FLUIDS/SOLIDS HANDLING Design Two-Phase Separators Within the Right Limits Ss Here is a proven, step-by-step method. WY. Svrcek, W.D. Monnery University of Calgary iquid/vapor separators are one of the most common types of process equipment. Many technical papers Jhave been written on separator design and vast amounts of information are also available in corporate process engineering design guidelines. The basic equations used for sizing are widely known; however, subjectivity exists dur- ing the selection of the parameters used in these equations. This article attempts to address the basics of two-phase separator design and provide step-by-step proce- dures and examples for two-phase vapor/liquid separator design. ‘Two-phase separator types and selection ‘Two-phase separators may be oriented either vertically or horizontally. In some cases, it may be necessary to compare both designs to determine which is more economic. Separators may be designed with or without mist eliminator pads and may also have inlet diverters. Some sepa- rators may have proprietary impingement or settling internals. The vendor should be contacted to design these types of v Vertical vapor/liquid separators are pre ferred for separating liquid from mixtures with a high vapor/liquid ratio while hori zontal separators are preferred for sep: rating vapor from mixtures with a low vaporfliquid ratio. Background Vapor/liquid separation is accompli ly ed in three stages. The first stage, primary separation, uses an inlet diverter so that the momentum of the liq- uid entrained in the vapor causes the largest droplets to impinge on the diverter and then drop by gravity. The next stage, secondary separation, is gravity separa- tion of smaller droplets as the vapor flows through the disengagement area. The final stage is mist elimination where the small- est droplets are coalesced so that larger droplets are formed which will separate by gra For secondary separation, the allow- able velocity must be calculated so that disengagement area can be subsequently determined. Performing a force balance on the liquid droplet settling out provides the necessary relationship, When the net gravity force, given by Eq. 1, a balances the drag force, given by Eq. 2, (2/8) RU py the heavier liquid droplets will settle at a constant terminal velocity, U,. Equating Eqs. | and 2 results in, / 48 D.(e.-Pv) 3CpPv Hence, as long as U, < U,, the liquid droplets will settle out. Typically, the allowable vertical velocity, Uy, is set U,= CHEMICAL ENGINEERING PROGRESS © OCTOBER 1993 * 53FLUIDS/SOLIDS HANDLING between 0.75U, and Ui, Bg, 3 can be rearranged as Eq. 4. a Sauders-Brown ype equation (1) P. - Ps) «a 1 (e 5 Ps) 4 where ‘4eD. 5 where K= “ab. 6 Practically, very sn II droplets ea not be separated by These droplets are coalesced to form er droplets which will settle by avity. Coalescing devices in separa tors force the gas to follow a tortuous path and the moment droplets causes them to collide with other droplets or the coalescin: device, formi ravity alone. m of the droplets. The coalesced droplet diameter is not ade: ——— When calculating U, for a horizontal separator, a “no mist eliminator K value” should be used. quately predictable so the K values Tor mist climinators are typically empirical. Thi Tirst enters separator design. There I Jterature sources of Kv isk where subjectivity ues such as the Gas Processor’s Supplier Association (GPSA) ME Data Book” /2), numer fous technical publications and ven- dors recommendations, The GPSA 2) and York Mist Eliminator (3) val- SECA Te Cl OCU Mist Eliminator 1s PENS ‘K=0.1821 + 0.00297 + 0.0480Ln (A) 15s Ps K=035 Posie 40s P5500 K=0430 0023001 ‘For compressor suction scrubbers, mole sieve scrubbers and expander inet ‘Separators multiply Kby 07-08. ‘Theoretical (no mist eliminator) Ye R411 ~2248X + 0.273X'~ 1.8656 -2%+ 5201 ~ xt ed fossa at He Notes: o,f itn wer 1 cron = 328084 x 108 54 © OCTOBER 1993 © CHEMICAL ENGINEERING PROGRESS ues have been curve given in Table 1 If there is: no mist eliminator, it is, recommended to use ene half of the above values (2) or the be ealeulated from Eg. 5 if Ue liquid droplet size is known. The drag coetficient. € ted and is given in Table I or can be obtained from Figure 7-3 in the GPSA Engineering Data Book” (2) it is worthwhile ted and are theoretical has been curve fit Before proceedin; to clarify some definitions and criteria. Holdup is defined as the time it takes to reduce the liquid level from normal (NLL) to empty (LLL) while maintain ng 3 normal outlet flow without feed makeup. Surg time it tikes for the Tiquid level to rise from normal (NLL) to maximum (HLL) while maintainin feed without any outlet flow jdelines buse “surge” on the volume between low (LLL) and high (HLL) liquid levels, Holdup time is based on ime is defined as the a normal Some the reserve required to maintain good control and Stream Ewilities. Su mased on requirements to accumulate safe operation of down ye lime is usually liquid as a result of upsizeam or down Stream variations or upsets, for exam= ple. slugs. In the absence of specific requirements, surge time may be vaken as one half of holdup time For vertical separators, the vapor disengagement area is the entire cross-sectional area of the vessel so thatt vapor disengagement diameter ean be calculated trom Ey, 6: Vertical separators ) Technically, this is the mist etiminator Giameter and the inside diameter of the vessel must be slightly larger 90 that the mist eliminator can be installed inside the vessel, Typically, the calcu lated value is taken up to the next six in, This value is tken as the required diameter of the vessel, P. and the cor responding cross-sectional area, A, is calculated using this diameter The next step in sizin; separator is to determine the hei a vertieFor a two-phase vertical sep. total height can he broken tions, ay shown in Fi rator he adding the hy as per Eq. 7. ated by ts of these sections, MeHg +H, tH eH, +H, Ia mist eliminator pad is used. additional height is added, as shown in Figure 1, The calculations of diameter and height are detailed in the “Design Procedures” section of this article Horizontal separators, For hoti- zontal two-phase separators, the cross-section is oceupied by both vapor and liquid, as shown in Figure 2. When sizing horizontal two-phase separators, usually a diameter is assumed, LLL is selected or caleulat= ed. NLL is set by liquid holdup, and HLL. is set by liquid surge. The eross- sectional area between HLL and the Lop of the vessel is used for vapor disengagement. The length of the vessel is then calculated lo accom- modate holdup and surge or to facili- tate vapor liquid separation. Hence, this approach to sizing horizontal separators. oF variations of it, are iterative calculations, The following will develop the asic equation used for calculating the size of a horizontal separator. For a horizontal separator cross section, a ‘volume balance” iy written. Vy tMELA, AA) (8) Holdup and surge volumes are caleu- lated from holdup and surge times which are selected according to guide- lines. The low liquid level area is a Function of the lov tiquid level height, ‘obtained from guidelines, and the ve sel inside diameter. The minimum vapor disengagement area , Ay, is lypically specified as or wo two fh or 20% of the vessel inside diameter, whichever is greater, The sizing method in the “Design Procedures” section of this article assumes this for A,,,and only inereases it if the length Wahaxt Mist liminator wat Vapor Mist Eliminator Outlet igus urlet Nozzle Figure T. Vertical two-phase separator. required for vapor-liquid separation is much greater than the length requited for holdup and surge, for # given diameter, Equation 8 is then a fune- tion of the inside diameter. D, and length, l. For horizontal separators, the lig- uid droplet to be separated from the sas has @ horizontal drag force which is not directly opposite to gravity yin the vertical case, Without detailed treatment of two-dimensional particle motion, most literature sources reco- nize that the allowable horizontal velocity can be higher than the termi- nal velocity (2, 5, 7). This ean be ioereececeoenr reel If a mist eliminator pad is used, additional height is added. shown simply by equating the “resi- ce" times of the Fiquid dropiet w be settled. That i, the time it takes ¥9 travel the horizontal Hength between inet and outlet must be greater than dei the time it tikes wo settle the vertical distance 1 the Figuid surface Lath Wu? U, ° This can be rearranged in terms of the allowable horizontal velocity i) ‘The length, L, divided by the heigit of the vapor disengagement area, Hy, would always be greater than unity The allowable horizontal velocity isa very subjective topic with several empirical approaches to modify the vertical "A" value available in the lit erature (2, 5. 7, 8). For horizontal sep- arator design, the subsequent design procedures use a “droplet settlin approach” similar to the API proce- dure (6) which does not require empirical modification of the "A val- ue for vertical settlers. It should be noted that when calculating U, for a horizontal separator. “no mist elini- nator K value” should be used. Design Procedures ‘The Following design procedures and heuristics are a result of a review of literature sources and accepted industrial design guidelines. The hori- vontal design procedure incorporates optimizing the diameter and length by minimizing the weight of the shell and heads. To add a degree of conser vatism to the design, the volume available in the heads iy ignored. Vertical design procedure (See Figure 1): 1, Calculate the vertical terminal vapor velocity x [Pu Pe k [Pa u Set U, = 0.75U, for a conservative design. Culeulate the K value from Table 1. CHEMICAL ENGINEERING PROGRESS © OCTOBER 1982 © 55FLUIDS/SOLIDS HANDLING ——— Nomenclature G D >, D, SS ES PETXFSSESE SREP OOS = vertical vessel eros sectional Area, = cross section For LLL (horizontal vessel), ff ‘= total oss sectional area (ovinootal vessel), ‘vapor disengagement area = droplet diameter, fi. = nozzle diameter, in Galt or out et vaporliquid as spesitca) = vapor disengageroent dete, ft = welded Joint efficiency rag force, ‘gravity force, 1b, «= gravitational constant 82.17 fs? £32.17 (bovis) «= disgengagement height, ft = holdup height 2 Hy, t0inket nouae centerline net high quid evel + Jow Liquid Level height, f ‘= mist imino top tnk ei, = surge ei ft ‘= total vertical separator height, ft ‘= vapor disengagement area height, & ‘terminal velocity constant, U3 = vesellongh, low liquid level ‘= Yapor/fiquid separation minimum ‘ength, ft = dropletimass lb, ‘= normal fiquid tevel = pressure, psig or pia ‘liquid volumetri¢ fow, Yimin ‘= minture volumetric flow, 178, fffoin Q, = vapor volumetric ow, is, [min S/—_#yessel material suess value, psi T, — =boldup time, min td mbeadthichnes, in 1 zshalthickness. ia Uy) allowable horizontal velocity, 105 LU) -=termina velocity. ts U,_ =misture velociy.1%5 Uy = vapor velosity, fs V,, holdup volume. Vig =LLL volume, Vi" surge volume. V,_ total volume thorzootal vessel We vessel weight, Greek Letters 2 = mixture liquid fraction By = Yapee viscosity, cP liquid density, vt Py = mixture density, Iv? Py = Napor density ht liquid dropout time, s 2. Calculate the vapor volumetric 3, Calculate the Vessel (inside) flow rate: diameter: rf aM yy (12 40v 13) Qs aap ts 1% Duy (SBE) no Eee CURT CRC) 56 © OCTOBER 1993 + CHEMICAL ENGINEERING PROGRESSthere is a mist eliminator, ad 3 to 6 in. to D,,, 10 accommodate & support ring and round up to the next 6 in increment to obtain D. If there is no mist eliminator D = Dy 4, Calculate the liquid volumetric flow rate: ft Imin (14) (1) 5, Select holdup time from Table 2 and cafculate the holdup volume: T,)(Q,) as) even Teer 6. If the surge volume is not specified, select a surge time from Table 2 and calculate the surge volume: Vs (7) (Q,) fP (6) 7. Obtain Jow liquid level height, H,,,. from Table 3. 8. Calculate the height trom low lig- uid level to normal liquid level: 1 ft minimum 9. Calculate the height from normal jwid level to high liquid level (or high level alarm): Vs (x/4)D 6 in minimum 10. Calculate the height from high liq- uid level (o the centerline of the inlet norzke: Fy = 12+ dy in. (ith inlet diverter) Hy = 12+ Yedy.in, (without inlet diverter) a9) Note: dy is calculated as per Table 4. 11. Calculate the disengagement height, from the centerline of the inlet nozzle to: if as Feed inlet suge 0 op 1 Figure 2. ue Horizontal two-phase uate separator. : a. the vessel op langent line if there is no mist efiminator or b. the bottom of the demister pad. H, = 0.5 D, ora minimum of H,,= 36+ Ysdy. in. (without mist eliminator) 20) H, = 24+ dy, inches (with mist eliminator) 12, If there is a mist eliminator, take 6 in. for the mist eliminator pad and take 1 f. from the top of the mist eliminator to the top tangent line of the vessel 13, Calculate the total height, #7, of the vessel Hy=H,, +H) tHe Hy Hy +H, ay where H,,, is the height from step 12; if there is no mist eliminator H,,, = 0. Horizontal design procedure (See Figure 2). 1, Calculate the vapor volumetric flow rate, Q, using Eg. 12. 2. Caleufate the liquid volumetric flow rate, Q,, using Eq. 14. 3. Calculate the vertical terminal vapor velocity, U,, using Eq. 13, (K value as per Table | for no mist eliminator). Set U, = 0.75 U, for a conservative design. Fx Lu Vapor Outlet (CHEMICAL ENGINEERING PROGRESS © OCTOBER 1993 ° 57.FLUIDS/SOLIDS HANDLING Orne ECMO a 4X os oxy AND DX aR IK 2) > 5660973 018448 g= ate h=— 1800705 F-04838 4, Select a holdup time from Table and calculate the holdup volume. V,, using Eg. 15 If the surge volume is not speci fied, select the surge time from Table 2 and calculate the surge volume, V,. using Eq. 16. 6, Obtain un estimate af L/D trom Table 5 and initially calculate Whe diameter according to: 4th) I/D) Hoo (Round to nearest 0,5 fi.) Caletlate the total cross-sectional area A, =2D° 2 yD 33 Calculate the low liquid level height, H,,y- using Table 3 or Hy =0spern OD where D in land round up to the nearest in., iD <4°0", H,,, = 9 in. 5B» ocroaer 1993 « PrAN uncut Seu CR nT Ne Wal Tickno ‘Surface Aros ‘ind We) She 2D sol BE-1aP ** 21 lips Heads 10808 are-07P te Hemispherical Heads Pou a ase E-ONP Dishes Heads oene Seine weenie) P. design pressure, psig (typically, operating pressure + (1530) psi or 10-18%, whichever greater Appropriate Vessel Height Notes: design pressure, "F typical, operating pressure +2550°F i, > 20°F I Ty < 200°, 250°F + under 60°F does not reduce wall thickness + ifoverpressure caused by boing, shauld Be Ty, ‘D,diameter, in. ‘S. allovrabie stress, psi (Reference 9) E joint efficiency, (08-1.0), 0.85 for spot examined {corrosion allowance, in, picaly oto qi. tin, orger of and gto nearest in) 1.0 for 100% x-ray joints 8. Using H,,,/D, obtain A, using Table 6 and calculate the Low Hiquidl area. A) 9. 1F there is1no mist eliminator pad. the minimum height ofthe vapor disen= sment area (A, isthe larger of 0.2D or 1 fi. If thete is 4 mist eliminator pad, the minimum height of the vapor disen- area isthe larger of .2D or 2 fi, Henec. set H, to the larger of 0.2D or 24.(1 Nit there is no mist eliminator. Using H1/D. obtain AA, using Table 6 and calculate A 10, Calculate the minimum length to accommedate the liquid holdup 11. Calculate the liquid dropout time: Hl “U 6 s 26) CHEMICAL ENGINEERING PROGRESS “A, 12. Calculate the actual vapor velocity. Uy, 1 2 27 Uys fells en 13, Calculate the minimu required for vapor-liqutid dis ment. Ly n length Lag = ge (8 IIL < dg then set L = Lyyy (Vaporiliquid separation is control- ing), This simply results in some extra holdup. If Ly, >> £, then increase H, and repeat [rom the step OEE > Lys the design is acceptable for vapor/liquid separation, If 1, >> Lyya (Liquid holdup is controlling). L can only be decreased and Ly, increased if H, is decseased. Hy, may only be decreased if itis greater than the minintum specitied in the step 9.Fee Aer esos TU CT CCle eis cner culation) (Calculations would have to be repeated [rom the step 9 with reduced H,). Calculate L/D. IF L/D > 6.0 then increase D and repeat calculations. from the step 6. If L/D < 1.5, then decrease D and repeat calculations From the step 6. 15. Calculate the thickness of the shell and heady according to Table 7. 16, Calculate the surface area of the shell and heads according to Table 7. 17, Cateulate the approximate ves- sel weight according to Table 7. 18. Increase and decrease the diameter by 6 in. increments amd repeal the calculations until £/D has ranged from 1,5 10 6.0. 19, With the optimum vessel size (minimum weight), calculate normal and high liquid levels Ay, =n + Vfl 2%» With A,, /A, obtain #,,, from Table 6 Hy, = D-H, G0) Example: Size « horizontal sepa- rator with a mist eliminator pad w 145,600 ti p ~ [4197.90 98.95))"" - (0.6)(5.0) Equation F. 10.09 fr'ls 19.79 fri 1 0.28. fis 01 fils, use 5.0ft G 293.84 2955 fi ee ee Size a horizontal separator with a mist eliminator pad. separate the following mixture. The ‘operating pressure is 975 psig and the holdup and surge are to be 10 min and 5 min respectively. Use a design tem- perature of 650°F, See Table 8. 1, Calculate the vapor volumetric flow rate (Eq. A). 2. Caleutate the liquid volumetric flow rate (Eq. B). 3, Calculate the vertical terminal velocity (Eq, C): K=0.13 (GPSA val- ue divided by two since “no mist eliminator” value is used) U, = 0.75, =0.29 fs 4, Calculate the holdup volume: V,, = (10 min.) (19.79 ftmin,) = 197.90 1 5, Caleulate the surge volume: V, = (5 min, (19,79 t/min.) = 98.95 fe 6. Assume L/D = 5.0. Initially set the diameter (Eq. D) A, = Id (5.0/1) = 19.63 1. 7. Caleulate low liquid height: Hy, = (0.55.0) 47 =9.5 in,, use 10 i. 8. Caleulate the low liquid fevel area: H,,,/D = 0.167 USing Table 6, 4,,,/AT=0.110 Ay, = (0-110) (19.63 ft) = 2.16 f° 9, Set Hy = 2, H/D = 25 = 04 From Table 6. A/a, = 0.374 Ay = (0.373)(19.63) = 7.34 12 10, Calcutate the length to aecom- modate holdup/surge (Eq. B), 11. Calculate the liguid dropout time (Eq. F) CHEMICAL ENGINEERING PROGRESS © OCTOBER 1993 © 59FLUIDS/SOLIDS HANDLING ee 12, Calculate the actual vapor veloc 9 Ai 13. Calculate Ly, = (1.37 Uy, = LS «137 ss fsy6.908) =9.45 fe 7.34 fi 14, 15> Lig, but ff iy minimum and cannot be teduced so Leannot be |W guatinn 6 reduced | LD=29.5)5.0=59 | Calculate the thickness of the shell and heads according 10 Table 7. Uor2x60) * Table 9, use 2:1 elliptical heads + Assume E= 0.85 # Assume SA 516 70 Carbon Steel, Design Temp. = 650°F 2(17,500\0.85)- 1.21 Bquation H. + From (9), $ = 17,500 psi + Assume corrosion allowance = — Vivi + P2975 x 1.1 = 1,072 psig ¢ Eq. H) (1072}(60) f= J (1 7500V(0.85)— (0.2)(107. user in. (See Bq. 1) use, in, andl use 1 = 2-4 in. 16. Calculate the surface area of the shell and heads aecording to Table 7. Ag = R(S.0011) 9.5 1) = 463.38 fe and Ay = (1.09) (5.0 192 = 27.25 10 17. Cateulate the approxi weight (Bg. J): = 50.224 tb _|t 18. Try P= 5.5 te and repeat calcula Wo= |[490 weight of shell Equation I. wate vessel \ lia inl fr tions untit minin and heads is obtained, For a free copy af this article, send in the Reader Inquiry card in this issue with the No. 156 ciecled. ——+(463.38 2.240 in, | fl +(2)(27.25)_fP)| Fics Fgincoring, Faeitiew Literature Cited ‘Chemical Engineering Conference (1988). 6, American Petroleum Institute, 1, Saudets M, and G.G. Brown, Ind Eng, | Chem. 26418, p. 98 (1938), Recommentied Practice $21 (1982). 2. Gans Processors Supplies Association, 7 Wathlat, RN. “String Sepattor and “Accumilters” Hedrmearbon gineering Dita Boo. Leeann Pressing 461}, p. 288.256 (1967). ‘Vol. Chapter 7 (19875. ‘8.Gerunda, Arthur, ‘How To Size Liquid O40. York Company toe, "Att aa ‘Vapan Separators,” Chem. Eng. p- 81-84 ‘Elimination in Gas Treatment Plans and Refierion” Engincring. Parsippany. 2, (EBD. 9. ASME Pressure Vessel Cad Section 4. Perry, Robert HL and Cecil H, Chilton, bss ‘eds, “Chemical Engingers’ Handbook,” en ewe 1, Table UCS-23, p, 270- 11986), Sihedtion, Chapter 2, p, 2-6 (1973), eee W. ¥. SVACEK isa pratesso In th department ‘of chanics and petesoum engineering atthe 5 Carpentier, PUL. Aomportant Parameters or Cost Effective Separator Design.” ‘Shell Oil Cormpany-Head Ortice ‘Universi of Calgary, Caigary, Abo, Canada 1et20 578; Fn B22 348; Eomait sent 2 scsucsigny car Seek previous work a esnirsystms engingorin he onl systons group of Monsante Company, St Lous, MO. Upon teasing Monsanto, he joined the Univers of Waser Ota. He tateivd bis 8Se and PHD egreesin chemical nthe University of Albena, W. MONNERY isa PhO candidate at ne Univesity of Calgary, Calgary, Alberta, (4220-751; Fo: 409/282 3046; E+ a) Hels ravoarching tte preictan of physical proper ties He prevoutly worked es a process: ‘eer tor Colt Engineering Corp. and Laval. Ine. Mr. Monnecyceceived tis BSc and MS ‘Gegrees in chomicalengincerngtiem tha University et Calgary 60 + OCTOBER 1993 © CHEMICAL ENGINEERING PROGRESS: