SUCCEEDING AT SIMULATION Don’t Gamble With Physical Properties For Simulations hemical engineers use process you always will have to make assump- simulation to perform a variety tions in terms of physical properties, Finding good values of important work. This work however. The goal of this article is to out- Cranges from calculations of line the appropriate assumptions and to for inadequate mass- and energy balances of large flow- provide techniques when properties are sheets to prediction of the performance of missing. or missing physical process alternatives that can save millions of dollars. An engineer very quickly can The five important tasks property parameters define a complex flowsheet and all the Successfully describing the physical process conditions. Desktop computers properties to be used in a simulation in- is the key to a now allow rating, sizing, optimization, volves five tasks: and dynamic calculations that previously 1. selecting the appropriate physical successful required large mainframe computers. In property methods; the past, these simulations were often 2. validating the physical properties; simulation. And built by a group of experts, including a 3. describing nondatabank compo- physical property expert. Now, simulators nents (chemical species or compound) this depends upon such as ASPEN PLUS, ChemCAD III, and missing parameters; HYSIM, PRO II, and SPEEDUP are easi- 4. obtaining and using physical prop- choosing the right er to use and more powerful than the erty data; and estimation methods. standalone programs of the past. Today, a 5. estimating any missing property single engineer can set up the basic simu- parameters. lation specifications, including the physi- It can be argued that these tasks are cal properties, in very little time. not sequential and, to some degree, they Missing or inadequate physical prop- are concurrent. During simulation devel- erties, however, can undermine the accu- opment, however, you will need to visit racy of a model or even prevent you from each area to be confident that your simu- performing the simulation. That some re- lation is as accurate as possible — so Eric C. Carlson, quired information is missing is not an that important decisions can be made Aspen Technology, Inc. oversight in the simulator. After all, for based on the results of your simulations. most compounds, physical property pa- rameters are not known for every thermo- Selecting the appropriate dynamic model or for all temperature or physical property methods pressure ranges. Models have built-in as- This essential first step will affect all sumptions and practical limits that should subsequent tasks in developing accurate apply. physical properties in your simulation. In this article, we will provide practi- Indeed, the choice of the physical proper- cal tips and techniques to help you accu- ty models for a simulation can be one of rately describe the physical properties the most important decisions for an engi- needed in a simulation. As an engineer, neer. Several factors need to be consid- CHEMICAL ENGINEERING PROGRESS • OCTOBER 1996 • 35 SUCCEEDING AT SIMULATION ered, and no single method can han- Table 1. Thermodynamic property models dle all systems. Table 1 lists some available in a simulator. thermodynamic models available in simulators. Equation-of-State Models Activity Coefficient Models The four factors that you should Benedict-Webb-Rubin(BWR)-Lee-Starling Electrolyte NRTL consider when choosing property Hayden-O’Connell* Flory-Huggins methods are: Hydrogen-fluoride equation of state for NRTL • the nature of the properties of hexamerization* Scatchard-Hildebrand interest; Ideal gas law* UNIQUAC • the composition of the mixture; Lee-Kesler (LK) UNIFAC • the pressure and temperature Lee-Kesler-Plocker Van Laar range; and Peng-Robinson (PR) Wilson • the availability of parameters. Perturbed-Hard-Chain Predictive SRK Special Models To ease the selection of the right Redlich-Kwong (RK) API sour-water method physical property methods, we sug- Redlich-Kwong-Soave (RKS) Braun K-10 gest using the decision trees shown in RKS or PR with Wong-Sandler mixing rule Chao-Seader Figures 1–3. These trees are based on RKS or PR with modified-Huron-Vidal-2 mix- Grayson-Streed the four factors for selecting property ing rule Kent-Eisenberg methods, and can be used when the Sanchez-Lacombe for polymers Steam Tables . chemical components and approxi- * Not used for the liquid phase mate temperature and pressure ranges are known. While these diagrams are simplifications, they do show the Non-electolyte basic steps of the decision-making See Figure 2 process, while the notes in the sidebar amplify some of the key points. The nature of the properties of in- Polar terest. A question that you may ask E? yourself when starting a simulation is “Does the choice of physical property Electolyte Electolyte NRTL methods matter?” The answer is an or Pitzer emphatic YES. The choice can ? Real strongly affect the prediction of the Peng-Robinson, simulation. You should be selecting a Redlich-Kwong-Soave, collection of methods that will best Lee-Kesler-Plocker predict the properties or results of in- All Nonpolar terest to you. R? Chao-Seader, Grayson-Streed or Because many chemical process Braun K-10 simulations include distillation, strip- Pseudo & Real ping, or evaporation, one important P? potential consideration for the choice of physical property models is vapor/liquid equilibrium (VLE). This Vacuum Braun K-10 or Ideal is the area in which the most physical property work is focused in chemical engineering. Liquid/liquid equilibri- ? Polarity E? Electolytes um (LLE) also becomes important in processes such as solvent extraction Real or and extractive distillation. R? Pseudocomponents P? Pressure Another critical consideration is pure-component and mixture en- thalpy. Enthalpies and heat capacities are important for unit operations such Source: (7) as heat exchangers, condensers, dis- tillation columns, and reactors. ■ Figure 1. The first steps for selecting physical property methods. 36 • OCTOBER 1996 • CHEMICAL ENGINEERING PROGRESS Navigating the decision trees Here are some pointers to help you navigate the decision trees that ap- recent years, there have been improvements to UNIFAC (see Table 3) pear as Figures 1–3. that can better predict VLE, heat of mixing, and LLE over a wider temper- What are pseudocomponents? In many applications where only non- ature range. Recent extensions to UNIFAC proposed for molecules such polar molecules are present (such as in hydrocarbon processing and re- as refrigerants and sugars may be useful, and you can add the groups fining), the mixture is so complex that instead of representing it by all the and parameters to your simulation. Simulators may have the ability to known constituents, it is easier to group the constituents by some useful generate binary interaction parameters for Wilson, UNIQUAC, or NRTL property such as boiling point. In this way, a mixture of hundreds of con- from UNIFAC. stituents can be reduced to 30 or fewer. The properties of these grouped Not all components can be described using UNIFAC, however, and constituents, called pseudocomponents, are represented by an average not all group interactions are available. Examples of components that do boiling point, specific gravity, and molecular weight. If you do not use not have UNIFAC groups include metals, organometals, and phosphates. pseudocomponents, the constituents should be described by a molecu- So, we highly recommend always doing a search for available data on lar formula and are referred to as real components. binary or ternary systems of interest. Why are electrolyte mixtures different? Electrolyte mixtures include How should the vapor phase be treated? The choice of the VLE components that are charged molecules (ions) or that form salts. Some method using an activity coefficient model also requires a choice of simulators allow calculation of electrolyte reaction equilibrium with model for the vapor phase properties. If vapor phase association is ob- phase equilibrium. This is a very powerful method and its usage covers served (as in the case of acetic acid), then the vapor phase model many applications such as caustic scrubbing, neutralization, acid pro- should be Hayden-O’Connell or Nothnagel. A system containing hydro- duction, and salt precipitation. The nonideality of electrolyte solutions, gen fluoride may require a special model to represent the high degree of usually containing water, can be observed in boiling point elevation, salt- association due to hydrogen bonding. Association in the vapor phase ing out of gases (that is, adding salts to the solution to change the solubil- can have a strong effect on phase equilibria and enthalpy. ity of gases), and salt precipitation. The most common electrolyte meth- When should defaults be overridden for other physical property ods are the Pitzer model, and the modified-NRTL activity coefficient methods? Prediction of density, enthalpy, and viscosity also are impor- model of Chen and coworkers. Some electrolytes, like formic acid and tant in simulators, and you shouldn’t automatically accept the default acetic acid, are very weak and an electrolyte method is not required. methods. Check the simulator documentation for the default method and Which type of method should be chosen for mixtures containing mixing rules. polar components but no electrolytes? There are two groups of methods Vapor density is calculated by an equation of state or the ideal gas — based on activity coefficients or equations of state. Use activity-coef- law. Mixture liquid densities can be calculated by an equation of state, a ficient-based methods when pressures are low to medium (typically less temperature-dependent model such as that of Rackett, or by a tempera- than 10 bar or 150 psia) and if no components are near critical point. Ac- ture- and pressure-dependent model such as the COSTALD. For psuedo- tivity coefficient models also often are used to accurately predict non- components, an American Petroleum Institute (API) method typically is ideal liquid behavior such as for VLE and for LLE.