A hands on guide to using UniSim ThermoWorkbench

Solutions Tree UniSim ThermoWorkbench

Topic #: 4238-1108 Date Created: 05/03/2012 Last Modified Since: 17/09/2014 Viewed: 154

Summary

A hands on guide to using UniSim ThermoWorkbench

More Information

This solution is a tutorial example to demonstrate how to perform fluid phase data regression using UniSim
ThermoWorkbench. The regressed data is also verified with UniSim Design. Fluid phase data regression
allows users to tune fluid package binary interaction parameters (binary coefficients) so that vapour-liquid
equilibrium calculations match with measured experimental data.

The objective of the current solution is to provide a hands-on guide demonstrating the powerful
thermodynamic data regression features offered in UniSim ThermoWorkbench and how tuned data can be
used with UniSim Design.

Introduction

This case considers a set of vapour-liquid equilibrium calculations in UniSim Design using the NRTL activity
coefficient model, for a binary mixture of 10 mol % n-B-Acetate and 90 mol % Methanol. In this specific case,
the results from UniSim Design using the default fluid package parameters were found not to be in
agreement with some specific laboratory data (Table 1).

Mole fr. Mole fr. Mole fr. Mole fr.

Methanol Methanol Methanol Methanol
in liquid in liquid in vapor in vapor
USD USD
Temperature Pressure Actual Actual
Calculated Calculated
(K) (kPa) Lab Data Lab Data
Value Value
296.60 11.81 0.760 0.540 0.997 0.910
313.10 27.40 0.713 0.614 0.991 0.920
333.10 78.45 0.834 0.900 0.984 0.966

Experimental data for this binary pair was found in the literature (DECHEMA) and is presented in Table 2.
The data is in the form of PXY data, where P is the pressure in kPa, x is the mole fraction of methanol in
liquid phase and y is the mole fraction of methanol in vapor phase.

T=333.15 K T=296.6 K T=313.1 K

P x y P x y P x
9.319 0 0 1.52 0 0 3.52 0
24.771 0.076 0.65 3.653 0.05 0.6 12.132 0.106
 34.664 0.145 0.752 5.893 0.106 0.751 14.132 0.1
44.863 0.251 0.827 7.813 0.2 0.833 17.305 0.2
49.423 0.306 0.852 10.866 0.403 0.898 20.905 0.2
58.955 0.478 0.893 11.812 0.54 0.91 22.771 0.3
64.701 0.606 0.912 13.252 0.745 0.943 24.758 0.4
70.021 0.718 0.935 13.959 0.831 0.957 25.958 0.5
73.314 0.793 0.941 15.132 0.948 0.986 27.064 0.5
78.447 0.9 0.966 15.465 1 1 27.398 0.6
84.366 1 1 -- -- -- 28.971 0.7
-- -- -- -- -- -- 29.784 0.7
-- -- -- -- -- -- 30.984 0.8
-- -- -- -- -- -- 31.597 0.8

Regression procedure

The step by step procedure for a fluid phase data regression technique using UniSim ThermoWorkbench is
outlined below:

1. Open UniSim ThermoWorkbench and select the Fluid Package Manager from the Managers
menu (Fig - 1).

2. Click the Add Fluid Package button (Fig. -2). The fluid package selection window will appear.
3. Select the Property Package tab (Fig. -3).

Select the Liquid radio button and select NRTL from the list for the liquid phase property model.
Similarly, select the Vapour radio button and choose Ideal Gas for the vapour phase model. Note
that any combination of liquid and vapor models can be selected but it is the responsibility of the
user to ensure the chosen models are suitable for the particular components and conditions.
4. Select the Components tab and add the two components n-B-Acetate and Methanol. This is
similar to adding components in UniSim Design. At this point the default fluid package name
Fluid1 can be changed if desired.
5. Select Thermodynamic WorkBench Manager from the Managers menu.
6. Select FluidPhaseRegression from the list and click Add. The Fluid Phase Regression window
appears (Fig.-4).
 7. Select the fluid package name Fluid1 using the Fluid Package Selection dropdown list.
8. Click the Add button from the Dataset Selection box. The first dataset, Dataset1, will open
(Fig. -5). Note UniSim ThermoWorkbench does not allow changing the dataset name.

9. In this tutorial we have only PXY data. So select PXY from Type dropdown list and choose the
appropriate units (mole fraction in this case) from the Basis dropdown list.
10. Now we need to select the objective function. This selection is important because the choice of
objective function will control the variables to which the error function will be minimized. For
example, one might want to select the vapour composition optimization function (VapourComp) t
minimize with respect to measured vapour composition, or relative volatility (RelVolatility) to
minimize with respect to relative volatility, and so on. For further details refer to the UniSim
ThermoWorkbench documentation.

In this case we shall select Activity Coefficient as the objective function. This is the default
function for activity models like NRTL and is very useful when regressing data reported at infinite
dilution.

Note that the same objective function can be accessed from two different locations: from the
Fluid Phase Regression view, Summary tab and from the Fluid Phase Dataset view, Basic Data
tab. Note also that the available objective functions in the dropdown list depend on the dataset
type.
11. Now the first set of laboratory data is entered in Dataset1 (Fig -6).
 Similarly the second and third datasets are entered from Table 2.
12. The next step is to set the degrees of freedom of the binary interaction parameters. Go to the
Variables tab of the Fluid Phase Regression view and select the Degrees of Freedom radio
button (Fig.-7).

13. In the Interaction Parameters table, select the cell containing the interaction parameters and
select the desired status (Locked, Glued or Free). For this tutorial, we shall use Free which
allows the optimizer to freely change the values of the interaction parameters independently.
14. Now we are ready to invoke the optimizer. Click the Optimize button located at the bottom of
Fluid Phase Regression view and the interaction parameters will be optimized to match the i
data.
15. The results with the new optimized interaction parameters can be checked by performing flas
calculations in UniSim ThermoWorkbench by defining a stream, using the Stream Manager f
the Managers menu.
16. Additionally, the fluid package data can also be transferred to UniSim Design by clicking the
Export to UniSim Design icon on the fluid package manager window ( ). After exporting t
data to UniSim Design the vapour liquid equilibrium is calculated again and this time it is fou
that the calculated data are in close agreement with the laboratory data (Table-3).

Mole fr. Mole fr. Mole fr.

Methanol Methanol Methanol
in liquid in liquid in vapor
USD USD
Temperature Pressure Actual
Calculated Calculated
(K) (kPa) Lab Data
Value Value
296.60 11.81 0.529 0.540 0.919
 313.10 27.40 0.575 0.614 0.921
333.10 78.45 0.883 0.900 0.967
Note that before finally exporting and using the data in UniSim Design, it is highly recomm
to examine the calculated and lab data using the Errors and Error Plots tabs in the Fluid
Regression window. In this case it was found that the calculated data and the literature d
closely matched. It is worth mentioning here that UniSimThermoWorkbench can also be
pure component regression as well as for phase equilibrium calculation. The Pure Comp
Regression operation allows the user to create new components or modify existing ones
specific process considerations. Coefficients for more than twenty-five thermo physical p
can be regressed using a wide range of equation forms. The Phase Equilibrium operatio
one to assess the quality of the thermodynamic model that has been selected to represe
mixture. The operation enables the visualization of key properties for binary and ternary

Results and conclusions

In this tutorial, vapour liquid equilibrium data of n-B-Acetate and Methanol are optimized using t
regression functionality in UniSim ThermoWorkbench. It was found that results obtained using r
interaction parameters closely matches with the original laboratory data.

Keywords regression, BIPs

Type Hands on
Product UniSim ThermoWorkbench