modelos termodinamicos de aspen plus.pdf

8/20/2019 Modelos Termodinamicos de Aspen Plus.pdf

1/12

S U C C E E D I N G A T S I M U L A T I O N

CHEMICAL ENGINEERING PROGRESS •OCTOBER 1996 •35

Chemical engineers use process

simulation to perform a varietyof important work. This work ranges from calculations of

mass- and energy balances of large flow-sheets to prediction of the performance of process alternatives that can save millionsof dollars. An engineer very quickly candefine a complex flowsheet and all theprocess conditions. Desktop computersnow allow rating, sizing, optimization,and dynamic calculations that previouslyrequired large mainframe computers. Inthe past, these simulations were oftenbuilt by a group of experts, including aphysical property expert. Now, simulatorssuch as ASPEN PLUS, ChemCAD III,HYSIM, PRO II, and SPEEDUP are easi-er to use and more powerful than thestandalone programs of the past. Today, asingle engineer can set up the basic simu-lation specifications, including the physi-cal properties, in very little time.

Missing or inadequate physical prop-erties, however, can undermine the accu-racy of a model or even prevent you fromperforming the simulation. That some re-quired information is missing is not an

oversight in the simulator. After all, formost compounds, physical property pa-rameters are not known for every thermo-dynamic model or for all temperature orpressure ranges. Models have built-in as-sumptions and practical limits that shouldapply.

In this article, we will provide practi-cal tips and techniques to help you accu-rately describe the physical propertiesneeded in a simulation. As an engineer,

you always will have to make assump-

tions in terms of physical properties,however. The goal of this article is to out-line the appropriate assumptions and toprovide techniques when properties aremissing.

The five important t asksSuccessfully describing the physical

properties to be used in a simulation in-volves five tasks:

1. selecting the appropriate physicalproperty methods;

2. validating the physical properties;

3. describing nondatabank compo-nents (chemical species or compound)and missing parameters;

4. obtaining and using physical prop-erty data; and

5. estimating any missing propertyparameters.

It can be argued that these tasks arenot sequential and, to some degree, theyare concurrent. During simulation devel-opment, however, you will need to visiteach area to be confident that your simu-lation is as accurate as possible — sothat important decisions can be made

based on the results of your simulations.

Selecting the appropriatephysical property methods

This essential first step will affect allsubsequent tasks in developing accuratephysical properties in your simulation.Indeed, the choice of the physical proper-ty models for a simulation can be one of the most important decisions for an engi-neer. Several factors need to be consid-

Don’t Gamble WithPhysical PropertiesFor Simulations

Finding good values

for inadequate

or missing physical

property parameters

is the key to a

successful

simulation. And

this depends upon

choosing the right

estimation methods.

Eric C. Carlson,

Aspen Technology, Inc.


2/12

36 • OCTOBER 1996 • CHEMICAL ENGINEERING PROGRESS


ered, and no single method can han-dle all systems. Table 1 lists somethermodynamic models available insimulators.

The four factors that you shouldconsider when choosing propertymethods are:

• the nature of the properties ofinterest;

• the composition of the mixture;• the pressure and temperature

range; and• the availability of parameters.To ease the selection of the right

physical property methods, we sug-gest using the decision trees shown inFigures 1–3. These trees are based on

the four factors for selecting propertymethods, and can be used when thechemical components and approxi-mate temperature and pressure rangesare known. While these diagrams aresimplifications, they do show thebasic steps of the decision-makingprocess, while the notes in the sidebaramplify some of the key points.

The nature of the properties of in-

terest. A question that you may ask yourself when starting a simulation is“Does the choice of physical propertymethods matter?” The answer is anemphatic YES. The choice canstrongly affect the prediction of thesimulation. You should be selecting acollection of methods that will bestpredict the properties or results of in-terest to you.

Because many chemical processsimulations include distillation, strip-ping, or evaporation, one importantpotential consideration for the choiceof physical property models isvapor/liquid equilibrium (VLE). Thisis the area in which the most physical

property work is focused in chemicalengineering. Liquid/liquid equilibri-um (LLE) also becomes important inprocesses such as solvent extractionand extractive distillation.

Another critical consideration ispure-component and mixture en-thalpy. Enthalpies and heat capacitiesare important for unit operations suchas heat exchangers, condensers, dis-tillation columns, and reactors.

Equation-of-State ModelsBenedict-Webb-Rubin(BWR)-Lee-Starling

Hayden-O’Connell*

Hydrogen-fluoride equation of state for

hexamerization*

Ideal gas law*

Lee-Kesler (LK)

Lee-Kesler-Plocker

Peng-Robinson (PR)

Perturbed-Hard-Chain

Predictive SRK

Redlich-Kwong (RK)

Redlich-Kwong-Soave (RKS)

RKS or PR with Wong-Sandler mixing rule

RKS or PR with modified-Huron-Vidal-2 mix-

ing ruleSanchez-Lacombe for polymers

* Not used for the liquid phase.

Activity Coefficient ModelsElectrolyte NRTL

Flory-Huggins

NRTL

Scatchard-Hildebrand

UNIQUAC

UNIFAC

Van Laar

Wilson

Special Models

API sour-water method

Braun K-10

Chao-Seader

Grayson-Streed

Kent-EisenbergSteam Tables

Table 1. Thermodynamic property modelsavailable in a simulator.

All Nonpolar

Pseudo &Real

VacuumBraun K-10 or Ideal

Chao-Seader,Grayson-Streed orBraun K-10

Polar

Real

Non-electolyte

Electolyte

Electolytes

Electolyte NRTL

or Pitzer

See Figure 2

Peng-Robinson,Redlich-Kwong-Soave,Lee-Kesler-Plocker

P?

E?

R?

E?

P?

?

R?

? Polarity

Real orPseudocomponents Pressure

Source: (7 )

Figure 1. The first steps for selecting physical property methods.


3/12


Here are some pointers to help you navigate the decision trees that ap-

pear as Figures 1–3.

What are pseudocomponents? In many applications where only non-

polar molecules are present (such as in hydrocarbon processing and re-

fining), the mixture is so complex that instead of representing it by all the

known constituents, it is easier to group the constituents by some useful

property such as boiling point. In this way, a mixture of hundreds of con-

stituents can be reduced to 30 or fewer. The properties of these grouped

constituents, called pseudocomponents, are represented by an average

boiling point, specific gravity, and molecular weight. If you do not use

pseudocomponents, the constituents should be described by a molecu-

lar formula and are referred to as real components.

Why are electrolyte mixtures different? Electrolyte mixtures include

components that are charged molecules (ions) or that form salts. Some

simulators allow calculation of electrolyte reaction equilibrium with

phase equilibrium. This is a very powerful method and its usage covers

many applications such as caustic scrubbing, neutralization, acid pro-duction, and salt precipitation. The nonideality of electrolyte solutions,

usually containing water, can be observed in boiling point elevation, salt-

ing out of gases (that is, adding salts to the solution to change the solubil-

ity of gases), and salt precipitation. The most common electrolyte meth-

ods are the Pitzer model, and the modified-NRTL activity coefficient

model of Chen and coworkers. Some electrolytes, like formic acid and

acetic acid, are very weak and an electrolyte method is not required.

Which type of method should be chosen for mixtures containing

polar components but no electrolytes? There are two groups of methods

— based on activity coefficients or equations of state. Use activity-coef-

ficient-based methods when pressures are low to medium (typically less

than 10 bar or 150 psia) and if no components are near critical point. Ac-

tivity coefficient models also often are used to accurately predict non-

ideal liquid behavior such as for VLE and for LLE. In contrast, equation-

of-state methods excel in their ability to represent data and extrapolatewith temperature and pressure up to and above the mixture critical

point. Now, however, methods relying on cubic equations of state with

predictive mixing rules effectively combine the strengths of the two

methods. (See Table 2.) For higher pressures (and temperatures), these

special equations of state are better as they were developed to apply to

a wider range of temperatures. These methods incorporate activity coef-

ficients in the calculation of component interactions represented by ex-

cess Gibbs free energy. Most of the latter use a UNIFAC-based activity

coefficient model as the default, but you can use any activity coefficient.

At simulation pressures less than 10 atm and where there are no

near critical components, for the best results use the Wilson, NRTL, or

UNIQUAC binary parameters that may be available in built-in databanks,

or fit binary parameters to experimental data (if available) using activity

coefficient models. These parameters may have been determined at dif-

ferent temperatures, pressures, and compositions than you are simulating,though, so you may not obtain the best possible accuracy. If interaction

parameters are not available, however, you can use the UNIFAC method.

When should UNIFAC be used?UNIFAC and other UNIFAC-based ac-

tivity coefficient models are predictive approaches that use structural

groups to estimate component interactions. From structural information

about organic components usually available in the built-in databank,

UNIFAC is able to predict the activity coefficients as a function of com-

position and temperature. You can make use of UNIFAC when you do not

have experimental data or binary parameters or when an approximate

value is acceptable (for instance, for a component with low priority). In

recent years, there have been improvements to UNIFAC (see Table 3)

that can better predict VLE, heat of mixing, and LLE over a wider temper-

ature range. Recent extensions to UNIFAC proposed for molecules such

as refrigerants and sugars may be useful, and you can add the groups

and parameters to your simulation. Simulators may have the ability to

generate binary interaction parameters for Wilson, UNIQUAC, or NRTL

from UNIFAC.

Not all components can be described using UNIFAC, however, and

not all group interactions are available. Examples of components that do

not have UNIFAC groups include metals, organometals, and phosphates.

So, we highly recommend always doing a search for available data on

binary or ternary systems of interest.

How should the vapor phase be treated? The choice of the VLE

method using an activity coefficient model also requires a choice of

model for the vapor phase properties. If vapor phase association is ob-

served (as in the case of acetic acid), then the vapor phase model

should be Hayden-O’Connell or Nothnagel. A system containing hydro-gen fluoride may require a special model to represent the high degree of

association due to hydrogen bonding. Association in the vapor phase

can have a strong effect on phase equilibria and enthalpy.

When should defaults be overridden for other physical property

methods? Prediction of density, enthalpy, and viscosity also are impor-

tant in simulators, and you shouldn’t automatically accept the default

methods. Check the simulator documentation for the default method and

mixing rules.

Vapor density is calculated by an equation of state or the ideal gas

law. Mixture liquid densities can be calculated by an equation of state, a

temperature-dependent model such as that of Rackett, or by a tempera-

ture- and pressure-dependent model such as the COSTALD. For psuedo-

components, an American Petroleum Institute (API) method typically is

employed. The Rackett model is recommended for general use.

Vapor enthalpy usually is calculated via an ideal gas assumption oran equation of state. The equation-of-state methods calculate a depar-

ture from ideality called the vapor enthalpy departure. For components

such as acetic acid, the Hayden-O’Connell model is best, and will calcu-

late a larger-than-normal vapor enthalpy departure.

Liquid enthalpies are calculated by a variety of methods. If the simu-

lator uses the ideal gas as the reference state, then the pure-component

liquid enthalpy is calculated from the ideal gas enthalpy and a liquid en-

thalpy departure. This can be written as

H *, l = H * , ig + (H *, l - H * , ig ) (1)

where H *,l is the pure-component liquid enthalpy, H *,ig is the ideal gas en-

thalpy, and (H *,l - H *,ig ) is the liquid enthalpy departure. This departure in-

cludes the heat of vaporization, the vapor enthalpy departure from the

ideal pressure to the saturation pressure, and the liquid pressure cor-

rection from the saturation pressure to the real pressure. Simulators

also allow separate calculations for a liquid enthalpy directly from theliquid-heat-capacity polynomial. For some components, the method in

Eq. 1 will not be accurate enough for liquid-heat-capacity predictions.

This can be very important if you are exporting your property information

to another program such as one for rigorous heat-exchanger design. You

can use the latter liquid-heat-capacity (CpL) method to improve the ac-

curacy of liquid heat capacities.

Viscosity is another important property for sizing of piping, pumps,

heat exchangers, and distillation columns. There are various vapor and

liquid methods for calculating viscosity and, generally, the parameter re-

quirements for these methods are substantial.

Navigating the decision trees


4/12



In addition, density, viscosity, pH,and thermal conductivity may be es-sential for other process calculations.Transport properties are importantwhen doing equipment sizing calcula-tions. Also, processes such as metallur-gy and mining will require calculationsfor phase equilibria including solids.

The composition of the mixture.

Composition will influence all proper-ties, due to the way mixture propertiesare calculated. It will affect phaseequilibria greatly because of the inter-action of the components in the mix-ture. Usually, the interaction in the

liquid phase is the more important be-cause of the close proximity of themolecules in that phase. The nature of the vapor phase also can be significantif the components form complexes.The important intermolecular forcesare electrostatic, induction, attraction,and repulsion between nonpolar com-ponents, and chemical forces such ashydrogen bonding. A good overviewof these forces is given in Ref. 1.

P >10 bar

P


5/12


The magnitude of the electrostaticand induction forces is related to thepolarity of the components. Compo-nents such as water, acetone,formaldehyde, and methyl chlorideare strong dipoles. Many polar com-pounds are associative, and formcomplexes or dissociate into ions.Components like ethane and n-hep-tane are nonpolar. You can use your

simulator to report the dipole mo-ments of databank components as onemeasurement of polarity. In general,mixtures of nonpolar componentswill exhibit less nonideal behavior.

Figures 4–7 illustrate the effect of polarity on binary vapor/liquid equi-libria. Figure 4 shows the predictedand experimental VLE of two highlypolar components, acetonitrile and

water, at 1 atm. The azeotrope is ac-curately predicted at approximately0.7 mole fraction of acetonitrile. Fig-ure 5 presents VLE for a mixture of two slightly polar compounds,toluene and phenol, at 1 atm. The de-viation from ideality is shown bycomparing the predicted curve froman ideal liquid assumption to thatfrom a method predicting nonideality

100

95

90

85

80

0 0.2 0.4 0.6 0.8 1

T e m p e r a t u r e , ˚ C

Molefraction Acetonitrile

Source: (14 )

=VaporMolefraction

=Liquid Molefraction

1

0.8

0.6

0.4

0.2

0 0.2 0.4 0.6 0.8 1

V a p o r M o l e f r a c t i o n T o l u e n e

NRTL-RK Ideal Liquid

Liquid Molefraction Toluene

Figure 4 (above). VLE of acetonitrile/water system at 1 atm.

Figure 5 (right). VLE of toluene/phenol system at 1 atm.

81

80.5

80

79.5

79

78.5

78

77.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

T o t a l T e m p e r a t u r e , ˚ C

Vapor Molefraction C6H6Liquid Molefraction C6H6

MolefractionSource: (15 )

16

14

12

10

8

6

4

0 0.2 0.4 0.6 0.8 1

T o t a l P

r e s s u r e , A t m

Vapor Molefraction C2H6Liquid Molefraction C2H6

Molefraction

Figure 6 (above). VLE of cyclohexane/benzene system at 1 atm.

Figure 7 (right). VLE of ethane/propylene system at 40°F.


6/12



(the nonrandom two-liquid activitycoefficient model (NRTL) andRedlich-Kwong equation of state forthe vapor phase). Figure 6 depicts theVLE of a mixture of cyclohexane andbenzene at 1 atm. Here, the interac-tion of seemingly similar moleculeswith a difference in boiling point of less than 1°C causes an azeotrope at acomposition of about 0.54 mole frac-tion of benzene. A mixture such asethane and propylene (Figure 7) is analmost ideal one, and does not deviatemuch from Raoult’s law.

Mixtures of nonpolar and polarcompounds, such as water and hydro-carbons, often will form two liquidphases that are very immiscible. Fig-ures 8 and 9 show examples of misci-ble and immiscible systems of liq-uid/liquid equilibria, respectively, at 1atm. In Figure 8, cyclohexanol is im-miscible in the water phase but theorganic phase contains up to 0.50mole fraction water (0.10 mass frac-tion water). Figure 9 shows the high

degree of immiscibility in both theorganic and water phases for a mix-ture of benzene and water wherethere is less than 0.06% by mole ben-zene (0.3% by mass). Because of thisbehavior, some simulators have a spe-cial property method to treat thewater phase as organic-free (alsocalled Free-Water).

Most simulators offer collectionsof property methods in predefined

sets based upon methods that fre-quently are used for certain types of mixtures. Usually the sets are identi-fied by the method used for phaseequilibria. When these sets use anequation-of-state model, the samemodel is used for many properties, in-cluding those for phase equilibria.

The pressure and temperature

range. This is especially important inchoosing the method to performphase equilibria calculations. Meth-ods that are based on Raoult’s law orthat use activity coefficients are notaccurate at high pressure or when thetemperature is above the critical tem-perature of a component. You can useHenry’s law when you have lightgases in subcritical solvents, but itgenerally is not recommended forconcentrations of solute greater than5%. In general, equations of state arebetter suited to predict VLE over awide temperature or pressure range,especially at high temperature andpressure.

The availability of parameters.Without sufficient pure-componentand binary parameters, you will beunable to calculate pure-component ormixture properties. You must chooseamong obtaining and using experi-mental or literature data, estimatingparameters, or choosing a less rigor-ous method. This should be investigat-ed for all physical property methods in-cluding those shown in Figures 1–3.

Validatingthe physical properties

A necessary step in any simulationproject is validation of the physicalproperties. This involves reporting, tab-ulating, or plotting pure-component andmixture properties and comparing theresults to known data or expected be-havior. This is an important step in anysimulation and should be performed fordatabank as well as nondatabank com-ponents. Simulators can provide thesecalculated properties in tabular and plotformat. This is a useful tool for under-standing how pure-component and mix-ture properties, such as density, heat ca-pacity, and excess properties, vary withtemperature, pressure, and composition,and how they behave when extrapolat-ed. Similarly, such results can be used togenerate plots of VLE and LLE to com-pare to diagrams in the literature and ac-tual field data. Some simulators havethe capability to generate residue curvesfor distillation of ternary mixtures. Theresidue plot capability also is a powerful

tool for distillation analysis.Use the tabulation and plotting

tools to determine the cause of dis-crepancies in properties. If a mixtureproperty is incorrect, investigate if asingle component is the cause by re-porting pure-component properties.Another useful technique is to com-pare the same flowsheet or property-table results while using differentphysical property methods.

100

80

60

40

20

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6

T e m p e r a t u r e , ˚ C

Liquid 2 Molefraction C6H12OLiquid 1 Molefraction C6H12O

MolefractionSource: (16 )

0.0006

0.00055

0.0005

0.00045

1

0.998

0.996

0.994

0.992

0.99

0.9880 10 20 30 40 50 60 70

Temperature, ̊ C

L i q u i d 1 M o l e f r a c t i o n C 6 H 6

L i q u i d 2 M o l e f r a c t i o n C 6 H 6

Figure 8 (left). LLE of cyclohexanol/water system at 1 atm.

Figure 9. (above). LLE of benzene/water system at 1 atm.


7/12


By default, most phase equilibriacalculations are performed assumingvapor and liquid phases. If your pro-cess involves two liquid phases(VLLE), be sure to specify three-phase calculations. If not, you willget incorrect results. As a part of thevalidation, you also should check thatyour property methods do not falselypredict two liquid phases.

Simulators let you specify that onlyone phase is present in a stream or aunit operation. If vapor and liquidphases are possible, however, youshould use the two-phase specification.

Nondatabank components

and missing parametersWhen you want to simulate non-

databank components or have compo-nents for which parameters are miss-ing, ask yourself the fol-lowing:

• Is this a major compo-nent in the mixture? If it isminor, can I take it out of the simulation?

• Does the componenttake part in VLE?

• Is the component non-volatile?

• Is it polar or nonpolar?• Will reaction (including decom-

position) cause this component to bedepleted?

• What properties need to be accu-rate for the chosen property methods?

These questions will help you toidentify the parameters that are need-ed based on your choice of physicalproperty methods. If these parametersare not available or cannot be deter-mined through literature search, re-gression, or estimation, then you will

have to reevaluate your choice of physical property methods or obtaindata by measurement.

You should determine what the pa-rameters will default to if the simula-tor does not find any available. It isdangerous to assume that the physicalproperty parameters were available just because the simulator did notgive you an error message. Use thesimulator manuals and on-line help to

create a list of parameters that aremissing. You should detail this infor-mation when communicating the as-sumptions of the simulation to otherusers or your management.

Certain property parameters alwaysare required for a simulation. Thesecan include molecular weight, vaporpressure, and ideal-gas heat capacityconstants. The need for other parame-ters depends upon your choice of physical property methods. The simu-lator manuals should include the infor-mation about the parameter require-ments (7). There also are parametersthat will be required for calculating theheat of reactions or the reaction equi-

librium constants. This includes theheat of formation and the Gibbs freeenergy of formation of all componentsthat participate in the reactions.

You can use your judgment aboutthe importance of a parameter to setnominal values for unimportant prop-erties. For example, if you know thata component is very nonvolatile andare using Antoine’s equation forvapor pressure (ln P = A + B /(T +C )),you can set the value of parameters A, B and C to -100, 0, and 0, respective-ly. (T is temperature.) This will assignthe vapor pressure used in Raoult’sLaw a very small value, almost zero(3.7 × 10-44!). This and similar tech-

niques to remove or minimize the im-pact of specific parameters should beused with caution, however.

If you can’t find a component inthe simulator’s databanks, make sureyou check for synonyms. For exam-ple, methoxybenzene may be listed asmethyl phenyl ether or anisole. Agood approach is to search for thecomponent using its formula. Whenselecting the component by formula,

check for different ordering of atoms.For instance, ammonia can be de-scribed as H3N instead of NH3. Ref. 2contains a formula index of organiccompounds and is a good resource foralternative names.

Once you have determined the pa-rameter requirements that are not sat-isfied, the next stage should be ob-taining and using physical propertydata.

Obtaining and usingphysical property dat a

Sources of data. To provide pa-rameters for nondatabank compo-nents or to do regression for pure-

component and binary parameters,you will need to search for availabledata. Such data may be found in a va-riety of sources, including data-com-

pilation references, hand-books, journals, and inter-nal data collections.

While most streams insimulations contain mix-tures, accurate property cal-culations are not possiblewithout accurate pure-com-ponent properties. The im-portance of pure compo-

nent data should not be underestimat-ed as they are the basis for both pure-component and mixture properties.For instance, pure component proper-ties such as vapor pressure will beused in phase equilibria calculations.Table 4 contains common sources forpure component properties, whileTable 5 lists common sources formixture properties.

The recommended order of datasearch is:

1. critically evaluated data sources;

2. nonevaluated sources;3. experimental measurements; and4. estimation techniques. Binary parameters for phase equi-

libria. Because of the large numberof binary pairs in even a simulationof only ten components, we recom-mend ranking the components so asto prioritize the pairs and focus theliterature search and measurement ef-forts on the most important parame-

Techniques to remove or minimize

the impact of specific parameters

should be used with caution.


8/12

ters. First, divide the componentsinto three groups: high, medium, andlow priority. Base the priority on cri-teria such as composition, and thepurity specifications of the process— if a component purity is specified,that component is important even if itappears only in low concentrations.Second, pair the components intohigh/high, high/medium, high/low,medium/medium, medium/low, andlow/low groups. Search the availablesources, including in-house ones, forany data for all groups. If certaincomponent pairs are known to be-have ideally, they can be excludedfrom the search. Next, use the UNI-FAC method for the missing pairs inthe medium/medium, medium/low,

and low/low categories. UNIFAC isnot recommended, however, for anypairs that include the components of high priority. A secondary literaturesearch can be used to find binary datafor similar compounds and those pa-rameters then substituted. Proposeexperimental work if any binary pa-rameter data are still missing or if data regression exposes data as inad-equate (3).

Regressing dataData regression is a powerful tool

for engineers not just to make thebest of available data, but also to ana-lyze the goodness of fit of a physicalproperty model to the data. Most sim-ulators include a data regression fea-ture. Examples of commonly re-gressed data include binary VLE andLLE, vapor pressure, heat of vapor-ization, density, and heat capacity.

Data regression finds the best fit of parameter estimates to the experimen-tal data. The best fit is represented byfinding the lowest value of an objec-tive function while matching thephase equilibrium or other constraints.One common regression technique iscalled Maximum Likelihood Estima-

tion. The objective function for thismethod is:

Σ j w j (Σi((C im - C i

e)/ σi)2) (2)

where j is a data group, C im and C i

e aremeasured and estimated variables, re-spectively, such as temperature, pres-sure, composition, or heat capacity, σiis the standard deviation or the error inthe measurement of the variable, andw j is the weighting of the data group.

When fitting phase equilibria data, theregression algorithm attempts to re-duce the objective function while thephysical property method is being usedto check that the components meet theconstraints of phase equilibria.

The work of a successful regres-sion involves selecting the right phys-ical property model and parameters,representing the data properly, choos-ing appropriate standard deviations of the data, and starting with suitableinitial estimates of the parameters.The following are general guidelinesfor data regression.

• Make sure that you are regress-ing the right parameters. Use the

same physical property method andbuilt-in databank that you will beusing in the simulation. Choose pa-rameters that have impact on the databeing used. For example, when usingan equation-of-state method such asPeng-Robinson or Redlich-Kwong-Soave, you should determine theacentric factor, ω. But, if you areusing an activity coefficient method,you should determine two or moreconstants for the Antoine model.

• Estimate as few parameters as possible. There is a tendency to use alarge number of parameters when fit-ting a model to data such as tempera-ture-dependent properties or binaryphase equilibria. Try to regress thedata with as few parameters as possi-ble. If the regression results reportthat the standard deviation of the esti-mated parameters is of the same orderof magnitude as the values of the pa-rameters, you may be estimating toomany parameters for your given data.The larger the temperature range of your data, the more parameters that

you can estimate.• Watch out for incomplete data. A

regression may yield poor results if there are missing data points, particu-larly composition data. For example,some authors do not report all com-positions in VLLE or immiscibleLLE. You may need to estimate themissing compositions so that phaseequilibrium can be calculated for allcomponents. Find out how your sim-



Source Evaluated Generally Reliable?

Critical Data of Pure Substances, DECHEMA* Critically Yes

CRC Handbook of Chemistry and Physics (Beilstein)* Noncritically Yes

DIPPR Data Compilation* Critically Yes

Encyclopedia of Chemicals, Drugs, and Biologicals Noncritically Yes

Encyclopedia of Polymer Science and Engineering Noncritically Yes

ESDU Validated Engineering Data Index Noncritically Yes

Handbook of Thermophysical Properties of Gases Noncritically Yes

and Liquids

J ANAF Thermochemical Tables Critically Yes

Lange’s Handbook of Chemistry Noncritically Yes

Perry’s Chemical Engineer’s Handbook Noncritically Yes

Properties of Gases and Liquids Noncritically Yes

Selected Values of Properties of Chemical Critically YesCompounds (TRC)*

Vapor Pressure of Pure Substances Noncritically Yes

* Parts of these sources are available on-line from DIALOG Information Services, STN Interna-

tional, or Technical Databases Services, Inc. (TDS)

Table 4. Examples and reliability of sourcesof pure component data.


9/12


ulator handles missing data to bestdeal with incomplete data.

• Specify the right number of phas-es. A regression will yield incorrectresults if the number of phases is notspecified correctly. This is a commonproblem in VLLE systems. For someliterature data, the number of phasesis hard to interpret because of the pre-sentation of the data or lack of de-scription. Often in VLLE data, only atotal liquid composition is reportedeven though two liquid phases werepresent. The author may be reportinga heterogeneous azeotrope — anazeotrope where the vapor composi-tion equals the total liquid composi-

tion but two liquid phases are present.When doing the regression of a het-erogeneous azeotrope, divide the datainto two groups, the VLE data and theVLLE data. This will ensure that thecorrect phase equilibria is considered.In regressions such as this, it is im-portant to use the property tabulationand plotting features of the simulatorto check that the parameter estimatescorrectly reproduce the original data.

• Use a model’s full functionality.Aphysical property model may be usedto calculate several properties. For ex-ample, you can use binary excess-en-thalpy ( H E ) data and binary VLE orLLE data to determine binary parame-ters for activity coefficient models. Forequation-of-state models, you simulta-neously can use liquid- and vapor heatcapacity, vapor pressure, and heat of vaporization data. If data are availablefor these properties, use these data to-gether to estimate the parameters. Datagroups of different types can be usedtogether in the same regression.

• If necessary, regress parameters

even if values are available in thedatabank. The physical property pa-rameters found in the built-in pure-component and binary databanks gen-erally are very reliable. You may find,however, that you need to determinenew parameters to replace the data-bank values for your application.Check the built-in parameters to en-sure that the recommended tempera-ture, pressure, and composition range

is not outside the range of your simu-lation. For example, vapor pressureparameters may not have been deter-mined at temperatures below the nor-mal boiling point. Most physical prop-erty models extrapolate outside thetemperature bounds reasonably well— but at some compromise in accura-cy. The parameter values also mayapply to a very wide range of tempera-ture and thus not provide as good a fitif you only need a narrow range in thesimulation. For phase equilibria calcu-lations, to improve the accuracy of VLE or LLE predictions, you maywant to use ternary or quaternary datato fine-tune binary parameters thatmay be available in the simulator.

• Check that the parameters repro-duce the data. The simulator will re-port qualitative results of the regres-sion, including the residuals (experi-mental minus estimated variables). Usethe property tabulation or plotting fea-tures to reproduce the data at the speci-fied conditions. This can be performedin the same regression run. Check that

the correct number of phases is pre-dicted by allowing two-liquid-phasecalculations for the property table orplot. In addition, your simulator mayhave an option where you can evaluatethe fit using the existing parametersand model with experimental datawithout doing a regression.

• Remove components not in phaseequilibria. If components that aresolids or ions do not appear in a

phase, you can remove them from thephase equilibria constraints. This isuseful in VLE.

• Generate equilibrium data. If you have binary parameters for an ac-tivity coefficient or equation-of-statemodel, your simulator may be able togenerate VLE or LLE data for regres-sion using these parameters. You canregress these “data” with anotherphysical property model. This allowsconsolidation of known parametersinto a single property method.

• Fit other data. Your simulatormay have a data fitting feature that canbe used for plant data. This methodmay not be as useful for predictivesimulation, though, if the data are notfrom a wide variety of conditions.

Estim ating missingproperty parameters

Property estimation usually is doneafter a data search is performed, tosupply missing property parameters.You can use built-in estimation meth-ods to fill in some gaps in your physi-

cal-property-parameter requirements.Simulators include one or more esti-mation methods for each of the mostcommon parameters. There are twotypes of estimation methods for purecomponent parameters: structuralgroup, and corresponding states.

Structural group methods are basedon the idea that contributions of theparts or structural groups of the compo-nent are additive for properties such as

Sources

Activity Coefficients at Infinite Dilution, DECHEMA Chemistry Series

Binary VLE Data file*, DIPPR

Dortmund Databank (superset of DECHEMA data collection)*

Heats of Mixing Data Collection, DECHEMA Chemistry Series

Liquid-Liquid Equilibrium Data Collection, DECHEMA Chemistry Series

Phase Equilibria and Enthalpies of Electrolyte Solutions, DECHEMA Chemistry Series

Vapor-Liquid Equilibrium Data Collection, DECHEMA Chemistry Series

Vapor-Liquid Equilibrium Data for Electrolyte Solutions, DECHEMA Chemistry Series

Vapor-Liquid Equilibrium for Mixtures of Low Boiling Substances, DECHEMA Chemistry Series

Selected Values of Chemical Compounds, Texas A&M University

Solid-Liquid Equilibrium Data Collection, DECHEMA Chemistry Series

* On-line databanks

Table 5. Examples of sources of mixture data.


10/12



normal boiling point, critical tempera-ture, critical pressure, ideal-gas heat ca-pacity, and standard heat of formation.Some methods, such as that of Benson,contain additional corrections for next-nearest-neighbor atoms or for rings.Structural group contributions are deter-mined by taking an average contributionbased on known physical constants of many organic compounds. Because theBenson, Joback (10), and other struc-tural-group methods are based mainlyon data for organic compounds, theycannot be used for inorganics, including

metals, or ions. In addition, structuralgroup methods do not accurately repre-sent very large organic molecules (thatis, ones with a molecular weight > 200)such as proteins. New group-contribu-tion methods like that of Constantinouand Gani (4) potentially may providebetter estimations for organics. Otherpossibly useful methods are proposed inthe literature but may apply to only cer-tain families of components.

Corresponding states methods arebased on empirical mathematical re-lationships among properties. For ex-ample, the Letsou-Stiel method re-lates liquid viscosity to critical tem-perature, critical pressure, and acen-tric factor. These methods most likelywill be inaccurate when used forcompounds unlike those upon whichthe correlation was based.

A good approach for both groupcontribution and corresponding statesmethods is to check the accuracy of as many methods as possible for

compounds for which properties areknown and which are structurallysimilar to the compound you are esti-mating. The following exampleshows the use of this concept.

Estimating the properties of propyl

phenyl ether. Let’s say that you aremodeling a process containing propylphenyl ether (PPE), also calledpropyloxy benzene. The only datayou have are its boiling point

(189.9°C), density at 25°C (0.9474g/cm3), and molecular structure:

You want to estimate the proper-ties of PPE using the most appropri-ate methods.

Step 1. Determine the best estima-tion methods for a similar phenylether. Select other compound(s) chem-ically similar to PPE for which youhave experimental property data. (Of course, the more similar compoundsyou can use, the greater your confi-dence that you are selecting the mostappropriate methods.) In this case, forsimplicity, let’s choose only phenetol:

Data for phenetol is available fromthe DIPPR data collection (5).

Use the simulator’s built-in methodsto estimate properties for phenetol.Then, compare the results of the various

methods with the experimentally deter-mined values to identify which methodsgive the best estimates for this class of compounds. Table 6 lists the results forthe different methods for phenetol.

You can see that the Ambrosemethod gives the best overall predic-tions for critical temperature and pres-sure, the Fedors method for critical vol-ume, and the Joback method for stan-dard heat of formation for phenetol. So,

OCH2CH3

OCH2CH2CH3

Property Name Units Data Estimated Value % Error Estimation Method

Critical temperature K 647.15 657.1265 1.54 J oback

Critical temperature K 647.15 653.738 1.02 Lydersen

Critical temperature K 647.15 652.7763 0.87 Ambrose

Critical pressure N/m2 3,420,000 3,577,070 4.59 J oback

Critical pressure N/m2 3,420,000 3,509,780 2.63 Lydersen

Critical pressure N/m2 3,420,000 3,474,970 1.61 Ambrose

Critical volume m3/kmole 0.39 0.3935 0.90 J oback

Critical volume m3/kmole 0.39 0.391 0.26 Lydersen

Critical volume m3/kmole 0.39 0.389603 -0.10 Fedors

Standard heat of formation* J /kmole -101,600,000 -105,400,000 3.74 Benson

Standard heat of formation* J /kmole -101,600,000 -104,140,000 2.50 J oback

* at 1 atm, 25°C for ideal gas.

Table 6. Comparison of estimated and experimental parameters for phenetol (C8H10O).

Property Name Units Estimated Value Estimation Method

Molecular weight 136.1937 Formula

Critical temperature K 668.6672 Ambrose

Critical pressure N/m2 3,085,350 Ambrose

Critical volume m3/Kmole 0.442373 Fedors

Standard heat of formation* J /Kmole -125,330,000 Benson

* at 1 atm, 25°C for ideal gas.

Table 7. Estimated properties for propyl phenyl ether (C9H12O).


11/12


we will use these methods to predictthe corresponding properties for PPE.

Step 2. Enter the available dataand structure for PPE. Enter normalboiling point and molecular structureof PPE, and specify the methods thatgave the best predictions for phenetol.

Step 3. Examine the estimation re-sults for PPE. These appear in Table 7.

One area of property estimationthat is more difficult is differentiatingthe properties of stereo isomers. Somegroup-contribution methods have cor-rections for ortho, meta, and paraconfigurations, but few have built-incorrections for optical isomers. Theseparation of these isomers in a chem-

ical process is based on their slightlydifferent properties — relative volatil-ity in distillation is one example.

Employing simpler methodsIn addition to structural group and

corresponding states methods, anotheruseful estimation approach is providedby series and family plots. Series plotslook at the values of a property such asnormal boiling point with increasingmolecular weight or carbon number forcompounds in a series that differ byone substituent group, such as the CH2-unit in n-alkanes. Figure 10 is a seriesplot for the normal boiling point of n-alkylbenzenes. Family plots are simi-lar, but the number of groups is larger.For example, Figure 11 shows a familyplot of the critical pressure of methyl(hydrogen)chlorosilanes. Youcan use these plots to predict propertiesby extending the curve or to check your data for errors (6). To create auseful series or family plot, however,you must be careful about the compo-nents included.

When accuracy is not critical, consid-er the simple but powerful technique of component substitution. In this, you usethe properties of another, similar compo-nent for all properties of the componentof interest that you do not know. A simi-lar component is one that has a compa-rable volatility (vapor pressure), density,and heat capacity. This is useful if thecomponent is nonvolatile or is not in-volved in phase equilibria. For example,

you have a small amount of a non-volatile component in a stream that is at100°C and 1 atm. You can access theproperties of a nonvolatile component,say C20H42 (molecular weight = 282.55,and boiling point = 343.78°C), insteadof estimating properties. This method isvery efficient if you do not need accurateproperties of the component. Take care,though, if you use this approach and oneof the UNIFAC activity-coefficientmethods, as you may change the as-sumptions made about the liquid phase.

Another technique to simplify amixture of similar components is torepresent them with a single compo-nent. This is a useful technique when

components are not known exactly.For instance, Component C5+ couldrepresent hydrocarbons of 5 carbonatoms and greater.

Estimating binary parametersYou can estimate binary parame-

ters for Wilson, NRTL, and UNI-QUAC activity-coefficient modelsusing two approaches: UNIFAC andinfinite-dilution activity coefficients.UNIFAC-estimated binary parametersusually do not provide enough accu-racy and, so, only are recommendedfor early stages of physical propertydata investigation and to “fill in theblanks” for components with mediumor low priorities.

Better binary parameters can beestimated using infinite-dilution ac-tivity coefficient data. (Some simula-tors may include this feature under

their regression tools.) This method isbetter because it is based on the com-ponents of interest, unlike the groupcontribution method, which averages

700

600

500

400

300

6 10 14 18 22 26

N o r m

a l B o i l i n g P o i n t , ˚ K

Number of Carbon Atoms

5,200,000

4,700,000

4,200,000

3,700,000

3,200,000

2,700,000

20 40 60 80 100 120 140 160 180

C r i t i c a l P r e s s u r e , P a

Molecular Weight

SiH4

MeSiH3

SiH3Cl

MeH2SiCl

Me2HSiCl

MeHSiCl2

Me2SiCl2

MeSiCl3

SiH2Cl2

Me2SiH2

Me3SiH Me3SiClMe4Si

SiHCl3

SiCl4

Figure 10(left). Series plot

of normal boiling point for n-alkylbenzenes.

Figure 11(below). Family

plot of critical pressure of

methyl(hydrogen) chlorosilanes.


12/12



the effect of group interactions fromdifferent components.

Estimation of physical propertiescan get you started in a simulationproblem — but you should do an ex-haustive literature search to findmissing pure-component and binaryparameters.

It is important to enter any known

parameters before doing property es-timation. First, experimental datagenerally are more accurate than esti-mated values. Second, corresponding-states estimation methods requireother physical constants as input.Using an experimental value will im-prove the prediction of these propertyparameters. Otherwise, the error inestimating parameters such as normalboiling point, critical temperature,

and critical pressure propagates toother property parameters.

Documentingwhat you’ve done

Simulation projects often have along life at a company. New usersmay come along and be unfamiliarwith the assumptions and recom-

mended use of the simulation. Youmay find that you need to revisit asimulation a year or more later. Doc-umenting the data sources, the rangeof applicability, and physical propertyassumptions is extremely important.This can be incorporated using thecomment or descriptions fields in thesimulator. Include a statement aboutany properties that were not well de-fined or components that should not

be added given the physical propertymethod employed — for instance,electrolytes when an equation-of-state method is being used. Keeptrack of the references for data andlist them in the simulation, if possi-ble. Include comments about proper-

ties, such as densities or heats of mix-ing, that were not of interest or notvalidated in the simulation. Keep theestimation, regression, and simulationfiles together. If possible, create a filecontaining all pure-component andbinary parameters including those ac-cessed in the built-in databanks. Thisway you will be able to reproduceyour results in the future with upcom-ing simulation-software releases.

Keeping the right perspectiveThe physical property system of

the simulator is not a black box, but awell developed set of rules and rela-tionships that can execute very com-plex calculations very quickly. It doesnot replace that most useful of alltools of a chemical engineer — com-mon sense. Always use your judg-ment to evaluate simulation errors orsuspicious results to find their source.That way, you’ll make the best use of your simulator, and avoid unneces-sary mistakes. CEP

E. C. CARLSONis a staff engineer at Aspen

Technology, Inc., Cambridge, MA

(617/577-0100; Fax 617/577-0303;

email:[email protected]). He is

responsible for providing technical guidance

and training in the areas of physical properties

and reactor modeling to process simulation

users. He received a BS from the Univ. of

Rochester and an MS from North Carolina

State Univ., both in chemical engineering. He is

a member of AIChE.

AcknowledgmentThe techniques and guidelines presented inthis article are the results of experience help-

ing simulation users solve engineering prob-lems. I would like to thank the followingcolleagues at Aspen Technology for adviceand information: Valentijn DeLeeuw,Marcelo Marchetti, Bill Mock, AndreaTakvorian, and Suphat Watanasiri.

1. Reid R. C., J. M. Prausnitz, and B.E.

Poling, “The Properties of Gases and

Liquids,” 4th ed., McGraw Hill, New

York (1987).

2. “CRC Handbook of Chemistry and

Physics,” D. Lide, ed., CRC Press, Boca

Raton, FL (1994).

3. Dewan, A. K., and M.A. Moore,

“Methodology to Develop an ASPEN

PLUS Model in Shell Development Com-

pany,” presented at ASPEN WORLD,

Cambridge, MA (1994). (Available from

Aspen Technology, Inc.)

4. Constantinou, L., and R. Gani, “New

Group Contribution Method for Estimat-

ing Properties of Pure Compounds,”

AIChE J., 40 (10), p. 1,697 (1994).

5. Daubert, T. E., R. P. Danner, H. M.

Sibul, and C. C. Stebbins, “Physical and

Thermodynamic Properties of Pure Chem-

icals: Data Compilation,” Design Institute

for Physical Property Data (DIPPR),

AIChE, New York (1989–onwards).

6. Smith, A. L., “Family Plots for Evaluat-

ing Physical Properties of Organosilicon

Compounds,” AIChE J., 40 (2), p. 373

(1994).

7. “ASPEN PLUS User Guide,” Vol. 1, Re-

lease 9, Aspen Technology, Inc., Cam-

bridge, MA (1995).

8. Gmehling, J., J. Li, and M. Schiller, “A

Modified UNIFAC Model. 2. Present Pa-

rameter Matrix and Results for DifferentThermodynamic Properties,” I&EC Res.,

32, p 178 (1993).

9. Hansen H. K., P. Rasmussen, A. Fre-

denslund, M. Schiller, and J. Gmehling,

“Vapor-Liquid Equilibria by UNIFAC

Group Contribution. 5. Revision and Ex-

tension,” I&EC Res., 30 (10), p. 2,352

(1991).

10. Joback K. G., and R. C. Reid, “Estima-

tion of Pure-Component Properties from

Group-Contributions,” Chem. Eng. Com-

mun., 57, p. 233 (1987).

11. Kleiber, M., “An Extension to the UNI-

FAC Group Assignment for Prediction of

Vapor-Liquid Equilibria of Mixtures Con-

taining Refrigerants,” Fluid Phase Equi-

libria, 107, p. 161 (1995).

12. Magnussen, P., P. Rasmussen, and A.

Fredenslund, “UNIFAC Parameter Pre-

diction Table for Prediction of Liquid-Liquid Equilibria,” I&EC Proc. Dev.,. 20,

p. 331 (1980).

13. Larsen, B., P. Rasmussen, and A. Fre-

denslund, “A Modified Group-Contribu-

tion Method for Prediction of Phase Equi-

libria and Heats of Mixing,” I&EC Res.,

26, p. 2,274 (1987).

14. Dortmund Databank, VLE Data, System

No. 980, Univ. of Dortmund, Germany

(1995).


No. 4,785, Univ. of Dortmund, Germany

(1995).


No. 752, Univ. of Dortmund, Germany(1995).

Literature Cit ed

modelos termodinamicos de aspen plus.pdf

Documents