Sei sulla pagina 1di 34
«2 United States Patent Hanley (10) Patent No. 4s) Date of Patent: US 8,449,727 B2 May 28, 2013 (54) METHOD OF DETERMINING FLOOD, POINTS OF PACKED COLUMNS (75) Inventor: Brian Hanley, Lowell, MA (US) (73) Assignee: Aspen Technology, Ine. Burlington, MA (US) (7) Nose: Su any ime, theta f his rat is extended of adjusted under 35 USC. 1580 by 452 dyn (21) Appl. Nos 127765,611 (22) Filed: Apr 22, 2010 ws) Prior Publication Data US2011/010079 AL May 5,201 Related US. Application Data Provisional application No. 61/276,098, fled on Sep. 8, 2088, (60) (1) neck. BOLD 342 US.C usb Field of Classification Search spc’ 20511,2 See application ile for complete search history. (2006.01) (2) 2032 68) 36) References Cited PUBLICATIONS ister otal, Realistically Prodi Capacity and Presure Drop for Packed Columns, Ja 2007, AICHE, pp 2838 Billee R, "Recent Investigations of Metal Pall Ring", Chom, Eg. Prog. 640) (1987) Broadbent SR; Hammersley, JM, “Peesaton Processes Cry tals and Mazes, Proc. Cambridge Pils Soc. 53 (1957) Ender, C5 Hanky. 8 "Packing Pressure Drop Estimation via Bilinear Repression”. 2003 AICHE Spring Mig, New Orleans. LA, piper fe stam JW, “Percolation and Cluster ize" (Gitte! Phenomena,s0l2,C-Domb ans Press, New York 1972), Fiz, CW, al, "Performance of Structured Packing ina Commer: ‘dal Sale Colum at Proauresof0.021027.6 Bat" dad Eng Chom, Fes. 382) 1999). [E:Hammoisty. 1M, "Pecoaton Processes and Related Phase Pramiions and ‘Greeneds, Academie Topics" J So: Pads. Appl. Math 11) (1963), (Continved) Primary Examiner — Robert A Hopkins (74) Attornes, Agent, oF Firm — Hamilton, Brook, Smith & Reynolds, PC. 6 ABSTRACT A flood point fora packed column i determined by providing ‘dataset of gas pressure drop vals asa function of gas Row ‘ate values at Several liquid Now rates through a packed eol- ‘uma, known food point value for ane liquid How rate, sting ‘ood point values for higher liquid flow rates at values lower than the known flood point value, and setting flood point values for lower liquid flow rates at values higher than the known flood point value, followed by expressing gas fle ‘ates for Hiquid flow rates as factions ofthe oo point vale {oreach respective liquid low rte. Ata constant gs presse ddop, the method then includes ealeuating an average frac- tional flood point vale forthe liquid How rates and minimiz~ ‘ng the standard deviation between the fretiona flood point valuc at different liquid flow rates and the calculated average {actional flood point value by iteratively resetting fractional ‘Mood point wales and recalculating the average Irational flood point valve for the Tiguid flow rates, thus resulting in {determining a flood point forthe packed column at any liquid flow rate, and thereby producing a plot of pressure drop as a unetion of tration of ood point at any liquid Naw mate, oF a ‘mathematical expression thereof that can be used in com pter-implemented coluina design ad process modeling. 7 Claims, 24 Drawing Sheets a US 8,449,727 B2 Page 2 PUBLICATIONS GGalam, : Mauger... Sito Percolation Thresholds in All Dimen- sons" Plyuica 4208 (1998), Gingold, D :Tab,C J, “Peeoatve Condution in Thee Dime Sons" Phys es 17 42(13) (1990) Hanley, B; Duabobbia, B; Benet, D. “A Unified Model for ‘Countcreanont Vapor guid Packed cols 1. Pressure Drop. nd ‘51g. Chem. Res. 38 (1994), Hong. DC; Stanley. HE; Conili, A: Bunde, A. "Random- Walk Approach othe T#0-Component Random-Conducor Misr: Per Turbing Asay’ Tom the Perec Random Resistor Network a Ran dom Superconducting: Nawork Limits", Phys. Rey. B 337) (1986) Kean, 1A. Tumer, HIM: Price, BC. “Suuctued Packing Proven Supetoe for TEG Gas Drying" Oi © Gas Sep.28. 19, p-4L ister, HZ. Gil. DR. "Polit Flood Point and Pressure Drop for Modern Random Packings, Chem. Eng. Prog 872) (199), Krchenvinke, Ht Knapp, Hy “Pressure Drop and Fhoding in Packed Clams Operating at igh Pressures Chem. Emg Tech 10 987) Landau, DP, “Computer Simulation Suuies of Cita! Phenom- ena", Piped 4208 (109, Lemor BJ.,Grove, CS, "Cia Copdtins oF Two-Phase Flin Packet Column, Ind. ng Chom. &X(1) (1981) ‘Meier, W Stoesker WD. Weinstein, B.,"Peformance of « New, igh Efcieney Packing", Chem. Ene Prog. Now. 1977.7 Rukovena, E: Koshy. TD. “Packed Dilation Tower Hydeaubic Desig Method an Mechanic Considerations”, Ind Eng Chm Bes. 521993). ‘Schulles, "Now Comparisons of High Eetive Modem Dumped Packing and Systematic Packed Columas", Chem Emp. Ser 1220)(1997, ‘Shoting, W; Zuosiang, 2: Xuan. “nvestigntion on Restifhing (Characteristics of Two New Tower Packing: Chen Ind & Eng. (China) 21990), ‘Wallis .., One-Dimensional Tvo-Phaxe Flow; MeGraw Hill, 2nd dion, New York 1979. Yeomans, IM, Stasical Mechanics of Phase Mansions Clarendon Pres, Oxron. 1995 Zanati, Rs Shot Hs Hope 4. “Boosting Tower Performance by ‘More han a Tree", Chem. ig « May 27, 1985.22 Zenz, FA. "What Every Engincer Should Know About Packed ‘Tver Opciations", Chem. Eag, Aug. 1983, p. 176. * cited by examiner US 8,449,727 B2 Sheet 1 of 24 May 28, 2013 U.S, Patent gz pe 22 of oF ot ca VIO 2d zh ot 80 90 vo 20 00 9s°0 US 8,449,727 B2 Sheet 2 of 24 May 28, 2013 U.S, Patent zupze © 9zeze © eee» veoee + arvec + eevee « i weorzee — MezOSE + en 4s | wea ee", fialo oe DIRE U.S. Patent May 28, 2013 Sheet 3 of 24 US 8,449,727 B2 FIG.2 5 <—- Loading g lg tone A, /; rs ja Ory packing feg gas rate FIG.3 U.S. Patent May 28, 2013 Sheet 4 of 24 US 8,449,727 B2 Air mass velocity, 1b/ FIG. 4 Bunoed 4/5810 0 U.S. Patent May 28, 2013 Sheet 5 of 24 US 8,449,727 B2 a § E 2 3 5 7s g = S & <4 <4 1 os os os oa 03 System Ait, be tomer in Sencar | on Oa OSer0ee 0 a 2 3 sos Fs, fi/s-(Ib/ft3)!2 FIG. 5 U.S, Patent May 28,2013 Sheet 6 of 24 os o.1sE[ mMTP 50 Air/Water 0.14 FE) 4 to 30 gpmvt? 0.12 Zo10 0.08 0.06 0.04 0.02 0.00 0.0 1.0 2.0 fu/sec [Ib/ft"]* FIG.6 3.0 US 8,449,727 B2 US. Patent 79 May 28,2013 Sheet 70f 24 US 8,449,727 B2 NO NO 100 Fix one vapor load n at flood, Estimate the others | Calculate fractional Rn Moods: x=Fy/Fsp for all (Fs, Cu, Ap) I Fora given (%,C1, Ap) ple, get other values of Ap at the same x, from remaining liquid load curves. 74 15 ALG, Cu AB) done? YES Calculate the ue variance or standard deviation ofthe Ap values ¥ Calculate toad n variance or standard deviation from all values calculated above. 8 ee Estimate new Foy values via least squares regression 701 Report Fee values, FIG.7 U.S. Patent May 28, 2013 Sheet 8 of 24 US 8,449,727 B2 0.18 IMTP 50 0.16 4 Air/Water 0.14 -4|4 to 30 gpmvift? 0.124 = < 0.10 4 8 & 0.084 0.06 4 0.044 0.02 4 0.00 T T T T 0.0 0.2 04 0.6 08 1.0 Fractional Flood FIG.8 4 = a 4 Flood Comparison 22 IMTP #50 air/water — Literature curve @ Data collapse method 1 L 0 10 20 30 GPM FIG.9 U.S. Patent May 28, 2013 Sheet 9 of 24 US 8,449,727 B2 Ambient Air-Weter Systems at 1 atm, 70°F (C-factor (f'sec) Laut Loading gpm) FIG. 10 U.S, Patent May 28, 2013 Sheet 10 of 24 (ssequo,suaunp) dova sunssang os 0 0a US 8,449,727 B2 FIG. 11 Fractional Flood U.S. Patent May 28, 2013 Sheet 11 of 24 US 8,449,727 B2 mH a . fos : Sos : ; . g 04 . g oe £024 we ° ae Fractional Flood FIG. 12 meee wee 1 0.00 Dovitisil (002s cao cueOsmenTom eT Fractional Flood FIG. 13 U.S. Patent May 28, 2013 Sheet 12 of 24 US 8,449,727 B2 00 02 04 06 08 10 Fractional Floed FIG, 14 TOT TooTeT 0.16 E-[3:4" plastic Fexiring :4 ag EL aia 4 te tiiat 00 02 04 06 08 10 12 Fractional Flood FIG. 15 U.S, Patent May 28, 2013 Sheet 13 of 24 ToT T ass Durapac 015 PL airwater a So10F 4 0.05 F a 4 0.00 Coated, L 00 02 O4 06 10 Fractional Flood FIG. 16 016 ro om - 4 . on | I 2g 0.10 4 Sos E 4 F006 F 4 oot 4 0.02 4 0.00 Oeil 0.0 02 04 06 08 10 Fractional Flood FIG.17 US 8,449,727 B2 U.S, Patent May 28, 2013 Sheet 14 of 24 020 ToT Tuo Nuer Ring # : ous 7 4 rs a s ' Sok : 4 oos 4 0.00 0002 04 06 08 10 Fractional Flood FIG. 18, eae T T T T T ous ed oe 4 ona . 4 S010F 4 § Boose 4 cos F- 4 aoa 4 002 “ 4 - 000 Exact at 0002 04 06 08 10 Fractional Flood FIG, 19 US 8,449,727 B2 U.S, Patent Sheet 15 of 24 May 28, 2013 : oor 7 4 0.00 00 02 04 06 08 10 Fractional Flood FIG. 20 ols 0.16 ola (apZnyg 00 02 04 06 O8 10 12 Fractional Flood FIG. 21 US 8,449,727 B2 U.S, Patent 12 os 06 Pressure Orop (dimensionless) Sheet 16 of 24 May 28, 2013 US 8,449,727 B2 04 0608 Fractional Flood wo 12 FIG. 22 = cata sth Degree Polynomial oa 06 os 1 Fractional Flood FIG, U.S. Patent May 28, 2013 Sheet 17 of 24 US 8,449,727 B2 39 54 mol % EB ~ 120 tom Liquid Ethylbenzene Density (kg/m) 98.5 mol % EB 100 torr 5.94 0.014 FIG. 24 880 840) 830 820 210) L % 100 Temperature (°C) FIG. 25, U.S, Patent 000 Sheet 18 of 24 US 8,449,727 B2 May 28, 2013 rene Monomer 200) T= 100% P= 182117 tor Temperature (°C) FIG. 26 te) FIG. 27 U.S. Patent May 28, 2013 Sheet 19 of 24 US 8,449,727 B2 a pe ‘Aiivater Flood Date goto £ C009 cos F : 07 F Os L L L ‘0.000 0.005) 0018 0.020 0.025 Cus (m/sec) FIG. 29 U.S. Patent May 28, 2013 Sheet 20 of 24 Mole Fraction Ethylbenzene in Vapor F 39.54 mol % EB 120 tom US 8,449,727 B2 1 wr 4 Ryan = 8.38 gp five penn oa oak oa oof 26 oo os eae 8 ‘Mole Fraction Ethylbenzene in Liquid x0 Vr 249 ass on D =4058 7 ly 98.5 mal °e EB 100 torr = 9.998 = | Ve A [sp35eq /. = Le 44.8968, S.94 kg see Ip FIG.31 0.01472 mol % EB U.S. Patent May 28, 2013 Sheet 21 of 24 US 8,449,727 B2 ‘Mole Fraction Ethylbenzene in Vapor 0.0 02 04 rs 08 1.0 Mole Fraction Ethylbenzene in Liquid FIG. 32 U.S. Patent May 28, 2013 Sheet 22 of 24 US 8,449,727 B2 144 z é T T T T 02 a4 os 08 10 Mole Fraction EB in Liquid FIG. 33 U.S. Patent May 28, 2013 Sheet 23 of 24 US 8,449,727 B2 860 960 840 230 780 i . 75 100 125 150 Temperature (°C) FIG, 34 U.S. Patent May 28, 2013 Sheet 24 of 24 US 8,449,727 B2 4-30, 60 Central Processor VO Devices Network Interface Unit Interface 82 a 86 “System bus 79 Memory 90 Disk Storage 95 Data 94 FIG 35 US 8,449,727 B2 1 METHOD OF DETERMINING FLOOD, POINTS OF PACKED COLUMNS. RELATED APPLICATION ‘This application claims the benefit of U.S. Provisional Application No. 61/276,088, filed on Sep. 8, 209, ‘The entire teachings of th above application ae incorpo rated herein by reference, ‘A portion of the disclosure of this patent document ‘ains material which is subject co copyright protection. The ‘copyright owner has no objection tothe facsimile produ: tion by anyone of the patent document o the patent disclo- sure a it appears inthe Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights ‘whatsoever BACKGROUND OF THE INVENTION Phase transitions in physical systems have been an active area of scicalific investigation for many years. Solid Liquid ‘por trnstions such as those that occur between ie, water, and steam are undoubtedly familiar examples. Another com ‘mon example is sugar erystal formation in honey that has been lft to stand fora long time, Phase transitions, however, ‘oceurin many oiler types of systems—for example, the tran sition from ferromagnetic fo paramagnetic behavior for fer romagnets atthe Curie temperature, oF the transition fram nonmal conductivity to supereonduetivity in eerain metals and ceramic oxides at critical emperatre,ortheconductor! insulator percolation threshold for electrical networks. ‘Phase transitions are classified as being ether Fistorder of ‘continuous or, more infrequently, second-order) phase tan fitions. First-order transitions such ae those between ioc, ‘Wate, and steam, iavolve the evolution or absorption of heat the transition point. Continuous transitions, by contrast, are not accompanied by heat transfer. In addition, continous phase transitions are accompanied by the growth offic tions on ever-longer length scales. Transitions are further ‘organized into groupings called universality clases. For sys= tems from the same universality class, renomalizing raw ‘experimental data to the proper ential conditions has the remarkable effect of “collapsing” data onto what is essen- tially a master curve. FIG. 1 shows two examples of this phesomenon: FIG. 1A demonstrates that pressurevotume! Temperature data for several different Aids fll on top of one ‘another when the data are properly renormalized to the eriti= ‘al point; FIG, 11 ilusiates thatthe same type of collapse ‘occurs for Ferromagnetic materials near the Curie tempera ture. SeeJ. M. Yeomans, The Statistical Mechanies of Phase Transiions,(1992)p. 28 for FIG. 1A, and p. 119 far FIG. IB. ‘Modeling of phase ransitions is important for designing , for the guid low rates, as shown in Fg. 2 above, and minimizing the standard deviation cr between the fractional flood point value at different ligud flow rates andthe calculated average factional food point valve, Stops 74-78 in FIG, 7 are ills- tnsive, The standard deviation is calculated as shown in Eg. 3 above In practice, the minimization procedure 78, 79 is car ried out forall sets of four points at constant pressure drop, for all available data, At least one fooding velocity needs to be fixedat the start of the fi. Leastsquares canbe used (step 701) to perform the optimization by setting the objsctive function to be zero forall points (the standant deviation or variance should be as close to er0 a8 possible) The overall procedure 100 as outlined above assumes that constant presse drop data for several liquid loads are readily available. In terms of aw data this will rarely be true— experiments are usually not performed in a mode wherein pressure drops are held eonstat while vapor load is varied at ‘constant liquid loa. Instead, most experiments, as shown in FFIG. 8, hold the liquid load constant while varying the vapor load, Pressure drop is the mensured variable. The required constant pressure drop data ean be estimated in a number of ‘ways. For example, smooth curves can be drawn through the {ata and then points at constant pressure drop can be read off the graph. Alternatively, one could incorporate some type of ‘oumerical interpolation procedure (Tinea, cubic spline, or smoothing spline for example) into the least squares routine 71 ‘The invention method 100 iteratively resets Fractional ‘ood point values (Step 701) and recaleulates (loop 79) the verge fractional ood point value forthe liquid flow rates thus resulting in determining a flood point for the pecked columa at any liguid low rae, and thereby producing at 703 plot of pressure drop as. funetion of fraction a fleod point atanyTiguid flow rat, or a mathematical expression there. ‘The results fr food velocities of IMTPX #50 were: epnf3.6882 sect] 13899 1Useo[I/10)"2 (6 sect ]* [30067 fuseefIb/R°]!? The method 100 described above defines new criteria for ‘ooding: the appropriate choice of food points forgiven set ofppacked column pressure drop data isthe choice tht results in the best possible collapse ofthe data onto anified curve ‘The graph of the airwater pressure drop as a function of faction of flood point for IMTPX #50 is showin in FIG. 8 FIG. Seompars these estimates tomaximum capacity data seported by a vendor of IMTPS #50 (Saint Gobsin Norpro, Akron, Ohio). The agreement between the literature curve US 8,449,727 B2 1 ‘andthe results obtained above is quite good, Further the data ‘collapse is also visually quite good. The only curve that noticeably deviates from the others, as indicated in FIG. 8, is the curve for liquid load of 4 gpm “Another example of using pressure drop and airhwater ow rate data will be used to illustrate that the method 100 described above yields the best unified curve of pressure rap ‘as. function of fraction of flood point, and produces flood Point values that are the appropriate choice Of flood points that are not necessarily the same values as the conventionally ‘chosen floed point values, The aitvater flow rate date for plastic Jaeger 2" Tr-Packs® provided by the packing manu- ‘actress shown in FIG. 10 lueger Products, Inc. Houston, ‘Tex. The conversion feetor between the C-Factor shown in FIG. 10 and the F, gas Bow rate is [p=]? thats Fis ‘obtained by multipiying the C-Factr vine hy the square root ‘of the diflerence between the liguid and vapor densities, and pp respectively, that is, [62.4-0 074] *-7.8947. The flood point value at liquid How rate 30 gp/t* was held fixed. The unified pressure drop as a funetion of fractional ‘ood point obtsind bythe method described above is shown in FIG. 11. The flood point values atthe water low rates showin inFIGS. 10 (6,20, 30, nd 40 gpa) obtained by the invention mothod are listed Table 1, and compared with the values shown in FIG, 10, TABLET feed epee enema eee owiite "“venion lec(lb fe)! bom ie fe insect » S Table 1 shows thatthe flood point values obtained by the method ofthe invention are significantly different fom the ‘conventionally obtained values, particularly at lower liquid flow rates. In particular, the results showa in Table | indicate that ata low Tiguid low rate (eg, 6 gpa") the column will ood a lower gas flow rate than expected. which could have negative consequences for eolumn opegtion, By contrast, at high liquid flow rates (e240 gpm/T?), the column ouput ‘could be inereased about 3% before flooding, a potetilly ‘gnifleant increase in capacity. FIG. 12 shows the unified data obtainedby usin the man {actrees lod points listed in Table 1, that is, without using the minimization ofthe standard deviation described above in eps 78, and 79 of FIG. 7. It can be seen from comparing FIGS. 11 and 12 that the Rood point values obtained by the ‘method 100 ofthe invention produce a beter unified curve of pressure drop as a function of fractional flood than the con- ‘ventional flgod point values FIGS. 13-22 show several other examples of unified pres+ sure drop data as @ function of fractional food point for @ variety of packing styles fom a number of different vendors. is important to note tha camrying out the data colepse procedure desribed above did not require any assumption of ‘a funetional form forthe pressure drop versus gasliqhid flow rate data. The advantage ofthis data collapse procedure is in the development of correlations to predict the pressure drop 1nd approach ta flood for a packing material. Instead of having to deal with thee variables —lguid loading, vapor o 8 Joauing, and pressure drop—the problem is reduved to one involvingoaly two variables approach to flood and pressure drop. Then, a mathematical expression can be obtained fora ceurve fitthrough the pints, suchas, for example, fifth order polynomial as shown in FIG. 28, where the data shawn in FIG. 11 is fi toa mathematical expression ofthe form, prALOSSS 3.606 ATI IH 280 faosnaeaaiot o=xel @ where xi the fractional flood at constant liquid load und y is te dimensionless pressure drop. Such an expression can be ‘sed ina computer implemented column design and process ‘mostling enabled by the present invention, ‘Te method ofthe present invention for optimizing «prod ‘vet output rate from a distillation column having atop and a bottom includes setting a desired product ouput rate from the

Potrebbero piacerti anche