is defined in terms of equilibrium concentrations:

.. (2)

an ideal mixture follows Raoults law and the relative volatility is the ratio of vapor pressure. Thus: p1 = P1x1
10

p2 = P2x2 y1 = p1/P . (3) y2 = p2/P therefore = P1/P2 (4) the final ratio does not change much over the range of temperature encountered in a typical column, so the relative volatility is taken as constant.

11

CALIBRATION CURVE
Prepare a calibration curve for Me OH- Water by plotting RI as a function of mole fraction Me OH at the ambient temperature 25C. Prepare different mixtures of Me OH + Water by volume and measure the RI of each mixture. Convert your volume fractions and plot mole fraction Me OH vs. RI on a simple graph. Mole% Me OH 0 10 20 30 40 50 60 R.I. at 25C 1.3321 1.3363 1.3394 1.3411 1.3411 1.3410 1.3398 Mole% Me OH 70 80 90 92 94 96 100 R.I. at 25C 1.3381 1.3356 1.3330 1.3300 1.3286 1.3280 1.3268

12

DESCRIPTION OF APPARATUS
The column is made of stainless steel material packed with borosilicate glass rasching rings. An electrically heated reboiler is installed at the bottom of the column. The bottom product is collected in the tank. The vapours form the top of column are condensed in the shell and tube type condenser by circulating cooling water, supplied by laboratory overhead tank. The condensate is divided into reflux and distillate by automatic reflux divider and R/D ratio can be varied. Reflux is fed back to the column and distillate is received in a receiving tank. The complete column is insulated for minimizing the heat loss.

UTILITIES
1. 2. 3. 4. 5. 6. 50 l of methanol Electricity Supply: single phase, 220V AC, 6KW with earth connection. Water supply: 2LPM at 5m head Floor drain required Required chemicals Refractometer for analysis.

13

EXPERIMENTAL PROCEDURE
1. Connect the cooling water supply to setup. 2. Fill the reboiler with Methanol-water solution. The total amount of solution should not be less than 15lts. The composition of should be in range of 1525% of methanol by volume. 3. Set a process temperature for the process using the digital temperature controller. The temperature should be in range of 85-95C 4. Start the heaters and cooling supply. 5. Adjust the cooling water flow rate to a moderate value. 6. Set the cyclic timer for total reflux. 7. Wait for 25-30 min for the system to achieve steady state. 8. Now take the samples from the bottom and distillate stream. 9. Cool down the samples to room temperature and measure RI 10.Now adjust the cyclic timer to a desired reflux ratio and wait for 5min 11.Now take out the samples from both distillate and bottom product. 12.Cool down the samples and measure RI. 13.The experiment can be repeated with different set point temperatures and reflux ratios.

14

OBSERVATION & CALCULATION

DATA:
= 3.32 (for Me OH water system)

Tc
100 96.4 93.5 91.2 89.3 87.7 84.4 81.7 78.0 75.3 73.1 71.2 69.3 67.5 66.0 65.0 64.5

x
0 0.02 0.04 0.06 0.08 0.1 0.15 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95 1

y
0 0.134 0.23 0.304 0.365 0.418 0.517 0.579 0.665 0.729 0.779 0.825 0.87 0.915 0.958 0.979 1

15

OBSERVATION: Tf = Rif =. RID= RIB= R = . CALCULATION (from calibration curve): xf = xD= xB= .
We can calculate it for different sets of values.We change the set point temperature and get various values.

16

ARTIFICIAL NEURAL NETWORK

INTRODUCTION

In our project, the artificial neural network plays a very important role. This helps us to predict the different values of data which we could not determine from our experiments. For creating a simulation of a BATCH DISTILLATION column we will require a huge number of data. But due to certain constraints, we are not able to conduct such huge number of calculations. This is where the ANN tool comes in handy. Using this tool we will be able to predict as many data values as required for our simulation, with a certain degree of approximation. For this we need to know a little about ARTIFICIAL NEURAL NETWORK itself.

17

WHAT IS ARTIFICIAL NEURAL NETWORK?

The ARTIFICIAL NEURAL NETWORKS (ANN) is nonlinear mapping structures based on the functioning of the human brain. They are powerful tools of modeling especially when the underlying data relationship is not known. The ANN can learn correlated patterns between input data set and corresponding target value. After training, ANN can be used to predict the outcome of new independent input data. The ANN imitates the learning process of the human brain and can process problems involving nonlinear or complex data even if the data is imprecise and noisy. Thus they are suited for the modeling of engineering data (in this case, distillation data), which are known to be complex and often nonlinear. The ANN has a huge capacity in the predictive modeling, i.e. we can predict any number of data as required, and then prediction of our system behavior is guaranteed. An ANN is a computational structure which is inspired by observed process in natural networks of biological neurons in the brain. It consists of simple computational units called neurons, which are highly interconnected. ANNs have become the focus of much attention, largely because of their wide range of applicability and the ease with which they can treat complex problems. ANNs are parallel computational models comprised of densely interconnected adaptive processing units. These networks are finely grained parallel implementation of non-linear static or dynamic systems. A very important feature of this system is its adaptive nature, where learning by example replaces programming in problem solving. This feature makes computational models very appealing in application domain where one has little or incomplete knowledge of the problem to be solved but where training data is readily available. ANN is being used in the applications & problems where statistical methods like discriminant analysis, regression etc were used. It is therefore time to realize ANN as a powerful tool of data analysis.

18

USE OF ANN IN THE PROJECT:

The software that we are going to use for this project is known as justNN. This is a freeware and very effective ANN software. The process of using the software is described as under: o DATA ENTRY AND TRAINING: first of all tables are created in the software and the input and output rows are designated. Then the data (got from experiments) are fed into the system. Then the system is trained to understand and find a correlation between the input and output.

Fig: this figure shows the softwares data entry sheet

19

o DATA PREDICTION AND FURTHER LEARNING: In this step we insert a query row and for different conditions of input, the ANN gives us suitable outputs. The ANN is then made to learn the new values that we have predicted. It must be kept in mind that the error range changes with every new stage of data prediction and learning. The software validates whether the given input in sync with the correlations that it had created and based on that gives the output.

Fig: the internal network of ANN, correlating the output and input parameters

20

o ERROR ANALYSIS: the error analysis function of the software allows us to check the amount of error that has occurred during the data prediction and further validation. The software allows us to set a certain range of error so that the predicted output falls within that range of approximation. After the learning process is complete, the software gives us a graph which shows us the amount of error that is present in the prediction. It must be noted that with increments in the number of prediction and validation, the amount of average error is reduced.

Fig:the error graph, showing the target error, maximum, minimum and average error values

21

PROCEDURE
The input and output rows are defined in the software in the data entry page. The input and output data, from the experiment, are entered in their respective rows. Then the software is made to learn these values. After series of iterations, the software creates a correlation between the input and output. Now query rows are inserted in the table. The input variables are inserted and the software predicts the output. Then the values predicted are learnt. The last two steps are repeated till we have the required amount of data for our simulation. Then the data is recorded up in a spreadsheet.

22

ChemCAD

Overview of the Batch Distillation Process

The process of building the flowsheet and simulating the batch distillation involves the following steps: o o o o o o o o o o Create a new simulation. Select engineering units. Draw the flowsheet. Select the components. Select thermodynamic options. Specify pot charge. Specify the distillation column. Define operating steps. Run the simulation. Review the results and print as needed.

Creating a New Simulation

Start by creating a new simulation and giving it a name. To do this, launch CHEMCAD and then Select File >Save to open the Save As dialog box. Navigate to the directory where you want to store the simulation (try MySimulations, located under My Documents) and give your simulation a name, leaving the type as CHEMCAD 6 (*.cc6). Then click Save to create the file and return to the main CHEMCAD window.

23

 Selecting Engineering Units

Select Format >Engineering Units to open the Engineering Unit Selection dialog box. The English units option is the default and is currently highlighted. To change the engineering units system, you would click the Alt SI, SI, or Metric button; you could then change any of the individual units as well. For our project we will be using the S.I. unit. So we will select it.

Drawing the Flowsheet

Creating a flowsheet is a matter of placing UnitOp icons on the screen, connecting them with streams, and then adding various graphical objects to enhance the drawing. We will finally come up with a flowsheet of this manner:

Fig: flowsheet of batch distillation column as displayed by ChemCAD

24

 Selecting Components
Now you need to identify the components to be used in this simulation. Start by selecting Thermophysical >Select Components. For this example, youll choose components from the standard CHEMCAD database. In the Select Components dialog box, find and add each needed component. From the available component section select the following components: a) Methanol b) Water Then add the components to the selected components section. Press ok

Thermodynamic Options
As soon as we have finished component selection, the Thermodynamics Wizard appears. This tool can suggest thermodynamics options to use with this simulation. Here we keep all the other constraints unchanged, just the global K option is changed to Peng-Robinson

Specifying Pot Charge

Now that we have thermodynamics and components defined for this simulation and we have a batch column in the flowsheet, we can define the pot charge for the column. We will specify the pot charge according to the following rules: The Temp (C), Pressure(psia), Vapor Fraction, and Enthalpy (MMBtu/h)fields arethe thermodynamic properties of the charge. We specify the temperature, pressure, totalflow and molefraction of methanol and water entering into the system. And press on the FLASH button. This will automatically calculate the enthalpy. In this case the mole fraction is kept 0.5 for both.

25

 Specify the distillation column. We will specify the distillation column details according to our condition. There the only fixed constraint will be the number of stages which we will take as 3. The other constraints like pressure etc will be fixed after we have the initial conditions from the experimental setup.

Defining the Operating Steps

When you have completed the initial column specification, the Batch Operation Parameters dialog box appears. Here we will set the reflux ratio for the process. During our simulation we will take up different values of reflux ratio and simulate the column operation.

Run the simulation. In this step we run the simulation for the column that we have designed. To run the simulation, click the Run All toolbar button. If there are errors that will be shown. So steps would be taken to correct those accordingly.

Review the results and print as needed

Once the simulation is complete, we can review the results interactively before printing a hard copy. The commands needed to do this are located in the Report andPlot menus. From this menu we can get all the required outputs. The following pages show a few outputs for a simple trial run that we had done to test the software.

26

OUTPUTS FROM chemCAD

PROCESS FLOW DIAGRAM OF THE BATCH DISTILLATION COLUMN

The process diagram has 3 main constituents: Unit 1: the distillation column with condenser and reboiler. Unit 2: process stream input containing methanol and water Unit 3: process stream output containing methanol. Unit 4:process stream output containing water.

27

DETAILS OF THE COLUMN: This page gives us stage wise details. We can know the temperature and pressure drop at every stage from this chart.

28

TEMPERATURE PROFILE AND HEAT DUTY: From this chart we can know the temperature, heat duty and reflux data of the column

29

MOLE FRACTION DATA: This gives the mole fraction of the components with the variation in the temperatures.

30

CONCLUSION

This project has been taken up solely as a teaching module that will help in the better understanding of the separation of the methanol water system in batch distillation column. This project can also be instrumental for small units in industries. A good amount of time and effort has been given in making this project and we hope that in the upcoming times our efforts will bear fruits.

31