CHAPTER 3 CALCULATION OF THERMODYNAMIC PROPERTIES ‘The thermodynamic properties of most interest to those working In natural gas are the vapor‘iquid equl- liprium ratio (K-value), enthalpy changes, and entropy changes. All of the equations of state discussed in Chapter 2 can be adapted to the calculation of these variables. ‘Several procedures for estimating equilibrium constants were discussed in Chapter of Volume 1 The equilibrium constants provided there In graphical form, and used throughout Volumes 1 and 2, were derived from the Soave-Redlich-Kwong equation of state. They were computed based on a composition that represents typical natural gas mixtures. They included, therefore, no corrections for composition ef- fects. They did incorporate changes caused by temperature, and the non-idealties introduced by high pressure. In Chapter 8 of Volume 1, enthalples were presented as a function of molecular welght and temperature for liquids, and as functions of temperature, pressure, and molecular weight for gases. Molecular weight has long been used as a correlating parameter for enthalpies used for hand calcula- tions for both natural gases and natural gas liquids. However, molecular weight Is not totally satisfactory ‘as a correlating parameter, particularly when the mbture contains molecules from different ‘families’ of compounds. As an example, carbon dioxide and propane possess almost Identical molecular weights, but thelr enthalpies will differ signicantly. In Volumes 1 and 2 there was no specific discussion of methods for calculating entropy changes. ‘The change in entropy Is primarily of interest in compression and turboexpansion processes. The methods utilzed in Volumes 1 and 2 for calculating compression and turboexpansion were based on constant entropy processes, but did not require explicit calculation of the entropy change for the process. In this chapter we will discuss equations of state, equation-of-state solution methods, and tech- niques used to calculate thermodynamic property values from contemporary equations of state. VAPOR-LIQUID EQUILIBRIUM ‘The vaporsliquid equilibrium constant (K) is defined simply as the mol fraction of any component in the vapor divided by the mol fraction of the same component in the liquid, K = yifk. As discussed in Chapter 5, Volume 1, the value of K can be found from the component vapor pressure In the vicinity of COMPUTER APPLICATIONS 34CALCULATION OF THERMODYNAMIC PROPERTIES atmospheric pressure when ideal behavior can be assumed. At elevated pressures, K depends on T, P and composition. Some thermodynamic means must be developed to reflect non-ideal behavior. The ‘Gibbs®” free energy is a property of particular importance because It can be related to the equilibrium state and, at the same time, can be expressed as a function of temperature and pressure: dG) = -SdT + VoP @1) |f Equation 3.1 is applied to an ideal gas at constant temperature: Gi = RTdinP @2) Equation 3.2 applies only at ideal conditions but itis In a form convenient to use with equations of state. To preserve this form a secondary thermodynamic concept was created that was given the name fugacity, represented by the symbol "t." Like all such concepts It has no theoretical basis but is a calculable quantity. Equation 3.2 now can be written for gases under all conditions. dGi = ATA¢In f) 33) Fugacity must have the units of pressure (by definion) but will only equal pressure under ideal ‘gas conditions. Thus, some refer to fugacity as a “corrected pressure" although this is a somewhat simplistic definition, thermodynamically ‘The fugacity Is an unbounded variable and ranges over the same span of numbers as pressure. A more valuable parameter for use in correlative/predictive procedures would be a variable with characteris- tics similar to fugacity but which ranged over a much smaller span of numbers, The fugacity coefficient for a pure component Is defined as: o = HP G4) For a component in a gas mixture the fugacity coefficient is: i = fiyP 85) ‘Another secondary thermodynamic function called activity coefficient is defined by the equation: y= fhe f° 68) Where: f° = component fugacity in the pure state ‘The equation of state can calculate reliable K-values, or equilibrium constants, only where the va- por and tiquid phase co-exist in equilibrium, The criterion used for equilibrium is that the fugactty of the ‘component in the vapor phase be equal to the fugacity of that component in the liquid phase, for all ‘components. When this criterion is met, the vapordiquid equilibrium constant is: K = dio @7 ‘The first equation used for computer solution of vapordiquid equilibrium constants was developed by Chao and Seader.** Thelr equation of state is worth noting from that standpoint alone, but it also has another attribute. The Chao-Seader equation of state uses different equations for a component in the vapor phase and for a component in the liquid phase. As shown in the equations presented in Table 3.1, the equilibrium constants also calculated from a slightly different equation. In Table 3.1, di Is the fugac- ity coefficient of the pure component at system conditions; Is the activity coefficient of the component in the liquid phase as defined by Hildebrand®?; and iY Is the fugacity coefficient of the component in the vapor phase. > VOLUME ItVAPOR-LIQUID EQUILIBRIUM TABLE 3.1 Chao-Seader Equations Jogi = a0 + ai/Tr + a2Tr + aa Ti? + aT? + (a5 + 96 Tr+ a7 T?) Pr + (ag + a9 71) Pr? —log Pr + wi fare + anTr + ata Tr + ata T® + ara (Pr - 0.6)] Ing” = @-1) B/B-In(Z- B) - AB [2(AVA)°> - ByB} In(t + h) Z = 1/(1-h)-AB [h(t + h)) h= BZ; A= (VAy)*; B= ZBiyi AL = 0.4278 Pa TH?® ; 0.086 67 Pai ‘The BWRS, the SRK, and the PR equations are all termed “single equations of state because, as shown in Table 3.2, the same equation is used to describe the fugacity coefficient in the vapor and the liquid phases. If the vapor phase Is being considered, the value of x Is the composition of the com- ponent in the vapor phase. If the liquid phase Is being considered, x is the composttion of the TABLE 3.2 Equation of State Forms for Calculating Fugacity (Coetficient) BWRS 2 RT Inf = AT In @ ATX) ple + Ba) RT + 293) [1h al (=) --S (Ky? 2 1 ee (=a)! - fe om] +2 [owen RT —3(@? a)" a? d) "2 | + pS [auzar® + ayesy"® ] + [p55] (@ + dT) (0? ai) 2oly T? (wy)? 3(c? «8 pF [ = expe") _ expo) ] a: ve 2 x [1 -expeyp?) (1 + yp? + 12? pM] ‘SRK b& bio @-1)-ine-6)-a [uae | In(t + BZ) PR A [2Bxiax 2 Jn tet | Ing = (bib) Z— 1) -INZ—B)— pe | Z-0414B COMPUTER APPLICATIONSCALCULATION OF THERMODYNAMIC PROPERTIES component in the liquid phase. The binary interaction parameter, ky, Is obtained from binary or mult component mixture data for the components in question. The interaction parameters have minimal in- fluence on denstties, but major impact on the vaporiquid equilibrium constants for the individual com- ponents in a mixture, ‘The basic equilibrium calculations are the bubblepoint, dewpoint and flash. In the bubblepoint, ‘dewpoint and fixed L/F flash, elther temperature or pressure may be the independent variable. The other (pressure or temperature) Is the dependent variable. In the general flash calculation, the temperature and pressure are usually fixed and LF is the dependent variable. In some cases, enthalpy or entropy will r- place temperature or pressure as an independent variable. In these cases the replaced variable be- ‘comes a dependent variable. Situations frequently occur where specification of the wrong independent variable can cause a particular equilibrium calculation to fall. Examples of these cases and the causes of fallure are reviewed later. ‘These calculations enable one to calculate the boundaries of the phase envelope as well as any point Inside. Special models can be used to calculate the critical, cricondentherm and cricondenbar points. BUBBLEPOINT CALCULATIONS In bubblepoint calculations, the liquid composition Is known and ether the temperature or pres- sure is fixed. ‘The vapor phase composition, and pressure or temperature of the system are unknown. The vapor phase is predicted by: yi = Koa. ‘Several different values of the dependent variable are assumed. The correct value is the one that yields: = Zy = Wwte 8) ‘An additional requirement when using composttion-dependent K-values is: lymtt-y" ls € @9) Where: = I = 1ton m = Iter. no. ‘The value of ¢ Is arbitrarily small (usually In the range 0.0001 to 0.000 01). If the composition of the vapor phase is not correct, the predicted K-values will not be correct. As a resutt, the composition of the vapor phase must be stabilized even though the correct value of the dependent variable has been determined. If Equation 3.8 has been satisfied and Equation 3.9 Is not satisfied, Iterations through the bubblepoint calculation must be continued until both equations are satisfied. ‘A further problem in the bubblepoint calculation Is the intial estimation of the vapor phase com- position, This problem is particularly acute when using one of the single equations of state. The vapor ‘and liquid phase compositions must be known to estimate the K values. One approach that has been successful is to assume that the mol fraction of the lowest bolling component in the system Is equal to Unity in the vapor with the remaining component mol fractions set to 10°. The vapor composition is ad- justed after oach teration, The adjustment procedure depends to a limited extent on the thermodynamic, property prediction procedure, but usually consists of a linear combination of the assumed and cal- culated vapor phase compositions. VOLUME IiiDEWPOINT CALCULATIONS A logic diagram ilustrating the basic bubblepoint calculation is shown in Figure 3.1. (The con- vergence procedure for temperature dependent bubblepoint calculations is shown in Figure 3.3. Pres- sure dependent bubblepoint calculation convergence procedures are shown in Figure 3.4.) All of the preceding procedures - vapor phase composition estimation, checking for vapor phase composition con- vergence, satisfaction of Equation 3.8 - are shown. DEWPOINT CALCULATIONS Dewpoint calculations are the opposite of bubblepoint calculations - the vapor phase composition Is known and the liquid phase composition and the system temperature or pressure is calculated. The equation used in the dewpoint calculation is: x: = y/K. ‘As in the bubblepoint calculation, several different values of the dependent variable are assumed. The correct value of the dependent variable has been chosen when: EwK = Ex = 1946 (8.10) subject to the condition that: [a t? am Lx € @.11) Where: = | = 1 ton m = iter, no. As in the bubblepoint calculation, ¢ will be set arbitrarily small (probably at the same value). In the dewpoint calculation, the liquid phase composition is initially unknown and must be assumed. A proce- dure that has been successfully used is to assume that the mol fraction of the highest boiling component in the syster * equal to unity. The remaining component mol fractions are set to 10°. Liquid phase ‘compositions ar: adjusted by linear combination of the assumed and calculated values during each itera- tion. A logic diagram illustrating the calculation sequence for dewpoint calculations Is given in Figure 3.2. ‘The convergence algorithms for temperature and pressure dependent calculations are shown in Figures 3.3 and 3.4 respectively. (The same correction calculations are used for both the bubblepoint and the dewpoint.) Flash Calculation T, P Fixed ‘A wide variety of flash equations and convergence algorithms has been published. All of the algo- rithms have the same objective - prediction of the composition and amount of the co-existing vapor and liquid phases at a fixed temperature and pressure. The majority of these procedures are essentially iden- tical. ‘One of the algorithms that has been found successful over a wide range of L/F conditions Is: a 2i(1 ~K) 1 (UF)(1- Ki) + Kr our)” = 00 12) Equation 3.12 can be developed from overall and component material balances, and the basic definition of K. This equation is applicable to a wide range of L/F conditions. It tends to fall at LF 0.000 01 unless extreme caution is used in the solution procedure. Most of the other flash calculation al- gorithms have similar problems either at LF = 1.0 or LF = 0.0. COMPUTER APPLICATIONS 3sCALCULATION OF THERMODYNAMIC PROPERTIES ENTER WITH x, , (T), P ORT, (P) ASSUME VAPOR PHASE COMPOSITION CALCULATE K VALUES ATT, PL XY CALCULATE y, & Ly; ADJUST DEPENDENT VARIABLE. EXIT WITH y,, ol ADJUST VAPOR PHASE COMPOSITION Figure 3.1. Block Diagram for Bubblepoint Calculation VOLUME IiiDEWPOINT CALCULATIONS ENTER WITH y,, (T), P OR (P), T ASSUME LIQUID PHASE COMPOSITION CALCULATE K VALUES ATT, P, xi. CALCULATE x; AND 2x; (.) INITIAL ASSUMPTION OF DEPENDENT VARIABLE ADJUST DEPENDENT VARIABLE ADJUST LIQUID PHASE COMPOSITION Figure 3.2 Block Diagram for Dewpoint Calculation EXIT WITH eee COMPUTER APPLICATIONS oYCALCULATION OF THERMODYNAMIC PROPERTIES (9) G1 - ty)" (BP), (x, - 1" (DP) O(L/F) (FLASH) ENTER WITH T™, mH T' T™ + ALLOWABLE CHANGE EXIT WITH T™T Figure 3.3 Temperature Adjustment Routine for Equilibrium In the basic flash calculation, the temperature, pressure, and composttion of the feed are fixed. ‘The unknown variables are the liquid and vapor phase compositions and the value of LIF. A conver. gence algortthm that can be used with Equation 9.12 Is: UFR” oUF” (UA) = (Ur @.13) Where: _ apm ttc oO = 2, URG-m) + ‘VOLUME ItDEWPOINT CALCULATIONS Ga™ = (1 Zy,)™ (BP), (x, - 0™ (oP) O(LIF) (FLASH) ENTER WITH P™, x™ |snp™! — on™) < ALLOWABLE CHANGE anP™*1 = gnp™ + ALLOWABLE CHANGE SAVE 2nP™, x™ AS anP' u EXIT WITH P™? Figure 3.4 Pressure Adjustment Routine for Equilibrium Calculations ‘This convergence algorithm is reliable provided the values of (UF)"*" are constrained to be valid by material balance considerations: 0.0 < (UF! < 1.0 (8.14) The classical bubblepoint/dewpoint checks: BKi2> 10 and EzKi > 10 3.15) (to assure that the mixture Is in the two-phase region) cannot be used in most equation-of-state based flash calculations because the K values for a given system are not known until the final solution has been COMPUTER APPLICATIONS 39CALCULATION OF THERMODYNAMIC PROPERTIES reached: Consequently, the flash calculation (and Its convergence algorithm) must be capable of performing “flash calculations" on single phase systems (subcooled liquids, super-heated vapors, or dense gas systems) as well as reliably predicting the amount of vapor and liquid present in a two-phase ‘system. When the above fiash equation/convergence algorithm is used on single phase systems, the final value of L/F will usually be outside the interval described by Equation 3.14 unless the material balance constraint is enforced. Should a value of (UF)"*" outside the limits defined by Equation 3.14 be detected in an iteration, the value of L/F predicted by Equation 3.13 should be replaced by the ap- propriate value described by the following equations: tt (UF)™** < 00, (un™*! = (UFA"/20 @.16) # UAT > 10, (Amt! = [1 + UA™20 6.17) This procedure eliminates most of the problems associated with flash calculations in single phase regions and yields excellent results In relatively few Iterations inside the two-phase region. Some problems still occur when attempting flash calculations in the dense gas region. Intial estimates of the phase composition must be made to initiate the flash calculation. Several procedures.are avallable. A combination of the bubblepoindewpolnt intial phase estimation procedures ‘works quite well. Set the vapor phase mol fraction of the lightest component in the system to 1.0 and the liquid phase mo fraction of the heaviest component In the system at 1.0. All other mol fractions are set to 10°. This procedure Is superior to the technique of basing the intial assumption of the phase compo- sition on some non-composition dependent K-value estimation procedure, particularly when a wide range of temperatures, pressures, component types, composition ranges, etc., Is to be considered. ‘The estimated vapor and liquid phase compositions must be compared with the calculated phase composttions. Equations 3.9 and 3.11 describe this checking procedure. If the restraints described by these equations for any component (in elther phase) are not satisfied, the calculations must be repeated even though an acceptable value for L/F has been determined. Some feel that this detalled checking procedure Is unnecessary. It probably Is for most problems involving moderate temperature-pressure- ‘composition conditions. However, at extreme conditions of temperature, pressure, and composition (low ‘temperature, high pressure, high acid gas compostton) failure to perform these composition checks will lead to results that are completely incorrect (poor estimates of the phase compostions and incorrect UF ratios). Unfortunately, the boundary changes In temperature, pressure or composition can completely alter the difficulty of a glven problem. Consequently, careful application of these checks in all calcula~ tions is strongly recommended since one can never be sure that a particular problem will not fall into the area of extreme conditions. A logic diagram ilustrating the basic flash calculation is shown in Figure 3.5. Flash Calculation at Fixed L/F and T or P Conditions Flash calculations at fixed L/F and temperature or pressure are frequently necessary. In these cal- culations, the dependent variable becomes pressure or temperature and the flash calculation becomes similar in principle to a bubble or dewpoint calculation. The flash calculation equation described eartler can be coupled with the temperature or pressure adjusting algorithms described for the bubblepoint/dew- point calculations to perform these calculations. nial estimates of the vapor and liquid phase compost tions must be made and the approach described In the flash at fixed temperature-pressure conditions can be used. The logic diagram for this type of calculation can be deduced from the earfier diagrams. 310 ‘VOLUME IIIDEWPOINT CALCULATIONS ENTER WITH T, P, ASSUME VAPOR AND LIQUID PHASE COMPOSITIONS CALCULATE K VALUES CALCULATE O(L/F), O'(LIF) —————> EXIT WITH carte] @— to continue. 2511 K VALUE 3.12917 1.41928 248772 123827 115165 111738 107328 106038 101262 9.65 MPA [1400 PSIA} COMPUTER APPLICATIONS 345CALCULATION OF THERMODYNAMIC PROPERTIES PRINTOUT 3.2 Calculation of Higher Dewpoint Temperature at 9.65 MPa THERMO (SRK) PAGE 9 UPPER DEWPOINT AT 9.65 MPA [1400 PSIA] DEWPT: VARY ‘TEMPERATURE: T ;FIX P =.39 DEG ¢ [31-3 F] , PRESSURE= FEED/PRODUCT RATES ARE KG-MOLS 9.65 MPA [1400 PSTA] COMPONENT FEED LIQUID «VAPOR xK NAME, MOLS MOL FR MOLS MOL FR MOLS MOL FR | VALUE N2 10.00 0100 +00 .0019 10.00 .0100 5.28593 CHa 939.00 .9390 00 14567 939.00 .9390 2.05610 caus 31.00 0320 100.0546 31:00 10310 © 56726 c3H8 10110 100.0475, 11:00 ‘0110 © 123154 reaH10 0030 00 +0236 3100 10030 ©. 12716 Ne4H10 +0020 00 +0210 2:00 +0020 09506 TesH12 10010 00 0196 1:00 ‘0010 = :05112 NeSHI2 10010 00 0244 100 ‘0010 =. 04095 cePLus +0020 too, 13506 2100 0020 00570 ‘TOTAL 1000.00 1.0000 +00 2.0000 1000.00 2.0000 HMSUL 6926.91 6.927 100 -1.579 6926.91 6.927 S;MJUL/K 1144 Too 128 0144.32 146 MOL Wr 51.954 17.397 D:KG/M3 588.558 97.383 MASS;KG —-17396.7 +0 17396.7 MOL $ VAP= 100.00;WT % VAP= 100.00;VOL % LIg= 00 VAPOR COMPRESSIBILITY FACTOR = «761 Pause. Please press to continue. a6 VOLUME ItDEWPOINT CALCULATIONS PRINTOUT 3.3 Calculation of Lower Dewpoint Pressure at -0.39°C THERMO (SRK) PAGE 8 LOWER DEWPOINT PRESSURE AT -0.39 C [31.3 F] DEWPT:VARY P ;FIX T TEMPERATURE -.39 DEG C [31.3 DEG F] , PRESSURE= .46 MPA [66.7 PSIA] FEED/PRODUCT RATES ARE KG-MOLS COMPONENT FEED LIQUID ‘VAPOR K ‘NAME MOLS MOL FR MOLS MOL FR MOLS MOL FR VALUE nz 10.00.0100 +00 0002 10.00.0100 142.41900 cHa 93900 19390 100 10272 939.00 .9390 34.54110 c2H6 31:00 10310 00 0069 31:00 0310 | 4.50169 c3H8 11:00 [0110 00 0107 21:00 ‘0110 1.03151 Ic4H10 3:00 +0030 100 0082 3:00 10030 © .36660, NeaH10 2:00 0020 100 0084 2:00 0020 123763, TeSH12 1100 ‘0010 +00 0122 1:00 0010 «=. 08215 NCSH12 1:00 0010 300 0172 1:00 0010 108799 CePLUS 2-00 0020 100 +9092 2:00 +0020 © :00220 TOTAL 1000.00 1.0000 +00 1.0000 1000.00 1.0000 HiMUL 9181.00 9.182 100 ~8.445 9181.00 9.181 S:MJUL/K 175.82 176 100.081 «175.82 1176 MOL WT 37.397 95.120 17.397 D:KG/M3 743.272. 31584 MASS;KG —-17396.7 +0 17396.7 MOL % VAP= 100.00;Wr % VAP= 100.00;VOL % LIQ= +00 VAPOR COMPRESSIBILITY FACTOR = 4987 Pause. Please press to continue. i (COMPUTER APPLICATIONS 347CALCULATION OF THERMODYNAMIC PROPERTIES ‘estimates of the dependent variable for subsequent calculations. Note that if the mixture being con- sidered is a narrow bolling range mbdure, and the phase diagram Is needle shaped like that shown in Figure 39, both bubblepoint and dewpoint calculations will probably fall, regardless of the type of cal- culation oF quality of the initial estimate of the dependent variable, unless good initial estimates of the phase composttion are also provided. Bubble/Dewpoint fallures very common Pressure ‘Temperature Figure 3.9 Example of a "Needle Shaped" Phase Envelope Specification of an equlibrium calculation in a region where It cannot exist is a relatively common error. These spectfications are tlustrated in Figure 3.10. Dewpoints at pressures greater than the cricon- denbar, or at temperatures greater than the cricondentherm, dewpoints in bubblepoint regions and/or ubblepoints in dewpoint regions are frequently requested calculations. The program simply cannot Dewpoint Calculation Specified Dew point Calc. Spec.’ Pressure ‘Temperature Figure 3.10 Forbidden Regions for Bubble/Dewpoint CalculationsDEWPOINT CALCULATIONS produce a valid answer with these spectfications. Split equation of state K-value correlations may produce a result that appears to be reasonable. However, this answer Is completely wrong. The single equation of state K-value correlations will usually provide an answer that is unrealistic and glve K values for all components that are nearly equal to one. ‘The program user Is usually working from an operating point. He has (or thinks he has) some evi- dence from a process unit that indicates the bubblepoint or dewpoint should be in a specific tempera- tureJpressure region, and he will normally blame ether the thermodynamic property correlation or the Program or both. However, the fault with the solution probably lies with the characterization of the Cs + fraction. A simplified example would be to describe nG1o in a gas mbxture as nG7. There is no way the program can infer that you really mean your heavy fraction is, In fact, nC1o. Consider the pressure-temperature phase diagrams shown in Figure 3.11. All of the phase dia- grams shown are for exactly the same gas composttion, which Is shown In Printout 3.1. The only thing that has changed is the characterization of the Cé-+ portion of the gas. It has been characterized as Ce, C7, Ce, Cio and as a mbsture of undefined composition but with bolling point of 110°C [230°F], 60°API gravity and 100 molecular weight. Changing from Cs to nC1o changes the’ cricondentherm from about -1°C [30°F] to about 77°C [170°F] and the cricondenbar from about 8.62 MPa [1250 psla] to about 14.83 MPa [2150 psla]. Changes of this magnitude can certainly have profound influence on gas behavior and the selection of optimum processing schemes for the gas. 3000 2500 2000 1500 1900 900 800 700 600 500 400 Pressure, psia 300 250 200 150 100 =250 -200 -150 -100 -50 0 50 100 150 200 Temperature, °F Figure 3.11. Phase Diagrams for Different Hexanes Plus COMPUTER APPLICATIONS 349CALCULATION OF THERMODYNAMIC PROPERTIES Without experimental data for some points on the dewpoint and bubblepoint curves, there Is no way to say exactly which of the phase envelopes in Figure 3.11 Is most neatly correct. To avold fallures based on these problems: 1. Properly spectty the characteristics of the Cs + fraction. 2. Use the proper type of equilibrium calculation. You may have to determine selected parts of the phase envelope for the system under considera- tlon to be able to determine If the system you have defined for the computer really represents the system ‘you Intend to study. K-value correlations based on a single equation of state are the most sensttive to the initial est mates of the dependent variable. Spit equation-of-state based K-value correlations are relatively Insens!- tive to the initial estimates of the dependent variable. Both forms of K-value prediction should recelve the best estimate possible to minimize the amount of time required to reach a solution. As a general rule, if the Initial estimate of the dependent varlable was not good enough, the single equation-of-state approach will simply produce a bad result that may look reasonable. The required quality of the intial estimate de- pends on many factors - the composition of the system, the temperature range in which the calculations are being performed, the approach to the critical point, etc, For the normal full-ange natural gas system, guidelines for intial estimates of the dependent variables are: 1. At low pressures the temperature must be estimated within 50°C [100°F] and/or the pressure estimated within 1000 kPa [150 psia]. 2. At Intermediate pressures, the temperature must be estimated within 30°C [50°F] and/or the pressure estimated within 500 kPa [75 psla]. 3. Near the crtical point, the temperatures must be estimated within 45°C [10°F] and the pres- sure estimated within +200 kPa [30 psla]. 4, At high pressure but not near the critical, the Initial temperature must be estimated within “£15°C [30°F] and/or the pressure estimated within +500 kPa [75 psla]. For narrow bolling range mbdures, the intial estimates should be divided by two. Recognize that these guidelines are qualitative. Many examples can be shown where poorer intial estimates of the de- pendent variable yielded a solution. In other cases involving similar systems, better intial estimates than those listed above yielded an Incorrect or no solution. Unfortunately, when encountering these problems, ‘you simply must,use a trial-and-error approach - try a range of intial estimates of the dependent variable to determine the one that gives you a consistent answer. If you have any doubts about any answer, run a series of equilibrium calculations In the range of Interest and plot the results. Improper charactertzation of C+ fractions is an Insidious trap. All engineers - both the fresh ‘graduate and the experienced engineer - have been taught that the Ce+ In a natural gas stream can be considered to be equivalent to some n-paraffin, This assumption Is completely wrong in many cases. Proper characterization of the Ce+ requires additional information that is often dificult and expensive to obtain. Too often the engineer will assume that his training/experience applies and specify the Ce+ as an rvparaffin. The K-value correlation Is then blamed for is inabllty to produce good answers when, In fact, the engineer Is at fault. In the final analysis, responsibitty for checking the quality of the results and properly adjusting program input data to achlove good resuits rests solely on the shoulders of the engineer using the Brogram. That Indhidual must leam the strengths and weaknesses of the program being used and how fo overcome these weaknesses by improving the quality of the input data submitted. aon. VOLUME IttDEWPOINT CALCULATIONS Flash Calculation Fallures AAs In the preceding section, the word failure means the inability of the program to property predict the thermodynamic behavior of an existing system. This fallure may be caused by: 1. Specification of a temperature/pressure outside the two-phase envelope. 2. Deficiencies in the K-value correlation method, 3, Improper spectfication of the Ce + fraction characterization. If any difficuities with flash calculations are found, perform at least a partial, not complete, phase ‘envelope calculation and plot the results. By doing this, the problems of attempting to perform flash calculations outside the two-phase region will be avoided. Determination of the phase envelope permits evaluation of the qualty of the Ce+ fraction characterization. If, for example, a flash calculation at the specified temperature, pressure and Ce+ fraction characterization does not yield the expected distribu- tion of liquid and vapor, you generally can assume that the Ce-+ charactertzation is bad. Single equation-of-state based K-value correlations are notoriously sensitive to temperature and Pressure conditions outside the two-phase region. These K-value prediction methods usually will Produce a set cf component K values all nearty equal to one when flash calculations are attempted in single phase regions. Theappearance of a set of calculated K values all equal to one is the first and best indicator that a,flash calculation was attempted In a single phase region. (Note - all component K values are also equal to one at the crtical point; however, the chances are low that a random temperature-pres- sure spectfication will be the critical point of the mbdure.) if you have good reason to believe that the system should be two phase at your specified condition, you have no alternative but to change the Cs + characterization. Making the Ce+ heavier will usually expand the two-phase region. ‘The major problem with split equation-of-state based K values Is llustrated in Figure 3.12. Notice the spilt equation-of-state predicted phase envelope is clearty incorrect. Further, flash calculations per- formed between the bubblepoint and dewpoint lines using a split equation of state will produce liquid and vapor streams when, in fact, the system should be all liquid. Another deficiency of the spilt equation-of- state approach to predicting K values is llustrated in Figure 3.13. In this case, a cricondenbar Is never Predicted. This means that the pressure of a unit operation like a distillation column could be raised to ‘almost any level, and a reasonable separation would be predicted even If the pressure of the column was beyond the cricondenbar of the distilate product. This mistake has been made repeatedly. The reader should understand that a single equation of state might give erroneous predictions of ‘bubblepoints and/or dewpoints, as shown in Figure 3.12 and, particularly, 3.13. The difference between the two equations of state in circumstances such as this Is that the split equation of state "thinks" that it hhas a reasonable and correct solution to the bubblepoint and/or dewpolnt. Equilibrium constants will show a satisfactory range of values. In all other respects the solution from the split equation of state ap- ears to be a good one. Only the judgment and experience of the engineer will tll him that the solutions are incorrect. When this kind of behavior Is predicted by a single equation of state, there are indications (generally that all K values are equal to 1) that the solution Is not a good one. The examples shown here will occur when using all split equations of state when the correlation Is Pushed beyond its range of fit. Within the range of ff, the split equation of state can provide excellent results and will frequently perform better than even the best of the single equations of state. Unfortunate- ly, precise definition of the range of fit of spit equations of state Is difficult. There are too many cond tions that have Impact on the ability of this type of correlation. Consequently, a complete phase en- velope should be calculated for each stream (or each critical stream) In a given process simulation f a split equation of state Is being used. COMPUTER APPLICATIONS 321CALCULATION OF THERMODYNAMIC PROPERTIES split equation Pressure single equation of state Temperature Figure 3.12 Phase Envelope Predicted by Spilt and Single Equation of State Approaches split equation of state Pressure single equation of state Temperature Figure 3.13 Another Example of Phase Envelope Predictions by Split and Single Equation of State Approaches VOLUME ItDEWPOINT CALCULATIONS Equation of State Comparisons Using the mbdure composition provided on page 97 of Chapter 5 In Volume 1, the bubblepoint and dewpoint at 4 MPa and the amount of liquid formed at 100°C were calculated. Table 3.3 shows the com- parison of values calculated by the SRK, PR, and BWRS equations of state. For the three the. difference in the calculated dewpoint temperature at 4 MPa pressure is trivial. The mixture analysis is probably not sufficiently accurate to warrant this degree of agreement. The difference In the bubblepoint temperature is approximately 10°C and the liquid formation is within 2%, both of which are outstanding agreement. TABLE 3.3 Comparative Calculations for Example Calculated Values Equation Dewpoint Bubblepoint Mol% Liquid of State ‘Temperature, °C ‘Temperature, °C Liquid @ 100°C SRK 193.4 64 186 PR 1323 $8 182 BwAS 1920 163 163 PFGC 1428 Ex 23.0 Table 3.4 compares bubblepoint and dewpoint calculations for the same three equations of state around essentially the entire pressure-temperature phase envelope for the mixture. Agait there is litle difference between the calculated values. The exception to this is that the BWRS equation predicts a lower bubblepoint temperature at higher pressures. 1.0 MPa higher than do the other two equations. It also extends the phase envelope approximately TABLE 3.4 Comparison of Phase Envelope Calculations ‘SAK PR BWRS PFGC Pressure kPa BP DP BP: DPt BP: DPt Bp | DP, 1000 “83.7 95.5 63 | 851 9 | 268 | 984 2000 21 | 1155 as | 562 | 1148 | 550 | 1207 3000 277 | 1266 1259 | 345 | 1257 | 203 | 1339 4000 64 | 1334 1323 | 163 | 1920 87 | 1427 5000 132 | 1372 1987 | 01 | 1351 | 109 | 1489 6000 24 | 1985 . 1965 | 149 | 1952 | 297 7000 519 | 1370 | 510 | 1944 | 202 | 1919 | 4a1 | 1540 2000 vo | 1313 | 726 | 1280 | 433 | 1239 | 672 i L beat Rriets bens Table 3.5 compares experimental determinations of vapor-iquld equilibrium constants with those calculated from the SRK, the PR, and the BWRS. The mixture is supposed to be a gas that Is in equillb- rium with a 103 MW lean absorber oll. The normal paraffin breakdown shown for the Cs+ components yields a molecular weight of 113.7, which is in good agreement with the observed value. However, If nor- ‘mal paraffins are used to represent absorber oil which really contains a significant quantity of naphthenes COMPUTER APPLICATIONS 323CALCULATION OF THERMODYNAMIC PROPERTIES TABLE 3.5 Comparison of K Values from Flash at ~40°C [-40°F] and 6.695 MPa [1000 psia] and perhaps some aromatics, the calculated K values for methane and ethane should be low. Overall, there Is relatively good agreement between the calculated and experimentally determined vapordliquid ‘equilibrium constants. Table 3.6 shows the comparison of calculated and experimental equilibrium constants from the same source for three different composttions of methane, ethane, and propane. Again, there is excellent agreement between the experimental and calculated vapor-iquid equilibrium constants. TABLE 3.6 Comparison of Predicted and Experimental K Values EXPY BWRS PR SRK ‘A. Bubblepoint temperature °C [°F] ofP = 9447 kPa [500 psia} | 90 [190] | 891-1281) | 90: [1280) | 299 [412871 B. Bubblepoint pressure kPa [psia] fT = 54.4°C [190°F] 3447 {500} | 3994 [483.5] | 9354 [486.9] | 3977 [489.8] ‘= Thermo Data Refined for LPG" by Stating, K.E., and Han, M.S. Hydrocarbon Processing (June 1972), p.107-186. Table 3.7 shows a ‘circular’ calculation of bubblepoint for a binary mixture of methane and propane. The experimentally determined bubblepoint conditions were 54.5°C [-190°F] and 3447 kPa [600 psia}. The calculation of the bubblepoint temperature at 64.4°C [-130°F] and the bubblepoint pres- sure at 3447 kPa [500 psia] shows only trivial diferences between the results calculated from any of the three equations of state, 324 YOLUME HICALCULATION OF ENTHALPY AND ENTROPY TABLE 3.7 “Circular Bubblepoint Calculation for Binary 94.8% Methane and 5.2% Propane Kevalues Mix Component | Mol% | _BWRS PR snk __| expe T = -101°C [-150°F] C1 44.48 | 10.5573 | 11.2285 | 11.72453 | 11.441 P = 721 kPa [32 psia] Ca 4750 | 0.23405 | 0.29640 | 0.2028 | 0.234 Cs oz | 0.01526 | 0.01432 | 0.01997 | 0.013 66°C [150°F] Gt 92.60 | 1.21044 | 1.18476 | 1.2068 1.162 2068 kPa [300 psia] C 249 | 0.06305 | 0.06629 | 0.0682 | 0.076 cs 4g1 | 0.00863 | 0.0075 | 0.00685 | 0.0103 24°C [75°F] co e450 | 1.2093 | 1.1623 | 1.15008 | 1.156 5516 kPa [800 psia] Ce 14.76 | 0.3863 | 0.44714 | 4454 | 0.449 3 o74 | o.tao23 | o21853 | o.2t609 | 0.203 Staring, KR. and Han, M.S, Hydrocarbon Processing (June 1972), p.107. CALCULATION OF ENTHALPY AND ENTROPY ‘The basic thermodynamic equations outlined in Volume 1, Chapter 7 are summarized in Table 3.8. ‘These serve as the basis for the calculation of enthalpy (H) and entropy (S) changes by utlizing the first and second laws of thermodynamics. Equations (e) and (f) in Table 3.8 show that H and S depend on P- V-T behavior and composition. Thus, the equations of state may be used to calculate the AH and AS re- quired to obtain work and heat quantities. TABLE 3.8 Summary of Basic Thermodynamic Equations H = U+PV @ GH = VéP + TAS (ay/ze)_ @ G = H-TS © dH = CpdT + [V-T(V/Tp] dP) dU = TdS = PdV + ete. (©) dS = (CpdTM) - (@V/8T)p] OP © To relterate the discussion in Chapter 7, Volume 1, itis convenient to apply the equations in two or more constant temperature and constant pressure steps. Take all temperature change steps at a low pressure where a gas behaves ideally or liquid compressibilly Is trivial. All pressure change steps are taken at constant temperature. Remember AH and AS are exact functions of P, V, and T whose change in value is dependent only on initial and final conditions and not on the path taken between them. ‘These steps can be visualized by tracing an enthalpy change process on a pressure-enthalpy dia- gram as shown in Figure 3.14. In this diagram we want to evaluate the enthalpy change In going from Point 1 to Point 2. The enthalpy change Is, obviously, hz-hr; for purposes of solution convenience, break the path Into three steps Aha, Ahp, and Aho. Adding these changes algebraically: Ah = he- hi = Ahe + Ahp + Ahe @.18) COMPUTER APPLICATIONS 325CALCULATION OF THERMODYNAMIC PROPERTIES 1000 100 Pressure 10 Enthalpy Figure 3.14 Illustration of the Path of the Calculation of Enthalpy Change Steps Aha and Ahe represent the Isothermal effect of pressure on enthalpy, while step Ahy repre- sents the isobaric effect of temperature on enthalpy. Most enthalpy difference calculation procedures calculate the isobaric effect of temperature on enthalpy by an equation of the form: Bh = hee = hi? ~ [oor (at @.19) where Cp°() Is a polynomial equation in temperature for the ideal gas. If the system being considered is ‘a mbdure, Cp°() is calculated for the mbdure by Cp°(t) = Zyi Cpt). The isothermal effect of pressure, on enthalpy (Aha or Ahe) is calculated by: (e—toynt = [TP -TePIMW av + 1-2 (2.20) Equation 3.20 can be developed by mathematical manipulation of the basic equation of state. Table 3.9 shows the form of the equations used for calculating the Isothermal effects of pressure on en- thalpy for the BWRS, the SRK, and the PR equations of state. ‘The foregoing discussion on enthalpy changes has assumed that the mbture was in the gas phase. If the mbdure is in the liquid phase, liquid enthalpies at each point are obtained by subtracting from the gas phase enthalpy the latent heat of vaporization. Otherwise, liquid and vapor enthalpy cal- culations are carried out in exactly the same way. Review of the enthalpy departure prediction equations shown in Table 3.9 will Indicate that the volumetric properties of the mbxture must be known before the enthalpy departure can be calculated. 326 ‘VOLUME IITCALCULATION OF ENTHALPY AND ENTROPY TABLE 3.9 Equation of State Formulas for Calculating Enthalpy BWRS H-H® = p(BoRT - Ao ~ 4Co/ T? + 5 Do! T?-6 Eo! T4) + 1/2 p? (2bRT - 3a - 4d/ T) = 115 p® a(Ga + 7d/T) + ofyT? [9 = (@ + 112-yp7~%" p%) exp-y 97) SRK -AH/AT = [AB -B/Rb] In(1 + B/Z) + 1-Z PR ees + [Geeta ne | 2y2b Z-0.414B Consequently, the basic equation of state must be solved for the appropriate volumetric property, either molar speotfic volume, density, or compressibility factor, before predicting the isothermal effect of pres- sure on enthalpy. This means that the enthalpy prediction calculation consists of a two-step procedure: (1) solution of the appropriate equation of state, (2) prediction of the Isothermal effect of pressure on the enthalpy. Obviously, the solution of the basic equation of state must be reliable if the enthalpy prediction is to be satisfactory. Entropy calculations are handled in the same fashion as enthalpy calculations. A two-step pro- cedure Involving the isobaric effect of temperature on entropy and the Isothermal effect of pressure on entropy is utilzed. The isobaric effect of temperature on entropy Is assumed to occur at a low pressure, and is calculated by: 21) ‘The isothermal effect of pressure on entropy can be calculated by: (6-S9R = I IRW- @P/TVI BV 22) Equation 3.22 is equivalent to: (S-S°VR = -Zy In(fY/Py) - AH/RT - In (P2/Pret) (23) Equation 3.23 is more convenient i K value and enthalpy departure calculations are being carried out at the same time the entropy calculations are being performed. One other term must be considered, the entropy of mixing: Smix® = -Eyiny: (8.24) Combining all of these terms in proper algebralc fashion yields the entropy change of the system in going from the ideal gas to the real gas or liquid state. Table 3.10 shows the equation-of-state forms for calculating entropy changes from the BWRS, the ‘SRK, and the PR equations of state. As for enthalpy, the discussion has assumed the gas mbture. COMPUTER APPLICATIONS 327CALCULATION OF THERMODYNAMIC PROPERTIES TABLE 3.10 Equation of State Formulas for Calculating Entropy BWRS $-$° = -3q Rin (Tp x) — plBoR + 2Co/T® - BDo/T* + 4Ed/T®) - 1/2 p? (OR + A/T?) + U5 pP ad! + 2chyT® [1 — (1 + 1/2 9? 9) expt-y PIL SRK “ASR = Exingi- AH/AT + Eq Ina + In PPO Ben ay"? se = anges [HOON] gf zee | Z+ Bl +¥2) ‘The use of equations of state to calculate AH and AS at elevated temperatures or pressures - or a cryogenic temperature -. may prove unsatisfactory unless the constants and parameters used were devel- oped using test data at these conditions. Similar problems may occur with systems containing large amounts of hydrogen, HS, CO2, and other molecules substantially different in behavior from paraffin hy- drocarbons. Combination Fi: -Enthalpy/Entropy Calculations The determination of system temperature or pressure required to yleld a specified total stream en- thalpy or entropy at some set pressure or temperature Is a common calculational problem faced by engl- neers in the natural gas industry. Even though the enthalpy/entropy calculations are not basic equi- librium calculations, the two are interrelated and must be solved simultaneously. (One of the major difficulties In performing enthalpy/entropy matching calculations Is estimating the phase condition at the final state. There is equal probability that the final condition could be in elther the single phase region or the two-phase region. Because of the anomalous phase behavior encountered in the retrograde region, or with hydrogen-containing systems, the traditional rules of phase behavior and the existence of single or two-phase behavior must be Ignored. To ensure total rellabllty, complete flash calculations must be carried out during the leration procedure of the enthalpy/entropy matching calcula- tion. To ilustrate the problems Involved, consider the phase envelope shown in Figure 3.15. The gas misture described by this envelope Is intially at point A (a dewpoint). It Is to have @ certain amount of energy added to It (at constant pressure) such that it wil end up at point B (@ point well within the two-phase envelope). (Note that this problem Is an enthalpy matching problem, the final enthalpy Is He = Q + Ha) Traditional rules would say that any time a dewpoint gas has energy added to It at. con- stant pressure, the gas will be In the superheated region. Clearly, this specialized case Is an exception to the rule. At final conditions, the gas Is In the two-phase region. Any solution based on the traditional rules would clearly be in error. If the complete phase envelope for the gas were known at the time the calculations were iniated, the problem could be solved by careful analysis of the phase envelope. However, determination of the complete phase envelope for all streams in a glven simulation is impractl- cal. Hence the recommendation that flash calculations be performed even though the system may be single phase. Admittedly, this policy causes extra calculations to be performed in many cases. However, the additional reliability of the procedure Is felt to be worth the incremental expense. VOLUME ItCALCULATION OF ENTHALPY AND ENTROPY Q added tna \ T;! 1B Pressure Temperature Figure 3.15 Example of Difficulties in Performing Flash Calculations ‘at Fixed Enthalpy Conditions ‘The basic problem to be solved in the enthalpy/entropy matching procedure Is Illustrated In Figure 3.16. The temperature (or pressure) which yields a fixed enthalpy/entropy at a specified pressure or tem- perature Is to be determined. Direct solution Is possible only if a diagram like Figure 3.16 Is available for the midure of interest. Unfortunately, the diagram generally Is unknown and a trial-and-error calculation must be used. Constant Pressure ‘Temperature Figure 3.16 Illustration of Calculation of Fixed Enthalpy/Entropy Flash COMPUTER APPLICATIONS 329CALCULATION OF THERMODYNAMIC PROPERTIES ‘The calculations Involved are llustrated by the logic diagram shown in Figure 3.17. The calculation procedure involves the following steps (consider the line A-B-C-D In Figure 3.18). 1. Set arbitrary upper and lower temperature limits for the calculation. 2. Perform a flash calculation at some assumed temperature or pressure and determine the en- thalpy/entropy. 3. Change the intially assumed temperature or pressure by a small amount and repeat tho fiash/enthalpy/entropy calculations. 4, Use the two points determined in steps 1-3 to linearly extrapolate to a third point that the con- vergence routine indicates will satisty the enthalpy/entropy match, and repeat the flash/enthal- py/entropy calculations. 5. If the enthalpy/entropy calculated in step 4 Is arbtrarly close to the desired value, calculations cease. If the enthalpy/entropy calculated in step 4 does not agree satisfactory with the desired value, calculations must continue. In this case, the linear extrapolation procedure Is. repeated with Ts becoming the new starting point for the linear extrapolation as shown in Figure 3.16. 6. The linear extrapolation procedure is continued until the calculated and specified en- thalpy/entropy agree within the specified tolerance. 7. AS each linear extrapolation is made, the temperature-enthalpy/entropy (pressure-enthal- py/entropy) limits are changed to provide a boundary for the next temperature assumption. If the linear extrapolation indicates going beyond the limits, the average of the temperature limits is used instead of the linearly extrapolated value. ‘The last step In the procedure Is required to handle a pure component or systems with narrow bolling range which behave like pure components. The enthalpy/entropy behavior of such a system Is shown by the line E-F-G-H on Figure 3.18. The F-G segment of the line Is vertical for a pure component and nearly vertical for a narrow bolling range mixture at high pressure. If the difference between the temperature limits is sufficiently small (say 0.01K) and the specified enthalpy/entropy value Is In the two-phase region, the fraction vapor and liquid can be estimated by: WF = (nt nYy(nY hh (8.25) or VF = (s*- sts’ -s) (8.26) and UF = 1.0-ViF (8.27) ‘Should the pure component or close bolling mixture be in elther single phase region, the linear ex- trapolation works quite well. The procedure is flexible and reliable. Because flash calculations are performed at each tempera- ture or pressure during the procedure, it Is a relatively time consuming calculation. However, the addl- tional expense is regarded as being trivial in comparison to the reliability. 330 VOLUME IICALCULATION OF ENTHALPY AND ENTROPY ENTER WITH Z,, 1, D OR x SET UPPER AND LOWER DEPENDENT VARIABLE Limits x¥, sxt, oY, ot CALL FLASH CALCULATION CALCULATE x™ ————>| RepLace xt WITH x ot with D REPLACE XY WITH x™, pU with D CALCULATE PHASE CONDITION BY INVERSE LEVER RULE Figure 3.17 Block Diagram for Enthalpy/Entropy Balancing Procedure COMPUTER APPLICATIONS 331CALCULATION OF THERMODYNAMIC PROPERTIES ENTHALPY /ENTROPY Bubble Point, Constant Pressure TEMPERATURE Figure 3.18 Difference in Vaporization Behavior of Pure Component and Mixture. SUMMARY ‘The foregoing discussion covers all of the principal factors involved in calculating equilibrium be- havior and energy levels from equations of state. Remember that all are empirical in nature and possess the limitations of all empirical equations. Although the equations may appear to be complex, they are not. They are merely tedious and thus ideally suited for computer solution. ‘Their value and true usefuiness Is proportional to the user's understanding of the physical system being calculated and the limitations of the program being employed. In addition to these uncertainties, there will be changes In both the analysis and quantity of gas with time. Both the magnitude and timing of the changes will depend on the reservoir mechanisms In- volved and the producing practices employed. by the operator. The uncertainty In these predictions usually is greater than that associated with equation-of-state calculations. ‘A“good decision can resuit only from calculations that realistically Incorporate the range of uncer- tainty of all parameters affecting the result. Throughout the rest of this book many such comparisons will bbe made for your guidance. 332 ‘VOLUME IiREFERENCES at 3.2 3.3 34 35 36 37 REFERENCES Chao, K. C. and Seader, J. D., AILCh.EW., Vol. 7 (1961), p. 598. Starling, K. E. and Han, M. S., Hydrocarbon Processing, Vol. 51, No. 5 (1972), p. 129. Starling, KE. and Han, M. S., Hydrocarbon Processing, Vol. 51, No. 6 (1972), p. 107. Redlich, O. and Kwong, J.N.S., Chem. Rev., Vol. 44 (1949), p. 233. Soave, G., Chem. Engr. Sci., Vol. 27, No. 6 (1972), p. 1197. Peng, D. Y. and Robinson, D. B., J. & E. C. Fund., Vol. 15 (1976), p. 59. Gibbs, J. W., Trans. Connecticut Acad. Sci., Ill (Oct. 1875-May 1876), p. 108 and (May 1877-July 1878), p. 343. COMPUTER APPLICATIONS 333NOTES: 334