16 REFRIGERATION SYSTEMS A refrigeration system lowers the temperature of a fluid below that possible when using air or water at ambient conditions. A typical building air conditioner cools air to a temperature of 10-15°C. At the other end of the scale is the liquefaction of helium at — 268°C. The temperature produced de- pends on the process objective. If the objective is to recover marketable liquids (NGLs) from a pro- duced gas stream, basic economics controls the temperature specified. If it is to meet a hydrocarbon dewpoint, that specification and the processing pressure sets the required temperature. Several basic processes will be discussed herein. 1. Mechanical Refrigeration a. compression b. absorption 2. Valve Expansion 3. Turbine Expansion MECHANICAL REFRIGERATION A refrigeration system is a heat pump. Low temperature heat is removed from the process fluid and is “pumped” to high temperature (ambient) where it is rejected to the environment. Energy is required to pump heat. The amount of energy depends on the quantity of heat to be pumped (chiller duty) and how far the heat has to be pumped (temperature difference between the chiller and the con- denser). Energy used to drive a refrigeration process can be in the form of heat or work. The absorp- tion refrigeration systems use heat to pump heat, compression systems use work Ammonia Absorption System Figure 16.1 shows a flowsheet for a refrigeration system utilizing using two concentrations of ammonia-water solutions. The basic driving force is the heat input to the generator. Ammonia vapor is stripped from the water solutions in the rectifier or stripper. This ammonia vapor is condensed and passes through a receiver, a heat exchanger (optional) and across an expansion valve into the evapora- tor or chiller. Here it vaporizes while cooling the fluid to be chilled. The ammonia vapor from the evaporator is absorbed in a weak ammonia-water solution. The result is a strong solution. The ammonia is removed from this strong solution in the generator to begin its cycle all over again. CHAPTER 16REFRIGERATION SYSTEMS Rectifier NH Liquid NH Liquid ‘Aqua Film ‘Absorber Figure 16.1 Flow Sheet of an Ammor Absorption System One can write selected energy balances around this system to determine loading at various points, The starting point is the evaporator. First, determine the total duty required to chill the fluid to its desired temperature. This is the heat load in the evaporator. The heat absorbed by the ammonia per unit mass is governed by evaporator pressure or tem- perature. The AH is the enthalpy of a saturated ammonia vapor at evaporator conditions minus the enthalpy of the entering ammonia, which should be a saturated or subcooled liquid. A series of such balances, and an overall balance, enable one to determine sizes and energy loads of each component part. Reference 16.1 summarizes the typical loadings for the ammonia sys- tem in Figure 16.1. The heat for the generator may be obtained from any one of four sources: (1) low pressure steam, (2) fired heater, (3) a hot process stream and (4) waste heat. Since heat loads are large, the availability of (3) or (4) increases the economic attractiveness of this system. Water cooling is shown for the condenser and absorber but air cooling may be used. As with all refrigeration systems, the higher the cooling temperatures the lower the process efficiency and greater the energy input. Units have been designed with capacities to about 35 MW [10 000 tons of refrigeration (TR)} The unit is simple and has few moving parts to maintain, Ammonia solutions are not difficult to han- dle metallurgically. A major psychological problem is ammonia smell. In a confined space this can be a nuisance. However, the pungent odor is a safety item that immediately confirms leaks. ‘ Though not widely used in gas processing applications, ammonia systems may be competitive with compression systems, particularly in those situations where significant quantities of waste heat are 214 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION available. Operating cost comparisons are dependent on the source of heat and the cost of cooling. If the economics are competitive, these systems are a viable alternative to compression systems for pro- ducing refrigeration. COMPRESSION REFRIGERATION Compression refrigeration is by far and away the most common mechanical refrigeration pro- cess, It has a wide range of applications in the gas processing industry. + Chilling natural gas for NGL extraction * Chilling natural gas for hydrocarbon dewpoint control + LPG product storage + Condensation of reflux in deethanizers/demethanizers + Natural gas liquefaction (LNG) Figure 16.2 shows a simple single-stage compression refrigeration system. Saturated liquid re- frigerant at Point A expands across a valve (isenthalpically). On expansion some vaporization occurs. The mixture of refrigerant vapor and liquid enters the chiller (sometimes called the evaporator) typi- cally 3-6°C [5-10°F] lower than temperature to which the process stream is to be cooled. The liquid vaporizes. Leaving at Point C is a saturated vapor refrigerant at the P and T of the chiller. This vapor is compressed and then enters the condenser as a superheated vapor. Compressor Y D a) —pde- — -| 1 Expansion ¢ A SS Valve Condenser Accumulator Figure 16.2 Flow Sheet of a Simple Refrigeration System The refrigerant leaves the condenser as a saturated liquid or slightly subcooled. For air cool- ing, the condenser temperature will usually be 14-16°C [25-30°F] above the air dry bulb temperature. For water cooling the condensing temperature will be 5-10°C [9-18°F] above the water temperature. The accumulator, sometimes called a surge tank or receiver, merely serves as a reservoir for refrigerant as levels vary in the chiller(s) and condenser. CHAPTER 16 215REFRIGERATION SYSTEMS. 4 Calculation of a Simple System There are several discrete steps in the sizing of the system shown in Figure 16.2. These are summarized below: 1, Determination of refrigerant circulation rate. Develop an energy balance around ——s> the chiller and expansion valve shown at right, At point A the re- frigerant is a saturated liquid (or very close to it). At point C it is a saturated vapor. Qcritle is the chiller duty and is set by the process re- Ly —pqes— quirements? If one writes an energy balance around the system, Qohitter + Maha = Mche. But Ma = Me =m, so Qeviter_ (16.1) he-ha peace SLs _FPS Where: Qeritler = chiller duty kw Btwhr he = saturated vapor enthalpy kkk Btu/lbm ha = saturated liquid enthalpy kdhkg Bru/lbm m = refrigerant circulation rate kg/s Ibnvhr 2. Determination of Compressor Power. This is done by any appropriate method as outlined in Chapter 15. Calculate theoretical (isentropic) work and use an efficiency to find actual work, The circulation rate from Step (1) is used, _w = m(hp” ~he) . E 3. Determination of Condenser Heat Load (Qcond): (16.2) eae Ww There are two ways to do this. Knowing Qchinter and W, you can write the overall balance for Figure 16.2 to find Qoong- (Remember, W is negative, so Qeona is the sum of two negative numbers) Qeond = W-Qeritter (16.3) Qoone If you are performing the calculation manually / and wish an independent check of the previ- E ‘ous work, write the balance shown below. a 0 Qeond = M(ha —hp) (16.4) — 216 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Determination of the Enthalpies The calculation requires one to find the enthalpy per unit mass at points A, B and C. These can be found from a computer program or from tables and figures published for all of the common commercial refrigerants. Appendix B at the end of this volume contains data on some substances used as refrigerants. Appendix 16A at the end of this chapter contains pressure-enthalpy (P-H) figures for propane and Refrigerant 22 (Freon-22) as well as vapor pressure and physical property information on all common refrigerants. The P-H diagram is convenient for solving the energy balance for a simple system. The lefi-hand figure is a simple representation of the P-H diagrams in Appendix 164. The re- frigerant is liquid to the left of the saturated liquid curve, two-phase inside the envelope and vapor to the right of the saturated vapor curve. The lines of constant temperature (isotherms) are horizontal be- tween the saturated vapor and liquid curves and then rise almost vertically in the liquid region. The calculation process starts by choosing the temperature of point A. Will water, air or some process fluid be used for condensation of the refrigerant? What temperature can we realistically achieve in the condenser? This temperature sets point A. It is on the liquid saturation curve, since it leaves the condenser as a liquid. What is the temperature at points B and C? Normally, it will be 3-6°C [5-10°F] less than the minimum desired temperature for the fluid being cooled. This approach fixes the location of point C. It is on the saturated vapor curve, since it is in equilibrium with the liquid in the chiller (evaporator). The expansion across the valve from point A is an isenthalpic process; a vertical line on a P-H diagram. Draw a vertical line from A to B, the pressure of point C, and then go horizontally to C. One can read the Ah required for Equation 16.1. The theoretical compression is isentropic. Starting at point C, follow a constant entropy path until you intersect the pressure line of point A. This is theoretical (isentropic) discharge point D. Compressor power is calculated from Equation 16.2. The condenser heat load can now be determined from Equation 16.3. For a commercially pure refrigerant, use of a P-H diagram or a corresponding table is as reli- able as any method. CHAPTER 16 a7REFRIGERATION SYSTEMS Example 16.1: SI Solution: FPS Solution: Calculate the refrigerant circulation rate, compressor power, and condenser duty for this system. Use the P-H diagram for propane on page 258 or 259. A refrigeration system is designed to refrigerate a natural gas stream for the pur- pose of NGL extraction. The following information applies. Chiller duty 2MW [6.8 x 10° Btw/hr] Gas temperature exiting chiller — 15°C [5°F] Air temperature 35°C [95°F] Refrig. Compressor efficiency 15% Chiller temperature is 3-6°C below required process temperature. Use ~ 20°C, note from Figure 16A.1(a) chiller pressure is 240 kPa, Condenser temperature is 14-16°C above air temperature for aerial cooler. Use 50°C. Note from Figure 16A.1(a) condenser pressure is 1750 kPa, Point A is a sat. liquid at 50°C, from Figure 16A.1(a) ha = 33 ki/kg Point C is a sat. vapor at - 20°C, from Figure 16A.1(a) he = 250 kJ/kg Refrigerant circulation rate, m: Q& 2000 ki/s = SE 292 e-ha (250-33) ki/kg ke/s Compressor power, W: hi" =341 kI/kg @ s=5.66 and P=1.75 MPa 9.2 (341-250) W = = 1120 0.75 eet Condenser duty, Qh: Qh = W-Qc =-1120-2000= -3120 kw Chiller temperature is 5-10°F below process temperature. Use -5°F, note from Figure 16A.1(b) chiller pressure is 34 psia.. Condenser temperature is 25-30°F above air temperature. Use 120°F, note from Figure 16A.1(b) condenser pressure is 240 psia.. Point A is a sat. liquid at 120°F, from Figure 16A.1(b) ha = — 780 Btu/lbm Point C is a sat, vapor at ~5°F, from Figure 16A.1(b) he = -686 Btw/lbm Refrigerant circulation rate, m: 6.8x10° Btu/hr -=— T= 72340 tbm/h (780 -(-686)] Bru/lbm i VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Example 16.1 (Cont'd): Compressor power, W: hS™ =—646Btu/lbm @ s= slightly less than 1,36 and P = 240 psia _ 72 340[-646 —(-686)] oS 15: (2545)(0.75) Condenser duty, Qh: Qu = W-Qe =(-1515)(2545) ~6 800 000=—10.7 x10° Btu/hr In the above examples, pressure drops were ignored. In an actual system some pressure drop occurs between the compressor discharge and the accumulator as well as between the chiller and com- pressor suction. These pressure drops increase the required compressor power. Pressure drop on the suction side of the compressor has the greater effect. Economizer Systems In the simple refrigeration system presented in Example 16.1, a significant amount of vaporiza- tion occurs across the expansion valve. Unlike the multicomponent phase envelopes discussed in Chap- ter 4, the quality lines inside the two phase region in Figures 16A.1 (a and b) are equidistant. Therefore, it is relatively easy to calculate the quality of a two-phase mixture. h x= liquid fraction = —Y (16.5) Vv y = vapor fraction = be abe (16.6) hy -hy Where: hg = two-phase enthalpy hy = saturated liquid enthalpy hy = saturated vapor enthalpy For Example 16.1, by Example 16.2: Calculate the fraction of liquid entering the chiller in Example 16.1. SI Solution: hy =33 kJ/kg, hy =250 kJ/kg, hy =-150 kJ/kg 250-33 _ 54 250-(-150) FPS Solution: hy =-780 Btu/lbm , hy =—686 Btu/Ibm , h, =-860 Btu/lbm ~686-(-780) _ 9 54 —686 —(-860) CHAPTER 16 219REFRIGERATION SYSTEMS. From Example 16.2, the quantity of vapor entering the chiller is 46%. This vapor provides no refrigeration effect in the chiller, but must be compressed from chiller pressure to condenser pressure, increasing the compressor power requirement. In refrigeration systems the word “economizer” refers generally to any device or process modi- fication which decreases the compressor power requirement for a given chiller duty. Two types of economizers are commonly used — flash tank economizers and heat exchanger economizers. Figure 16.3 shows a refrigeration system employing one flash tank economizer. In this system, the saturated liquid refrigerant leaving the accumulator is expanded across a valve to an intermediate pressure where vapor and liquid are separated. The separator liquid is expanded across the second valve to chiller pressure while the separator vapor goes to the compressor interstage. The refrigerant entering the chiller now has a higher liquid content. This reduces the refrigerant circulation rate through the chiller as well as the first stage compressor power requirement. 1st Compression 2nd Compression Stage Stage 5 aati Expansion Condenser Valve E Flash Tank (Economizer) A ‘Accumulator Figure 16.3 Flow Sheet of a Refrigeration System with an Economizer The P-H diagram for the economizer system in Figure 16.3 is shown at left. Point E is usually fixed at a pressure that results in equal compression ratios in each stage of compression. Expansion across the economizer expansion valve is isenthalpic. ‘The vapor formed goes to the second stage of compression as a saturated vapor. The saturated liquid, E, leaving the separator is expanded isenthalpically to pressure B. Notice that the Ah available from B to C for this system is larger than for the simple system. 220 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Calculation of Flash Tank Economizer Systems Determination of circulation rate, work and Qeona for the system in Figure 16.3 follows the same pattern as for the simple system. The first step is the same in principle. Now, however, the enthalpy of the refrigerant entering the chiller valve is determined by compressor interstage pressure and not by condenser pressure. With this change in ha, Equation 16.1 may be solved for “my,” the circulation rate to the chiller (and the amount of vapor to be compressed through Stage 1 of the compressor). What is the refrigeration circulation rate through the second stage? It is determined from a material balance around the economizer flash tank. The value of m, is known from the circulation rate calculation at the chiller. ft 7 i im, The second stage circulation rate, ma, may be calculated from the relationship m m, (16.7) x where x is the liquid fraction in the economizer vessel and is cal- culated from Equation 16.5. a 5 In the compressor power calculation, one finds the work ™; ‘ in the first stage (for flowrate “m,”) and adds it to second stage work (for flowrate “m,”) to find total work. The resultant temperature of the gases after they mix in the tee between stages may be found by an energy balance around that tee. In most cases, the tempera- ture effect here is small. The condenser heat load, Qeond, is found as before. Flash tank economizer systems are not limited to two stages (one economizer). Many installa- tions employ two economizers (three compression stages). An example of such a system is shown for the propane portion of the cascade refrigeration system in Figure 16.8. Anothér advantage of a flash economizer system is the ability to install chillers on the interstages. Figure 16.8 shows this capability. Rather than provide all of the refrigeration at — 40°C [-40°F], chillers are installed on the low-stage economizer (-19.9°C [-3.9°F]) and high-stage economizer (6.6°C [43.9°F]). While this scheme is not often used in a straightforward gas chilling plant it is frequently used in more complex processes employing significant heat integration. CHAPTER 16 221REFRIGERATION SYSTEMS. Example 16.3: Rework the simple refrigeration system in Example 16.1 assuming a two-stage system with one flash tank economizer. SI Solution: Estimate economizer pressure assuming equal compression ratios per stage. os Pecon = Pa (®) = [P,P 2 IPS = [1750)(240)]°* = 648 kPa (use 650 kPa) Note fromi Figure 16A.1(a) the economizer temperature is approximately 10°C. Ist Stage: Refrigerant circulation rate, m, me eecs he = 250 ki/kg (sat. vapor @ —20°C) he-hg hy =—75 kJ/kg (sat. liquid @ 650 kPa) 2000 = GIS ky ™ = 359-78) O15 HIS Compressor power, W; refrigerant is compressed in Ist Stage from 240 kPa to 650 kPa -w, = mAh" _ 6.15(295—250) _ 569 pw Ez 0.75 2nd Stage: Refrigerant circulation rate, mr @ economizer (650 kPa) ety rhe __ 282-33 _ hy—h, 282 -(-75) 15 =8.75 ky 7 s/s Cotpressor power, W2 Note: It is difficult to locate the actual second stage suction point on the P-H diagram. For purposes of this exercise it has been assumed that Anis" is 45 kJ/kg. This results in a total Ahi" of 91 ki/kg which is identical to the single stage system, mab _ (8.79945) _ kw E 075 -W, Total compressor power = 369 + 537 = 908 kW, a reduction of 18.9% from the simple system. Condenser duty, Qh Qy = W-Qe =-904-2000= 222 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Example 16.3 (Cont'd FPS Solution: Estimate economizer pressure assuming equal compression ratios per stage. os Pecon = Pi () = [1 )P2 1% = (0240)4))* 7 = 90 psia Note from Figure 16A.1(b) the economizer temperature is approximately 50°F. | _Ist Stage: Refrigerant circulation rate, m, he =— 686 Btu/Ibm (sat. vapor @ - 5°F) hy =~ 828 Btu/lbm (sat. liquid @ 90 psia) = 47900 Ibm/hr m, ~ [686 —(—828) Compressor power, W; refrigerant is compressed in Ist Stage from 34 psia to 90 psia mAh _ 47900 [-667 -(686)] =w, = REN _ 47.900 [667 —(686)) «477 4y : 1 = @545\E) (2545)(0.75) id 2nd Stage: Refrigerant circulation rate, my @ economizer (90 psia) =hp. _(673)~(-180) _ 5 6g hy—h, (-673)—(828) my = = 47900 «69.490 tbm/br x 0.69 Compressor power, W, Note: It is difficult to locate the actual second stage suction point on the P-H diagram, For purposes of this exercise it has been assumed that Ahi is 21 Btwlbm. This results in a total Ah" of 40 Btu/lbm which is identical to the single stage system. = MANN _ (69 400)(21) _ A 7 = 764 hp E (2545)(0.75) Total compressor power = 477 + 764 = 1241 hp, a reduction of 18.1% from the simple system. Condenser duty, Qh Qh = W -Qe =(-1241)(2545) —6 800 000: 996 x10° Btu/hr CHAPTER 16 223REFRIGERATION SYSTEMS: Heat Exchanger Economizer A second type of economizer configuration is the heat exchanger economizer shown in Figure 16.4, Cold, low-pressure chiller vapor is used to subcool the saturated liquid refrigerant. This de- creases the refrigerant circulation rate, and may reduce compressor power. Compressor D<] Expansion A Valve ‘Condenser ( Accumulator 1) Figure 16.4 Flow Sheet of a Simple Refrigeration System with a Heat Exchanger Economizer With regard to the compressor power, two factors offset the reduced circulation rate. The first is exchanger pressure drop. The pressure drop on the low pressure side of the exchanger may be 20-50 kPa [3-7 psi]. This can significantly increase compressor power, particularly when the chiller pressure is near atmospheric. Secondly, the refrigerant vapor entering the compressor is now superheated. Al- though this reduces the likelihood of liquid carryover into the compressor, it results in higher power consumption per unit mass due to the higher suction temperature. This can be seen in Figures 16A.1(a) and 16A.1(b). Note the constant entropy lines are less steep at higher temperatures. Calculation of Chiller Load (Qctitler) The chiller duty, Qehtien is set by process requirements and is determined by the energy bal- ance calculations presented in Chapters 8 and 13. Figure 13.17 on page 93 of Chapter 13 is a good ex- ample of a refrigeration application for natural gas chilling, Specification of Cold Separator Temperature The desired temperature of the gas-liquid stream leaving the chiller must be determined by the strategy governing the system. If hydrocarbon dewpoint control is the primary process objective and liquid recovery secondary, this temperature is often 3-5°C [5-9°F] below the temperature required to achieve the specified dewpoint to account for liquid carryover out of the separator. This required tem- perature may be significantly colder than the specified dewpoint if the gas is processed at high pres- sure. (See Chapter 4, page 103) If liquid recovery is the primary function of the unit, what products can be sold? The basic strategy is to condense the least amount of nonsaleable components (usually methane and/or ethane) compatible with the economic criteria. Anything condensed, like methane, that must be revaporized and maybe also recompressed, adds to the operating costs without contributing to liquid revenue. VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION For a given set of specifications, one should investigate a series of cold separator pressures and temperatures. The pressure for maximum liquid recovery is between 3.0-4.0 MPa [435-580 psial, if Cy is the saleable product. As the pressure increases, condensation of methane, ethane, and CO) in- creases. However, 3.0-4.0 MPa [435-580 psia] may not be the optimum pressure economically because of compression requirements. The optimum pressure must minimize total system cost, not merely that of the refrigeration and stabilization system alone. Separation is frequently carried out at the sales gas pressure to eliminate recompression. For the usual pressures chosen, what is a reasonable temperature? As noted before, this de- pends on the products desired. If the liquid product must have an RVP less than | atm. or is blended \with crude oil or condensate sold as crude oil, the common separation temperature may be ~ 20 to 0°C {-4 to 32°F]. When propane is the lightest saleable liquid, the temperature may be — 40 to - 20°C [40°F to - 4°F), Choice of temperature (and pressure) is a critical specification. Do not choose arbitrarily? Calculate the economic criteria for a series of conditions and choose the optimum one. Choice of Refrigerant The ideal refrigerant is nontoxic, noncorrosive, has PVT and physical properties compatible with the system needs, and has a high latent heat of vaporization. Many substances could be used as a refrigerant, The practical choice reduces to one which has desirable physical properties and will vapor- ize and condense at reasonable pressures, at the temperature levels desired. In gas processing applica- tions the usual choice is propane, ammonia, or R-22 at chiller temperatures above about ~ 40°C [40°F]. At cryogenic conditions, ethylene and methane might be used. In general, the lower practical limit of any refrigerant is its atmospheric boiling point. It is desirable to carry some positive pressure on the chiller to obtain better efficiency in the compressor, reduce equipment size and avoid air induc- tion into the system. Vacuum operation is feasible however, and has been successfully implemented in some systems, Different refrigerants will exhibit somewhat different power requirements per unit of chiller duty. These differences are typically small, < 10%. The primary factors in refrigerant selection are initial and replacement cost, availability, health and safety considerations and environmental impact. Propane’ is by far the most popular refrigerant in the gas processing applications. It is readily available (often manufactured on-site), inexpensive and has a “good” vapor pressure curve. It is flam- mable but this is not a significant problem if proper consideration is given to the design and operation of the facility, Freons are widely used as commercial refrigerants. They are non-toxic and nonflammable. CFC (Chlorofluorocarbon) refrigerants like R-11 and R-12 have been phased out due to environmental problems. The replacement for R12 is R-134a. HCFC refrigerants (Chlorofluorocarbons containing at least one hydrogen) are currently considered environmentally acceptable in most locations but the phase-out of some of these refrigerants has been mandated. R-22, an HCFC, is the most commonly used Freon in gas processing. Freons are expensive and system losses can represent a significant oper- ating cost. They are also difficult to ship to remote locations in large quantities. Certain Freons will form hydrates so it is necessary to keep the system dry. Many replacements for pure CFC or HCFC refrigerants are mixtures. Some of these are aezeotropes (these are designated R-5xx) and some are zeotropes (designated R-4xx). Aezeotropes act as a pure component in the refrigeration system. Zeotropes do not. Zeotropes do not have a constant CHAPTER 16 205REFRIGERATION SYSTEMS boiling temperature at a fixed pressure, i.e. the isotherms are not horizontal inside the phase envelope. The temperature difference between the saturated liquid and saturated vapor is referred to as “tempera: ture glide” and may be 3-5°C [5-9°F]. For kettle-type chillers, this effect increases compressor power relative to pure refrigerants. As of this writing the only replacement CFC/HCFC/HEC refrigerants de- veloped for R-22 are zeotropes. Table 16.1 presents the approximate compressor power requirement and condenser duty for propane, ammonia, and R-22 at various chiller and condenser pressures. TABLE 16.1 Comparison of Common Refrigerants ___ Evaporator emperature,*C=>| 4g | 40 | se | ae | as | as | on | + 4 Evaporator pressure | Ammonia | 529 | 716 | 95a | 126 | 163 | 210 | 265 | 3m 0 inka Propylene | mt | 1s | 17 | a | 29 | a | 400 | a8 «6 Propane | s6a_|_ur2_ |__| 176 | m6 | 263 | 320 | 06 31 [- Condensed Liquid Temperature 35°C; Condenser Pressure in KPa: Ammonia 1357; Propylene 1461; Propane 1220 grefigerant per [Ammonia | 0967 | 0958 | 0980 | ost | oss | os2s | 0922 | ovis | osm | ase second per MW of | mopyiene | 413 | 402 | 383 | 30s | 328 | 370 | 363 | 337 | 350 | 3a refrigeration Propane | 426_| ais | 40s | a9 | 393 | a7 | 367 | 359 | 350 | 30 wih ofretigenn | Ammonia | 7197 | s361 | aos? | aus | 2sis | 91s | isiz | im7 | oes | 19 Pe MW of Propylene | 796 | 4ase | 3526 | 2826 | 22u9 | ess | soz | i222 | 102 | 69 retigeation rronane | 7487_| 3706 | asz6_| 25m | 2707 | 2304 | sto | 1507 | 12st_| 10 Kilowar perMWot [Ammonia | 793 | ois | soa | sur | an | 377 | ais | 267 | zis | in refigeraton Propylene | 7448 | 657 | sv | ae | 4x7 | 309 | 310 | 256 | 22 | 176 | Propane | 727 | o2_| 566 | aor | ao | sis | 263 | 28 | wo (Condensed Liquid Temperature 52°C; Condenser Pressure in KPa: Ammonia 2088; Propylene 2164; Propane 1792 kg refrigerant per ‘Ammonia | 1.047 | 1.038 | 1.027 | 1.019 | 1.008 | 1001 | 0995 | 0989 | 0982 | 097% second per MW of Propylene | 5201 | soso | 4900 | 4771 | 4642 | 4534 | 4427 | 4319 | 4234 | aes ae Propane | s4s7_| 524 | sor | aoa | @ aoa | ass | ase | 4277 | a1 mimotretigean | Ammonia | 7760 | $796 | 439 | 3367 | 2008 | 2058 | 1633 | 1314 | 1058 | a9 Per MW of Propylene | 7408 | 3555 | as4o | 3536 | 2826 | 2290 | iso | uss | 1256 | 1089 iaceee Propane | _osis_| ras | sooo | aire | 347 | ase | ants | esi |_1si7_ | i270 Kilowatt perMWof | Ammonia | 1052 | 929 | aos | 706 | 19 | sxe | 46 | 403 | 346 | 203 refrigeation Fropyiene | 2s | 999 | sr | mm | os | sox | su | ax | arr | 3 Propone _|_o9e_| 97s] sor_| 76 | oe _|_ 596 | sis_| ao | 375 | is | _Evaporator Temperature, "P= | 9 | 40 | 30 | 20 ° w | 2» | 0 | «0 Evaporator pressure in | Ammonia | 767 | 104 | 139 | 183 soa | 385 | 480 | 597 | 130 a Propylene | 161 | 207 | 260 | 321 aso | sso | 700 | as | 960 Propase | 126 | 162 | 205 | 255 sei_| 64 | soo | 673 | 00 ‘Condeased Liquid Temperature 95°F; Condenser Pessre in pi: 197; Propylene 212; Propane 177 Pounds ofreigernt | Ammonia | 04s | o4as | oat | o438 | 043 | o4n2 | 0429 | 0426 | 04m | 0422 ermine pertonof | propylene | 192 | 17 | 1s | ie | 176 | a | 16 | 166 | 10 | 160 ‘etigertion Propane | 98 | 19s | tas | tas | are | ae | am | rer | 10 | 19 CFM ofretigerant per | Ammonia | 1490 | 1110 | 840 | 64s | 500 | 396 | 313 | 252 | 200° | 1.70 minute pertonof | propylene | 1200 | 918 | 730 | sas | are | age | aur | 253 | 202 | 180 eetenciis Propane | 1850 | 1200 | 937 | 729 | 19 | am | 387 | 312 | 259 | 205 Brake horsepower per | Ammonia | 374 | 323 | 280 | 241 | 208 | 178 | 150 | 126 | 103 | oms tonofretigetion | Propylene | 351 | 310 | 269 | 235 | 206 | 14 | 146 | 120 | 100 | oo Propane 343_| 303 | 267 | 232 | 203 | 195 | 149 | 124 | 101 | oso (Coodensed Liquid Temperature 125°F; Condenser Pressure in psis:_ Ammonia 303; Propylene 314; Propane 260 Pounds ofeftgeant | Ammonia | 0487 | 0483 | o47e | a47s | 0469 | 0465 | 0463 | 0460 | 0457 | o4s4 pecminatepertonot | Popyine | 242 | 235 | 228 | 2m | 216 | an | 20 | am | 197 | 19 Tatigeaton Propane | 259 | 244 | 236 | 229 | 222 | 246 | 210 | 208 | 199 | 19 CrM ofrefageant per | Ammonia | 16:10 | 1200 | 909 | 697 | 540 | 425 | 338 | 27 | 219 | 190 ton ofrefigeration | propylene | 1540 | 11so | 940 | 732 | sas | 472 | 3s | 308 | 260 | 218 Propane | 1970 | 1540 | rao | 916 | 720 | soe | 479 | 379 | 314 | 263 Brake honepowerper | Ammonia | 496 | 438 | aa1 | a3s | 22 | ase | 219 | 19 | 10 | 138 tonofretigeation | propyiene | 32 | an | 414 | 366 | 323 | 279 | 24 | 203 | 178 | 15s | Prine J ss 660 [406 | 959 | aus | 2a | ae | 207 | am | 10 226 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Compressor choice is linked to refrigerant choice as well as other considerations. Where weight and size are particularly important, a centrifugal or a screw compressor may be used. A recip- rocating compressor is an alternative for accessible land locations. Lube oil selection is particularly important for positive displacement machines. Lube oil carried into the refrigerant can congeal on chiller tubes reducing heat transfer rate. Effect of Temperature on Cost Figure 16.5 shows the approximate relative effect of temperature on refrigeration cost. The in- set to Figure 16.5 shows the refrigerant often used at various temperature levels. ‘The temperature lev- els are approximate. Actually, the refrigerant used at a given level depends on economic criteria, which will vary in different circumstances. Temperature, °C -250 200-150-100 TR Wet LLL 19.900 poy 2888 FT] 6000 f 4000 [1 2000 1900 800 Relative Cost of Refrigeration (Cooling Water = 1.0) Temperature, °F Figure 16.5 Relative Cost of Refrigeration Compared to the Cost of Cooling with Water at Ambient Conditions Operating Issues Refrigeration power, hence operating cost, is minimized by minimizing the condensing temper- ature and operating the chiller as close as possible to the process fluid temperature, i.e. minimum ap- proach. Generally, the condensing medium will be water or air, but in some applications a cold process fluid may provide a lower temperature condensing medium. Some gas processing facilities use the sales gas as the cooling fluid in the condenser, particularly if the inlet gas to the facility is below ambient temperature due to ground or seabed cooling. CHAPTER 16 227REFRIGERATION SYSTEMS Chiller fouling increases the required temperature difference between the process fluid and the refrigerant, and forces the refrigeration system to work harder to deliver the same amount of refrigera- tion. Fouling can occur on the refrigerant side of the chiller due to contaminants in the refrigerant — usually lube or seal oil from the compressor. Fouling can occur on the process fluid side due to hy- drates, wax, or freezing of TEG carried over from a glycol dehydration system. In climates where significant ambient temperature variations occur between summer and win: ter, it may be necessary to restrict the condenser cooling on the coldest winter days. There ate to reasons for this. First, sufficient pressure differential must exist between the accumulator and the chiller to cause the refrigerant to flow through the piping and valves. This is particularly true in economizer systems. Secondly, if the refrigerant compressor is a centrifugal, a low discharge pressure may cause the compressor to operate in stonewall. The most common operating problems in refrigeration systems are refrigerant losses, usually due to leaks and refrigerant contamination. Refrigerant losses occur due to leaks at flanges, valve packing, relief valves, etc. If Freons are used, installation of rupture disks below the relief valves can help reduce losses. In addition to lubricating oil, refrigerants can be contaminated by non condensables, usually methane or air, Methane can contaminate the refrigerant due to a leak in the chiller. Air can enter the system if the suction side of the compressor operates at vacuum. Both re- duce the efficiency of the refrigeration system because they artificially increase condensing pressure The primary objective of refrigeration system control is to cool the process fluid to the desired temperature (in some cases the desired temperature is the coldest temperature attainable given the com- pressor power and/or metallurgical constraints). The primary load change on the system is typically chiller duty, usually due to process fluid rate changes. The amount of heat transferred in the chiller is given by the relationship Q=UAAT,, For control purposes, either area, A or temperature difference, AT;, can be manipulated to con- trol Q. Area is manipulated by changing the level of refrigerant in the chiller and temperature difter- ence by changing chiller pressure (hence the boiling temperature of the refrigerant). The latter is preferred in most applications. A typical control scheme is shown in Figure 16.6. Compressor capacity is controlled using the methods presented in Chapter 15. ° ee eee Figure 16.6 Example Control Scheme for a Chiller 228 VOLUME 2: THE EQUIPMENT MODULESCOMPRESSION REFRIGERATION Anti-surge control for centrifugal type refrigeration compressors is somewhat different than the natural gas compression due to the phase behavior of the refrigerant. Cooling of the recycle condensation causes condensation of the refrigerant. A common recycle control scheme is shown in Figure 16.7. © Compressor Condenser ‘Accumulator 1 ‘Refrigerant to Chilier or Economizer Figure 16.7 Anti-Surge Control on a Refrigeration System The primary recycle flow comes from the hot discharge gas leaving the compressor. Without cooling this would quickly cause a high discharge temperature trip on the compressor when operating in recycle. Liquid refrigerant is injected into the recycle line to cool the recycle gas. The liquid rate is controlled by the suction temperature. Cascade Refrigeration When refrigeration must be provided at very low temperatures, below about < — 40°C [- 40°F], cascade refrigeration systems are sometimes used. Cascade systems employ more than one refrigerant and provide refrigeration at multiple levels. A propane/ethane cascade system is shown in Figure 16.8. In this system, refrigeration is provided at five levels Gaal o 2 4 6 8 10 12 14 16 18 20 100) ® Fa a 10 ’ o * 20] m10 © § 0 -20 § aH 8 20 |-30 & 3-40 40 8 bs ree The propane at - 40°C is used to condense" -80 pean the ethane refrigerant. All of the heat picked up in ~109 an the process is ultimately rejected to the cooling JL gg ke d-90 water at the propane condenser. A hypothetical 0 10 2 30 4 50 60 70 cooling curve for the process fluid has been devel- , MMBtumhr oped to show the amounts and levels of refrigera- tion, CHAPTER 16 229REFRIGERATION SYSTEMS 1.4 bara [15.7 psia] ~40.0°C [40.0] Ethane System 1.0 bara [142 psial Figure 16.8 Ethane-Propane Cascade Refrigeration System (Courtesy GPSA Databook) Cascade refrigeration systems employing propane, ethylene, and methane are used in ethylene plants and some LNG processes but are generally not common in gas processing. Low level reftigera- tion is typically provided using expansion or mixed refrigerants. Mixed Refrigerants ‘An alternative to cascade refrigeration is to use a mixed refrigerant. Mixed refrigerants are a mixture of two or more components. The light components lower the evaporation temperature and the heavier components allow condensation at ambient temperatures. The evaporation process takes place over a temperature range rather than at a constant temperature as with pure component refrigerants. This is illustrated in the P-H diagram (Figure 16.9) for a mixed refrigerant. The composition of the mixed refrigerant is sometimes adjusted so that its evaporation curve matches the cooling curve for the process fluid. Heat transfer occurs in a countercurrent exchanger, of- ten an aluminum plate-fin, rather than a kettle-type chiller. 230 VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION Mixed refrigerants have a somewhat better thermal efficiency than a cascade sys- tem because refrigeration is always being provided at the warmest possible tempera- ture, The amount of equipment is also re- duced compared to a cascade system. Dis- advantages include a more complex design and the tendency for the heavier compo- nents to concentrate in the chiller unless the refrigerant is totally vaporized. The mixed refrigerant process is used in some low- temperature NGL extraotion gas processing plants and is the most common process in large LNG plants. LNG processes will be Figure 16.9 Example P-H Diagram for a Mixed discussed in more detail later in this chapter. Refrigerant EXPANSION REFRIGERATION ‘Two types of expansion refrigeration processes are used in gas processing. 1. Valve expansion — often referred to as LTS, LTX or JT process _ 2. Turbine expansion — turboexpanders A third process type involves the expansion of gas through a tube where the gas attains sonic or supersonic velocity and liquids are removed from the tube wall. These processes will be briefly dis- cussed in a subsequent section, but are of limited commercial importance at present. Valve expansion processes have been used in the industry for over 50 years. They are simple, robust and relatively inexpensive, but have a low thermodynamic efficiency and generally require high pressure feed gas (10 000 kPa [1500 psia] or higher) to be economically viable. These processes are most commonly used in processing of gas to meet a hydrocarbon dewpoint. The turboexpander was first used in gas processing in the late 1960s. It is now arguably the most popular gas processing method. Turboexpanders can be used for very low temperature NGL ex- traction as well as for hydrocarbon dewpoint control. Turboexpanders (expanders) are very efficient, reliable and relatively inexpensive. VALVE EXPANSION Figure 16.10 shows an example LTS process. Feed gas enters a gas-gas exchanger where cool- ing is recovered from the cold sales gas stream, The gas is often not dehydrated, so hydrate formation in the exchanger and piping must be prevented. This is typically done by injecting an inhibitor (usu- ally MEG) into the gas upstream of the exchanger. Liquid hydrocarbons and the water-inhibitor mixture are removed from the gas in the Cold Separator. The cold, dry sales gas flows back through the gas-gas exchanger and on to the sales gas system. The temperature at the cold separator is usually controlled by bypassing a portion of cold sales gas around the gas-gas exchanger. CHAPTER 16 231REFRIGERATION SYSTEMS Dry Lean Gas Gas / Gas Ethylene Glycol i Scirwe | ASRS, JT Expansion Valve Lean Glycol D Injection Pump Low Temperature Separator NGL Hygroearbons wo Stubilzaon nee (Rich Glycol) Figure 16.10 Process Flow Diagram for a Typical LTS Process Vv An LTX process (as opposed to LTS) generally refers to a process where no hydrate inhibitor is injected into the gas. The process is designed so that any hydrate formation occurs downstream of the E-T valve and the cold separator liquid is heated by heat exchange with a warm fluid to melt any hy- drates which formed as a result of the valve expansion. The warm fluid is often the inlet gas to the process. LIX processes will not achieve as low a temperature as LTS, but they are simpler and less expensive. They have proven useful for field processing of gas to meet transportation or downstream processing specifications. Figure 16.11 shows an example LTX process. Pressure drop across a valve is an isenthalpic process, as noted previously. If no liquid forms, the following equation applies: (16.8) The symbol “1” is known as the Joule-Thomson coefficient. It is positive or negative, depend- ing on the relative size of the two terms in the numerator. VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION Separator ee Valve: re <- Steam i | | (Atternate Line) High Pr v igh Pressure Wellstream Waler io Disposal Figure 16.11 Example LTX Process a Curve A below shows a case where the instantaneous slope is greater than the average slope. Therefore, the gas will cool on expansion. The curve C gas is just the opposite and will heat on ex- pansion, Curve B is for an ideal gas, which will not change temperature on expansion. { 24) : (2) (2) is negative oT TY HH A Volume Temperature GF) Temperature Pressure Many gases exhibit a characteristic wherein the slope of the V-T curve changes sign. The tem- perature at which the slope changes sign ( = 0) is known as the inversion temperature. The right-hand plot above shows inversion temperature versus pressure. The shape shown is general for all actual gases. Outside the curve, the gas represented would heat upon expansion. Inside, it cools on expansion. Because of the location of the curve, hydrogen heats on expansion at normal pressures, whereas most light hydrocarbons cool, At very high pressures, of the order of 60 MPa [8700 psia], many naturally occurring hydrocarbon gases heat on expansion. Curves which show the temperature drop expected for a given pressure drop across a valve or choke are only applicable if no liquid forms on such expansion. CHAPTER 16 233REFRIGERATION SYSTEMS Liquid Condensation Across a Valve When condensation occurs, the calculation of the valve outlet conditions is inherently trial-and-error. L veer (he - hy) Calculate the specific enthalpy of the feed stream at P; and T;. If it is a two-phase stream, the PTs enthalpy is found by the weighted average that of — the liquid and vapor phases. ‘Assume the unknown temperature T. Make a flash calculation at P2 and T; to find relative amount and analysis of each phase. Find the specific enthalpy at Point Two from the above flash and the assumed T>. If hy = hy, you asgumed the right temperature, If not, repeats Steps 2-5 until the h’s are equal, within the desired limits of accuracy. Since the above is quite tedious, the calculation is usu- ally done on a computer using an EOS to calculate thermody- namic properties. For manual calculations, a reasonably good answer can often be obtained by assuming two different tem- peratures and plotting them on the following type of figure (at left). A straight line between the hy ~ hj, found for twe as- Te sumed temperatures, are connected by a straight line. The in- Temperature tersection at hp — hy = 0 gives approximate true T. For the system shown in Figure 16.10, the inlet gas stream is cooled by the sales gas before going to the J-T valve. All one knows are the inlet conditions and composition and the sales gas specifications. ‘These fix the pressure drop across the heat exchanger and valve in series. The drop across the former should not exceed 100 kPa [14.5 psia]. The following type of procedure is needed. 1 Ps er araye Assume temperature of gas downstream of the gas-gas exchanger. Run a flash calculation at this temperature and inlet pressure, minus exchanger pressure drop, say 70 kPa [10 psia]. Determing enthalpy of the stream at this point from compositions in Step 2. Use the previous procedure for the valve to find the temperature in the cold separator. Run flash at cold separator conditions. Find the enthalpy of the vapor leaving the separator. Assume a 5-10°C [9-18°F] hot-end approach on the gas-gas exchanger. Perform an energy balance on the inlet and sales gas streams in the gas-gas exchanger. a. if Abinjer > Absates, insufficient cooling is available to meet the assumed cold sepa- rator temperature. Reduce the exchanger approach or increase the valve pressure drop. D. if Ahgaies > Abjntes, excess cooling is available. Bypass sales gas around the gas-gas exchanger or increase the cold separator pressure. ©. if Absstes = Abjner, @ Solution has been reached. Once Step 8 is satisfied, the heat exchanger may be sized using heat transfer principles dis- cussed in Chapter 13. VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION Note: A flash calculation is needed on the inlet gas to the gas-gas exchanger if the inlet is two-phase and composition and relative quantity of each phase is not known. For applications where only small amounts of liquid condense across the valve, the process may be calculated using a pure component, such as methane. This is useful as a first approximation or reality check. Note: the calculated temperature drop for pure methane will be greater than for a natu- ral gas mixture. Example 16.4: Using the P-H diagram for methane, Figure 16A.2(a and b), estimate the tempera- ture drop across a valve for a pressure drop to 2000 kPa [290 psia] from 7000 kPa [1015 psia]. Inlet gas temperature is 10°C [50°F]. SI Solution: hy @ 7000 kPa and 10°C = 518 kJ/kg hy = hy, so at P, = 2000 kPa and hy = 518 ki/kg, Th, =-1"C FPS Solution: hy @ 1015 psia and 50°F = -1573 Btu/lbm hy = hy, so at P2 = 290 psia and h; = ~1573 Btu/Ibm, Th, <+7F TURBOEXPANDERS ‘Turboexpanders are single-stage radial-inflow turbines. A cutaway of a turboexpander/com- pressor assembly is shown in Figure 16.12. Gas enters the expander impeller radially and leaves axi- ally. Work generated at the expander wheel is absorbed by a single stage centrifugal compressor in most installations. This is because compression is almost always required downstream of the expander to restore the gas to sales gas pressure. In those cases where compression is not required, the expander can drive a generator to produce electricity. Expander efficiencies are quite high, approaching 85% (isentropic) for new installations operat ing at design conditions. Since their introduction in the gas processing industry in the late 1960s sev- eral thousand expanders have been installed and have generally proven reliable and robust. Bearings (2-radial and 1-thrust) have historically been oil lubricated. The first magnetic bear- ings were installed in turboexpanders in 1991. Several expanders currently use magnetic bearings”) and this technology is likely to become standard practice in the future. Seal gas is required to provide a barrier between the wet, cold process gas and the lube oil or magnetic bearings inside the rotor case. Adequate seal gas flow is necessary to prevent process gas from mixing with the lubricating oil and lubricating oil leaking into the process stream. Early seals were labyrinth as shown in Figure 16,12. Dry gas seals were first used in turboexpanders in 1989 and are commonly used today. They differ somewhat from the design used on centrifugal compressors in that a single seal (rather than tandem seals) is used due to space limitations. Regardless of the seal type, the seal gas must be clean and dry. Several turboexpander failures have been attributed to poor seal gas quality. The flowrate through the expander is controlled by inlet guide vanes. These are arranged radi- ally around the expander impeller and their movement is synchronized by rotation of a guide vane ring CHAPTER 16REFRIGERATION SYSTEMS a Seal Gas / Expander Discharge Variable Flow Control Lube Oil t ‘Compressor inlet lated by a flow or pressure controller. Figure 16.1 r Figure 16.13 Expander Inlet Guide Vanes Figure 16.12 Cross-Section of an Expander-Compressor (Courtesy Atlas Copco) which is attached to an actuator. The guide vane position, hence expander flowrate, is usually manipu- 3 shows expander inlet guide vanes. Turboexpanders are applied both in deep NGL Extraction (Cy+, Cy+) plants and hydrocarbon dewpoint control. The primary difference in these applications is the expansion ratio. For deep NGL extraction, expansion, rati are typically 3.0-3.5 and for dewpoint control ex- pansion ratios are usually about 1.3-1.5. An exam- ple hydrocarbon dewpoint plant using a turboexpander is shown in Figure 16.14. A simple deep recovery NGL process is shown in Figure 16.15. Modern deep recovery processes are more complex than Figure 16.15, employing significantly more heat integration as well as cooling technique: at the top of the demethanizer/deethanizer to in- cr se NGL recovery. Some of these schemes will be discussed in more detail section. process a later 236 VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION Dehydrated Feed Gas ‘Compressor Exchanger Cold Separator —— Expander Inlet Separator To Stabilizer Figure 16.14 Expander Process for Hydrocarbon Dewpoint Control Rasidue Gas Out Recompressor Dry Gas Feed from Mol Sieve Demethanizer or Dehydration Deethanizer Oe Heat Medium Supply/Return jottom Reboiler , NGL Product Product of Figure 16.15 Simple Expander Process for Deep NGL Extraction CHAPTER 16 237REFRIGERATION SYSTEMS Expander Calculations The theoretical thermodynamic path through an expander is isentropic. The theoretical work is calculated from an isentropic path and corrected to the actual work by use of an expander efficiency. For multicomponent streams and condensation across the expander, the calculation is iterative and te- dious, and is almost always done on a computer. The calculations proceed as follows: Th The known variables will typically be feed composition, w Tj, Py, and P>. ae 1, From P; and Ty, calculate hy and s}. Ps Assume a value of T,..,. 3. Runa flash calculation at the assumed T and known P) to establish the phase amounts and compositions. 4, Calculate hy and sp. (If the outlet is two-phase, these will be total stream values as dis- cussed in Chapter 8.) 5. If sp from (4) equals s}, you have assumed the right temperature. If not, repeat Steps 2-4 until s2 = s). 6. When sp = 5), calculated Ahigen = ajgeq — hi 7. Calculate expander power, W = MARigen Eisen: iu 8. Calculate hy,...4) = hy + (h2jseq — hi )Eisen- 9. Assume a value of T2,,, (this will be warmer than Tp,,.,)- 10. Runa flash calculation at the assumed T},., and P>. 11, Calculate h (hy,,,) from the results of the flash. 12. If hy from (11) equals hy,.,., you have assumed the correct temperature. If not, repeat steps 9-11 until ha... = Roy ca Example 16.5: Using the information in Example 16.4, calculate the outlet temperature and work if the gas was expanded across a turboexpander rather than a valve. Ejgen ~ 0.82 and the gas flowrate is 1 x 10° std m*/d [35.4 MMscfd] and MW = 16. SI Solution: 1, at Py and T;, hy = 518 kJ/kg and s, = 9.08 kI/kgK 2. s) =51, 80 at P; = 2000 kPa and s) = 9.08, Teen = 70°C agey = 380 KI/kg Abjsen = 380 — 518 = -138 ki/kg m = (1.0)(1762)(16) = 28 190 kg/h = 7.83 kg/s -W = mAhjsenEisen = (7.83)(380 — 518)(0.82) = -890 kW W = 890 kW (work is positive) 5. Abtgetuat = 518 + (380 — 518)(0.82) = 405 ki/kg 6. at Pp = 2000 kPa and hy = 405 kJ/kg Tagg = 60°C 238 VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION Example 16.5 (Cont’d.): FPS Solution: 1. at Py and T,, hy = -1573 Btw/lbm and s; = 2.17 Btw/lbm-°R 2. 8. = 81, $0 at Py = 290 psia and s) = 2.17, Tojen = 94°F a ygq = 1632 Btu/lbm Abigen = [-1632 — (-1573)] = -59 Btu/lbm 4, m= (35.4)(110)(16) = 62 300 Ibm/hr W = 1180 hp (work is positive) 5." Abgetuat = 1573 + [-1632 - (-1573)](0.82) 6. at P, = 290 psia and hy = 1621 Tayiy) = -76°F Estimation of Expander Efficiency The best way to estimate efficiency is from actual performance data, It is difficult to correlate efficiency data because many factors affect actual performance. As with compressors, normal manufacturing tolerances can affect performance measurably. Erosion can alter the shape of a wheel and thus efficiency. The presence of liquids likewise may have a dramatic effect. For all of these reasons, there may be a significant error in the estimated efficiency. In correlating efficiency data, one may use the basic similarity parameters governing all turbomachinery. Similar to centrifugal pumps and compressors, expander efficiency can be correlated against the specific speed of the impeller. Figure 16.16 shows such a correlation. Specific speed is one of the criteria for determining performance. It was discussed previously in the centrifugal pump section. The basic equation for expanders is: Na.) : N, _N@o) a (16.9) [AAD] aie SI FPS Where: N = shaft speed rpm 1pm Sh = isentropic Ah kiikg Btw/lb q = turbine exhaust volume m/s f/sec A. = conversion factor 31 718 For a radial inflow turbine, the optimum E is for a specific speed between 70 and 100. How- ever, this is not always possible because of the limitation in available standard expander frame and other parameters which affect performance. Figure 16.16 is for expanders where the amount of liquid formed is minimal. How much does liquid affect efficiency? This depends on where, and how, it was formed. Data from steam turbines has shown the predicted efficiency to be significantly less than the actual efficiency when too much liquid was formed in an axial turbine. CHAPTER 16REFRIGERATION SYSTEMS Isentropic Efficiency, & Specific Speed, Ns Figure 16.16 Efficiency of Expansion Turbines as a Function of Specific Speed The question of liquid formation is an important one. There is good reason to believe that liq- uid droplets do not actually form in the expander due to limited residence time. An equilibrium gas-liquid mixture only occurs at some point and downstream of the impeller. So ... it is likely that most of the liquid forms just downstream of the wheel and does not interact with the wheel proper. Initially, any liquid formed will be of submicron size. If there was significant liquid within the wheel proper, the denser liquid would be centrifuged outward, which would lead to lower efficiencies than we note because of flow distortions. Also, wheel erosion could be significant. How much liquid can be handled without significant loss of efficiency? Some units contain above 20 mol% liquid in the exit gas. Certainly 10% is being accommodated and 20% is not unrea- sonable. Expander efficiency will deteriorate over time due to normal wear and tear, This deterioration is accelerated by any erosion caused by droplets and particulates in the feed gas. Efficiency can also decline due to variation in flow from the design rates. Figure 16.17 shows how expander efficiency can vary with inlet flowrate.“6 Flow Parameter = Where: = mass flow inlet temperature = gas MW inlet pressure 224s i 240 VOLUME 2: THE EQUIPMENT MODULESEXPANSION REFRIGERATION 8 8 T 3 Expander Peak Efficiency, % fs el | 30 40 50 60 70 80 90 100 110 120 130 Percent of Design Flow Parameter Figure 16.17 Example Peak Efficiency vs. Percent Desgn Inlet Flow Parameter In the early planning stages, flow rates and gas compositions are rather inexact. Some use a design efficiency about 5-10% lower than that predicted by correlation. Obviously, E affects outlet temperature, liquid condensation and the composition of both phases which, in turn, affects all down- stteam equipment. Power transmitted to the booster compressor is also affected. In addition to specific speed there are other parameters affecting expander performance. Spe- cific diameter, presented earlier in Chapters 14 and 15, is one of these, defined by the equation: (@A)(any)}? ae 16.10) (@2)°* Bee aii __ SI isi FPS sisi Where: d = turbine diameter m ft Ah = isentropic Ah kJ/kg Btwlb q = turbine outlet volume mis f/sec ‘A = conversion factor 31 718 The specific diameter should be approximately 1.2 to 1.4 to achieve maximum efficiency. This is a useful number for estimating impeller size and shaft speed. Expander efficiency can be calculated using measured process data. The most convenient (and accurate) method is to use the expander compressor as a dynamometer. The following data is required “Expander Inlet: gas composition, T and P, and flowrate + Compressor Suction: gas composition, T, P, and flowrate + Compressor Discharge: T and P The calculation procedure proceeds as follows: 1. Accurately measure the pressures and temperature at the suction and discharge of the com- pressor. 2. Use a computer program to calculate hxc. and hgisch for the compressor Compressor Work = m (hgiseh — Bguct) 4, Add 2-3% to the work in step 3 to account for bearing and seal losses. ‘This represents the actual work generated by the expander 5. Use a computer program to calculate the isentropic work from the expander 6. Eisen = (Wactuat)/(Wisentropic) CHAPTER 16 241REFRIGERATION SYSTEMS APPLICATIONS OF REFRIGERATION IN GAS PROCESSING By far and away, the most common application of refrigeration in gas processing is the cooling of natural gas to meet a sales gas specification and/or the extraction of NGL components for recovery and sale. Selection of a suitable process scheme is based on several parameters including: + Process objective (hydrocarbon dewpoint control or NGL extraction) + NGL products to be recovered, sales revenue + Inlet pressure + Sales gas pressure + Capital and operating costs + Plant location, site conditions Hydrocarbon Dewpoint Applications Processing to meet a hydrocarbon dewpoint specification does not typically require low levels of refrigeration. Most specifications can be met by removal of the heavier C;+ components from the gas. There is little incentive to extract C2, C3, and Cy hydrocarbons from the gas, indeed the extraction of these components may reduce the gas value. If the process inlet pressure is less than about 8000 kPa [1160 psia] and the sales gas pressure ‘on the order of 6900 kPa [1000 psia] the most common process choice is mechanical refrigeration. There are exceptions to this, particularly for very lean gases where an adsorption system may be a better choice. If the sales gas pressure is low, e.g. gas turbine fuel (3500-4100 kPa [500-600 psia]) an expan- sion process is often more economic due to the availability of “free” pressure drop. Expansion processes are the preferred alternative when the inlet gas pressure is at least 20-30% greater than the sales gas pressure. This is particularly true at pressure above about 8000 kPa [1160 psia] because mechanical refrigeration systems are generally not effective at high pressure due to the retrograde phase behavior of the gas. If the inlet pressure is provided by the reservoir, i.e. no inlet compression is requifed, the valve expansion process might be more economical when inlet pressure to sales gas pressure ratio exceeds about 1.5. If the inlet pressure is provided by compression or when the ratio of inlet to sales gas pressure falls between about 1.1 to 1.5 an expander process is often preferred. Figure 16.18-16.20 shows some process temperatures and pressures for the three refrigeration schemes discussed above. The process path is also plotted on a phase envelope to better visualize the process differences. In each case the sales gas pressure is 6900 kPa [1000 psia], the minimum ap- proach on the gas-gas exchanger is 5°C [9°F], the dewpoint specification is -3°C [26.6°F], and the cold separator carryover is 2%. 242 VOLUME 2: THE EQUIPMENT MODULESAPPLICATIONS OF REFRIGERATION IN GAS PROCESSING 14000 12000|-— : - 10000) | oN i &F Ses Gos Chiller & i [pe lat Feod Ges ¢ Conditions: 3 27-0. 7050 KPa 3 s000}-— - a 4000}-— ~-4-3-1- 2000 —-- ° ‘90-80-70 -60 50 #0 20 20 10 0 10 20 30 40 50 60 Temperature, °C Figure 16.18 P-T Path for a Mechanical Refrigeration Process. 80 -80 -70 -60 -50 -40 -30 -20 -10 Temperature, °C Figure 16.19 P-T Path for a Valve Expansion Process CHAPTER 16 243REFRIGERATION SYSTEMS. 14000 T | 12000 |~ | arrears | : Expander | 0000 |~ int Separator Food Ga ‘ Pate Ee Gonaons: © ccoo |_| arena $ gains Gas a | Conctons: 3 6000 ~ Cold Separator aaa, Gan ea: g Corton, APC eo 4000 2000 | ° a sei 90 -80 -70 -60 -50 -40 -30 -20 -10 oO 10 20 30 40 50 60 Temperature, °C Figure 16.20 P-T Path for a Expander Process NGL Extraction The preferred processes for deep NGL extraction from the gas have historically been mechani- cal refrigeration, refrigerated lean oil absorption and turboexpander. Mechanical refrigeration is gener- ally not effective in achieving high extraction levels unless the gas is relatively rich in Cyt hydrocarbons and unless cold separator temperatures are below ~70 to -80°C [-94 to —112°F]. This requires a cascade or mixed refrigerant type refrigeration system. Refrigerated lean oil absorption can be very effective at removing NGLs. Ethane recoveries as high as 50% and propane recoveries of 98-99% have been achieved, These processes have high capital costs and are considerably more complex than mechanical refrigeration and expansion type processes. Few plants of this type have been built since the early 1970s, although new process developments have made this process attractive in some applications."* The lean oil absorption process will be dis- cussed in more detail in Chapter 17. The Petroflux® !69 process combines mechanical refrigeration and an absorption effect by using a refluxing heat exchanger. A simple flow diagram for the Petroflux process is shown in Figure 16.21. Feed gas is cooled by cross-exchange with cold sales gas and enters the cold separator. Vapor from the cold separator flows to the refluxing heat exchanger sometimes called a dephlegmator. Cooling is provided by external refrigeration. In the refluxing heat exchanger, liquids which condense from the gas flow downwards, countercurrent to the gas flow. The exchanger not only cools the gas but acts like a packed tower. The countercurrent flow through the exchanger also results in a sharper separation between the condensed NGL components (Cy+ or C3+) and the methane. This mass transfer effect significantly reduces the size and utility demand of the stabilizer. This process can also be com- bined with a turboexpander. Expander processes are the preferred process for deep NGL extraction. These processes are relatively simple, compact, can achieve ethane recoveries of 90% (propane 99%+), and have the lowest capital cost. If the sales gas must be returned to the pipeline, at high pressure, the largest plant 244 VOLUME 2: THE EQUIPMENT MODULES‘APPLICATIONS OF REFRIGERATION IN GAS PROCESSING Se Gas — | | { Refrigeration (/ systom Heat Exchanger \Retiuxing Heat Exchanger Cold Separator = Crude NGL ee Reethanize 7 iO) Feed Gas Recovered Hydrocarbon Liquids Cot Figure 16.21 Example Petroflux® Process operating cost is compression. Dehydration upstream of deep NGL recovery plants is almost always adsorption — using molecular sieves, although methanol injection has been used in some designs. A simple expander plant is shown in Figure 16.22. 7 Early processes used this scheme. Expansion ratios across the expander were 3 to 3.5 and ex- pander outlet temperatures were typically -90 to -100°C [-130 to -148°F]. While very simple, this Process scheme had limited thermodynamic efficiency. ‘Temperature approaches in heat exchangers were large and the potential refrigeration effect of methane revaporization in the demethanizer was not exploited. In addition, because there was effectively no temperature control at the top of the demethanizer, ethane recovery was limited and it was difficult to reject ethane without incurring signif- icant propane losses. Te Cooler and Sales_ = Hot Gas Dehydration exchanger Demethanizer Figure 16.22 Flow Sheet for Simple Expander Plant CHAPTER 16 245REFRIGERATION SYSTEMS The use of side reboilers on the demethanizer improved heat recovery. A portion of the feed gas was used as the heating medium, resulting in lower expander outlet temperatures (hence higher NGL recovery levels) for the same expansion ratio. In addition, the use of BAHXs decreased tempera- ture approaches in heat exchangers to 2-3°C [4-6°F]. Figure 16.22 shows an optional chiller. External refrigeration is sometimes used to assist in feed gas cooling. This is often the case when the feed gas is quite rich in Cy+ components. The rea- son for this is that less sales gas is available for back exchange in the gas-gas exchanger. Basic turboexpander plants, like the one shown in Figure 16.22 are limited to about 70% eth- ane recovery and 95% propane recovery. In addition, the basic plant is subject to CO; freezing and cannot maintain high levels of propane recovery during those periods of operation when ethane is re- jected. The design of turboexpander plants has evolved further for four reasons. 1. increase ethane recovery 2. increase CO, tolerance of the process 3. increase propane recovery during periods of ethane rejection 4. improved energy efficiency Several process designs have emerged, but they all share one common theme — provide a “re- flux” stream at the top of the demethanizer. One process which is now widely used in the industry is Ortloff’s Gas Subcooled Process (GSP).67) A simple process flow diagram is shown in Figure 16.23. In this process, a portion of the expander feed, say 20-30%, is bypassed around the expander and subcooled by heat exchange with the Rretrigerant Make-up > el as Reinjection Propane Cycle 1690 [-261°F] 43.9 barg [635 psig} 16.41 barg [93.0 psig] Copper 2.83 barg {41 psig] Z ies exchanger. (0.12 barg (1.8 psig] MR Cycle [le Fags ‘elo Q Figure 16.26 Basic LNG Process (Courtesy Air Products and Chemicals, Inc) 250 VOLUME 2: THE EQUIPMENT MODULESLIQUEFIED NATURAL GAS (LNG) Mixed Refrigerants Mixed refrigerants were discussed briefly in a previous section. An example mixed refrigerant might be a 50:50 mixture of ethane and propane. Such a mixed refrigerant will vaporize (i.e, boil) over a temperature range and not at a single temperature as shown in the P-H diagram, Figure 16.27. The temperature range is set by both pressure and composition. Ta- ble 16.2 summarizes the atmospheric boiling temperatures of the common components in LNG mixed reftigerants. Figure 16.28 shows the P-H diagram for the following mixture: 5% Np, 50% Cy, 30% Cp, 15% C3. This means the reftigeration effect will be distributed over a range of temperatures as opposed to a constant temperature, e.g. kettle-type chillers. This produces a smaller overall temperature differ- ence between the natural gas (being cooled) and the mixed refrigerant (being warmed up). These smaller temperature differences lead to a more efficient liquefaction process. Contrast such close tem- perature approaches with the “stair-step” temperature differences in the cascade refrigeration (Figure 16.8). As can be seen by comparing Figure 16.27 and 16.28 adding methane and nitrogen to the sim- ple 50:50 C)-C3 mixture (mixed refrigerant or MR) extends the boiling temperature range to lower temperatures. Most gas industry mixed refrigerants (MR) are a mixture of C3, C2, and Cj, often with an additional small amount of N2. Sa 10.000, Pressure, kPa Enthalpy, ki/kg Figure 16.27 _P-H Diagram for Simple Mixed Refrigerant, 50% C2, 50% Cy CHAPTER 16 251REFRIGERATION SYSTEMS TABLE 16.2 Refrigerants — Boiling Points (Single Component) _ Component Boiling Point | _‘Temperature gq | @1atm | 42°C [-44°F] Q : @1 am -89°C [-127°F] cG @1am _=162°C [-259°F] Ny @1.atm 196°C [-320°F] Gq @ 13.1 bara [190 psia}_|_ = ~120°C [-180°F 10.000 Pressure, kPa 8 Enthalpy, ki/kg Figure 16.28 P-H Diagram for Example Mixed Refrigerant, No, Cy, C2, Cy Cryogenic Heat Exchanger The main heat exchanger (MHE) is the cryogenic exchanger that efficiently condenses the pre-cooled natural gas to LNG, Figure 16.29. In its simplest form it contains two heat exchanger bun- dles with extremely large surface area. Standing vertically it is fed two mixed refrigerant (MR) streams: “Light MR” to the cold bundle on top and “Heavy MR” to the warmer bottom bundle. Some MHEs contain three bundles. The high-pressure mixed refrigerant supply (T) is already partially condensed in discharge coolers downstream of the MR compressor (see Figure 16.26). The saturated vapor (U) is the light MR containing predominantly the lighter components C; and C2 plus any nitrogen, if present. The heavy MR is saturated liquid (V), predominantly C, and C3, which is subcooled in the warm or bottom bundle before being depressurized at point X and distributed across the warm bundle cross-section. The heavy MR then progressively vaporizes in downflow to the bottom outlet (DD) which in tum leads to the MR compressor suction (see Figure 16.26). The refrigerating effect of the heavy MR is used on three streams: 1. subcooling itself, points V to W 2. first stage of cryogenic natural gas cooling and condensing, M to N 3. condensing the light MR stream, U to Y 252 VOLUME 2: THE EQUIPMENT MODULESLIQUEFIED NATURAL GAS (LNG) Light Refrigerant 4 Cold Cryogenic Bundle Heat } : Exchanger Heavy Refrigerant Warm Bundle Mixed . Refrigerant Supply LNG M Feed Mixed Refrigerant Gas Back to Compressor DD Figure 16.29 Main Heat Exchanger The light MR, predominantly C,, having been condensed at point Y is further subcooled in the top bundle to point Z. It is then depressured and distributed across the cold top bundle at point AA. It then downflows through the top bundle and progressively vaporizes to point BB. The refrigerating ef- fect of the light MR is used on two streams: 4, subcooling itself, Y to Z 5. final stage of cryogenic natural gas cooling and condensing, N to Note that the refrigeration effect generated in the warm (bottom) bundle is actually due to the combined cooling effect of the heavy refrigerant (X) and the vaporized light refrigerant leaving the cold bundle (BB). This combined refrigeration effect in the warm bundle broadly comes from: a. boiling (vaporizing) of the heavy MR b. plus a superheating of the already vaporized light MR The vaporized light MR and the liquid heavy MR that combined at CC eventually exit the bot- tom as vapor stream, DD. In base-load LNG plants this vapor line leading to the MR compressor suc- tion can be more than | m [3.3 ft] diameter. CHAPTER 16 253REFRIGERATION SYSTEMS: Cascade Refrigeration for LNG ‘A second type of LNG refrigeration system is the Phillips Optimized Cascade LNG Process. As the name suggests, it is a cascade system and employs three refrigerants: propane, ethylene, and methane. ‘A simplified process flow diagram for the Phillips Process is shown in Figure 16.30. Follow- ing the typical front-end treatment (acid gas removal, propane chilling/liquids removed, dehydration and mercury removal) the gas enters the propane refrigeration cycle. Here the gas is cooled to about 40°C [-40°F]. The propane refrigeration system also condenses the ethylene refrigerant and cools the methane refrigerant. The gas then enters the ethylene refrigeration system where it is further cooled to about -90°C [-130°F]. The ethylene refrigerant also condenses the methane refrigerant. The feed gas is finally cooled by the methane refrigerant to produce LNG. Inlet ‘Scrubber ae Chiller Treater Mercury Dehydrator Removal Ain Hea] Exchanger Plant Fuel Methane Compressor Heat Exchangers, Heat. Flash Tanks, and Exchanger Pressure Letdown Valves - Vapors from Ship — 2 nen Loading Tank Vapor Blower ‘Storage Ship Vapor ae _,| Tanks beacnel and Loading Facities | Pumps. To Ship Loading Facies 254 VOLUME 2: THE EQUIPMENT MODULESLIQUEFIED NATURAL GAS (LNG) The methane refrigeration system is open-cycle, i.e. the methane refrigerant stream is taken from the gas being liquefied. This allows the rejection of nitrogen (if required) and allows boil-off gases to be reintroduced back into the liquefaction process without the necessity for a large boil-off gas compressor. If necessary, NGLs can be removed from the gas downstream of the propane refrigeration sys- tem, Fuel gas is taken off the methane refrigeration compressor. Rather than one large proprietary spiral wound exchanger, the Phillip’s process uses conven- tional BAHXs. This allows more flexibility in design capacity and may reduce major equipment deliv- ery time. As in the case of the mixed refrigerant process the primary energy input is refrigeration com- pression. Modern LNG plants typically have an overall efficiency (LNG out vs. feed gas in) of about 90-92% depending on feéd composition and pretreatment requirements. REFERENCES 16.1 Briley, G. C., Hydr. Proc. (May 1976), p. 173. 16.2 Agahi, R. A., “Turboexpanders with Dry Gas Seals and Active Magnetic Bearings in Hydrocarbon Processing,” 78th Annual GPA Convention, Nashville, Tennessee (Mar. 1999). 16.3 Mclntire, R., “Increased Productivity of Turboexpanders,” 61st Annual GPA Convention, Dallas, Texas (Mar, 1982). 164 Mehra, Y. R., “Mehra Process Flexibility Improves Gas Processing Margins,” 66th Annual GPA Convention, Denver, Colorado (Mar. 16-18, 1987). 165 Limb, D. I. and B. A. Czamecki, “The Petroflux Process in NGL Recovery - Experience to Date and New Develop- ‘ments,” 66th Annual GPA Convention, Denver, Colorado (Mar. 16-18, 1987). 16.6 Finn, A.J. etal, “Modem Process Designs for Very High NGL Recovery,” 78th Annual GPA Convention, Nashville, Tennessee (Mar. 1-3, 1999). 16.7 Wilkinson, J. D. and Hank M. Hudson, “Improved NGL Recovery Designs Maximized Operating Flexibility and Product Recoveries,” 71st Annual GPA Convention, Anaheim, California (Mar. 16-18, 1992). 168 Pitman, RIN. etal, “Next Generation Processes for NGLILPG Recovery.” 77th Annual GPA Convention, Dallas, ‘Texas (Mar. 16-18, 1998). CHAPTER 16 255REFRIGERATION SYSTEMS NOTES: 256 VOLUME 2: THE EQUIPMENT MODULES