Вы находитесь на странице: 1из 13
Evaluation of Inclined-Pipe, Two-Phase Liquid Holdup and Pressure-Loss Correlations Using Experimental Data G.A. Payne,* SPE-AIME, U. of Tula C.M. Palmer,** SPE-AIME, U.of Tulsa J.P. Brill, SPE-AIME, U.of Tusa HD. Beggs, SPE-AIME, U. of Tulsa Introduction ‘Two-phase flow in pipelines located in hilly terrain is encountered in the petroleum industry frequently. In oilfield gathering systems, two-phase mixtures must be transported from the wells to the separation facility. Because of the problems associated with oil and gas production offshore, it is usually necessary to have a common pipeline for the liquid and the gas streams. It is expecially important to have good design methods for sizing these pipelines. When designing two-phase pipelines, pressure losses and liquid holdup must be predicted. The liquid holdup is defined as the fraction of pipe oc- cupied by liquid during two-phase flow. A value of liguid holdup also is an important consideration ‘when designing separation equipment, slug catchers, and pumps. ‘A great deal of research has been conducted in horizontal and vertical two-phase flow, and several good correlations exist for these cases. However, only limited research has been performed in inclined two-phase flow ‘The main objective of this study was to design and construct an experimental facility that could be used to investigate two-phase flow phenomena in pipelines laid in hilly terrain. The secondary objective of this study was to evaluate several existing correlations for predicting liquid holdup and pressure losses using data obtained from the test facility. Literature Review Several authors have investigated inclined two-phase flow to some degree. An actual field study of a 16-in. pipeline was conducted by Flanigan' in 1958. Pressure drops over various sections of the line were measured. He concluded that the inclination of the hills had no effect on the pressure drop and that no pressure recovery existed in the downhill sections. Flanigan’s design method included using the Panhandle equation to calculate friction loss and an elevation factor to determine the loss caused by elevation. It is possible that Flanigan’s elevation factor could include some pressure recovery in the downhill sections. This is because any pressure loss, not accounted for by the friction term, is assumed to be elevation loss on the uphill side. The overall pressure drop was used in the development of this correlation; consequently, any pressure recovery that might have been present is included in the elevation term. ‘A two-phase inclined flow study was conducted in 1967 by Guzhov ef al.? Their data were taken in 2-in. pipe inclined at angles between 9° from horizontal. In development of the correlation, two flow regimes were considered — plug and stratified. A mixture Froude number and the gas input fraction were used to predict flow pattern, In stratified flow, there is one liquid holdup expression for uphill flow and one for downhill flow. This holdup is used to find a two- J-cm) diameter, 550-ft (168-m) long pipeline was designed and con- 1 structed in a hilly terrain configuration. Two-phase-flow liquid holdup and pressure~ Toss correlations were evaluated using gas/water data obtained from experiments. Accurate predictions were obtained using the Beggs and Brill correlation and a combination of Beggs and Brill and Guzhov et al. correlations. ioe {JOURNAL OF PETROLEUM TECHNOLOGY phase density that will determine elevation pressure loss. In plug flow, the stratified uphill holdup ex- pression is used for holdup in both uphill and downhill flow. This results in complete recovery of the elevation pressure loss. A friction loss expression is given. Evaluation of the acceleration term requires an iterative solution. Beggs and Brill’ investigated two-phase flow at several inclinations between +90 and —90" from horizontal. Test sections of 1- and 1.5-in. pipe were used. Holdup and pressure drop were measured, and, holdup and friction factor correlations were proposed. A horizontal flow pattern map consisting. of segregated, intermittent, and distributed flow regimes is used to determine a horizontal holdup. The horizontal holdup is corrected for inclination and then used to determine the elevation pressure loss. Friction and acceleration terms also are provided. Recently, Robinson‘ showed that the accuracy of the Beggs and Brill correlation could be improved when applied to directional wells by using a tran- sitional zone between the segregated and intermittent flow regimes. When the flow is in this region, a weighted average of the segregated and intermittent holdup values is used. Sometimes, horizontal flow correlations are used ‘with Flanigan’s elevation factor to design pipelines in hilly terrain. The American Gas Assn. Design Manual’ recommends using Dukler’s* horizontal correlation with Flanigan’s elevation factor. Dukler's correlation is based on a large amount of ex- perimental data. Dukler developed a holdup iti ee woe correlation that requires an iterative procedure. The holdup is used to determine friction and acceleration pressure drops. The two-phase friction factor is a function of no-slip holdup and a smooth pipe friction factor. Eaton's’ correlation also is used for pipeline design. This method is based on data taken in 2-, 4-, and 17-in. pipe under field conditions. Eaton proposed a holdup correlation that is a function of several dimensionless groups. This holdup is used to determine the acceleration component of the pressure drop. The friction factor is also a function of several dimensionless groups, but this must be used with caution because the friction factor becomes un- bounded as single-phase flow is approached. Experimental Facility A schematic diagram of the hilly-terrain-pipeline test facility is shown in Fig. 1. The test facility consisted of 1,200 ft of 2-in. Schedule 40 line pipe (ID = 2.067 in.) with associated gas compressor, water pump, meter runs, and separator. The entire system was closed —i.e., both gas and liquid were recirculated. ‘The test section was 400 ft long and arranged in three hills. Entrance and exit effects were avoided by extending the pipe about 75 ft (450 pipe diameters) on each end of the test section. Fig, 2 is a profile of the test section, and Table 1 gives data about that section. General Procedure for Two-Phase Flow Testing ‘The system was pressurized initially to 400 or S00 Psia with a three-stage Worthington compressor. Fig. 1 Schematic diagram of hilly terain facility. SEPTEMBER 1979 ‘TABLE 1 —TEST SECTION DATA Elevation Length ‘Change Angle Sections , tc (degrees) 1 50 3.69 423 2 50 375 -430 3 50 611 702 4 50 -331 -3.80 5A 100 028 “4 5B 6 650 824 6 54 558 591 This provided the necessary volume to start a two- stage Knight compressor. The gas was circulated with the gas bypass valve fully open. After the desired ‘meter run was opened, the bypass valve was closed until the desired gas flow rate was achieved. With the flow controller set on the desired liquid rate and the correct meter run open, the pump was started and the water bypass valve was closed partially. The pump pressure had to be about 200 psi greater than the system pressure for the flow controller to function properly, After water reached the separator, the liquid-level control valve was set to maintain the desired liquid level. Once the liquid level was stabilized, the gas rate was set more exactly. ‘When the system had stabilized completely and all temperatures, pressures, and flow-rate data were recorded, transducer equalizing valves were opened and the ball valves were shut. The liquid was drained from each segment and then weighed. The ball valves were opened and the equalizing valves were shut. If only pressure loss data were desired, the holdup procedure was omitted. A complete discussion of the experimental equipment, including pressure traverse ‘and holdup measurement procedure, is found in Refs, 8 and 9. The holdup data taken by Palmer* and the pressure drop data taken by Payne? are presented in Tables 2 and 3. Evaluation Techniques PVT Prope Because the liquid used here was water, we did not Fig, 2—Test section profile have to consider mass transfer between phases. We assumed the water had a constant density of 62.4 Tbm/cu ft. The gas/water surface tension was assumed to be a constant 60 dyne/cm for the range of pressures and temperatures encountered in this study. Water viscosity was determined from the correlation of water viscosity as a function of ‘temperature presented by Beal. Specific gravity of the natural gas was 0.64. The gas compressibility factor was calculated using a subroutine provided by Brill and Beggs," which is an acceptable approximation to the Standing-Katz z- factor chart, The Lee ef al." correlation was used to calculate the gas viscosity. Statistical Analysis ‘The statistical parameters used to evaluate the ex- perimental data were average percent error and standard deviation. These parameters are defined as follows: calculated value ~ measured value %E io ue x 100, ‘measured value Ewe, wE= St, NW Ly (0, - %EY s N-1 To make any statistical inference from ex- perimental data, the data usually must be distributed normally. For this reason, the actual values of percent error for predicted pressure drop in the total {est section were sorted into ranges, and the cumulative relative frequency of the percent error was plotted vs the value of percent error on normal probability paper. A normal distribution will plot as a straight line intersecting the $0th percentile at the ‘mean, It also will intersect the 16th percentile and the 84th percentile one standard deviation on either side ‘of the mean, Figs. 3 and 4 show that the errors do approximate a normal distribution. Because the average percent error is based on a finite sample, we cannot expect it to coincide exactly with the actual population mean, However, since the percent errors approximate a normal distribution, it is possible to determine an interval that will bracket the true value of the average percent error, given specified odds. For a normal distribution where the sample variance is used as an estimator of the population variance, this interval is given by the following expression: Using the above confidence limits and a = 0.01 (ic., a 99% confidence interval), we can show that the Beggs and Brill correlation true average percent error lies between -20.51 and -28.53%. Con- sequently, at best this correlation underpredicts JOURNAL OF PETROLEIIM TECHNOLOGY ‘TABLE 2— HOLDUP DATA RECORDED BY PALMER® Pressure Temperature Uigud Gas (psia) CF) Rate __ Rate Uphill Holdup Downhill Holdup Fun Init Out init Oulet (BO) (GoD) Son.t $00.9 Son.88 Ge? seas Sed Yat 525 sig 98 SO5D GORY 07295 O2HB 02489 02198 02101 1847 Zoot S79 572% 72 Sone Srraro atm O24 ozs Otaa4 doe O08 imo Ste Ste suo & Saad © someae aN Gand OaI90 Ot6rs OIs10 OLbaT Yaroo S05 Gre 1023 Sous Gamoro aus sans Cater Oana Oslo Oania oor Si sa ‘ee 75 fone Saouoa Oak ag0s0 O00 Bist] O3m9 OA 220s or 390 63 8 SuSO Tarr arte OLtese Give Otse2 cones O499 Zao! 685 S59 73 ues Teouet amen 2am O247S see Ofos0 Cate Zow2 616 909 oF 3 ues sTousea Of6a0 OYG=9 OTe O83 Oren ONTa7 ios 575 Sos 8873 Zane “vewtea Orme? goad tess 1406.00 03270 Buoi Sar Se 7} Sah Tootgo Ousi0 Osta? Gates Oro! canes Gate 2uce Sar Set To 84 Seg tamees Ours OfMe0 Csast 0002 OTe? Conee Fisoo See 52 G6 70 Base lomo ogra Gaeze Dear? Oars Oauea Gora Zoe 509 503 63 Gene ‘Sonat sara Gane Gsree 02000 sat aeT0 Fis 50 562 sms 7 zaps Yuoos Ospoo Oster Osiee Oto Oaass CTO? Fives Set 545 107 ea73taotre Oop Banco castor O2m00 O4O10 UES Ziwor Sia Se Ho 4) tara ‘Sous Ossos Oaze2 Gsz00 Ores sory aaeee Fisos Soo ez Toe Se Mara aoaaar Osoor Score Geen? Gimme Sten Cava Ziwoo 473 fee is 7% sara focosa aang Gens oats Otass Oma ote Bo 72 es M0 Te M579. Herest asad sete Gaver O2uea Oana oases rion ou Sou ‘Go '6 Seat asran Osror tooo Gisgd BteHs Glas Oaret Fioo2 S755 Gr 5051 Jomooed Ovi OI779 tee Otees Ofe10 Oo0 Hoos Sis 542 suo 65 Sent fotsees O2uas G2WeT oame2 OanT OTN Oxer boos Sov S20 fos 93 Soot “Trooat ame era Gaets O22 Ose O20 Boos Sie S10 08 & © Sre1 Soused asm os602 Oart0 Osem O2sro Ore Hoos 513 508 MoS soos mayen Oaaze cee? Osan O20 O4oca OST oor Soe 50) ‘We 78 Sano daacre O's) Ogos Gooey Gtzes tase Os0ar Hoe te 2% ho tata foura Octo Gana Ozer Orit GtGae Coors Zoo eo 35 10 © 70 fats ionel OaraD osore o.gmes Dom Oana dsm Bivd 43 435 268 2450 Hatze0 do Ova Gatoe Dari? cater O.eo8 Bo Bow | Fig, 3—Cumulative frequency of percent errors_in pressure drop (total test section) using the Beggs ‘nd Brill correlation, SEPTEMBER 1979 Fig. 4—Cumulative frequency of percent errors. in pressure drop (total test section) using the Eaton- Bukier-Flanigan combination, 201 ‘TABLE 9 - PRESSURE DROP DATA RECORDED BY PAYNE® Inlet _ Average Liquid. Gas, Pressure Drop Pressure Temperature Rate Rate Bun (sia) FD (BID) (sctiD) See, Sea? 3068 12101529 64 92 1793716 -30 -06 1210332 6 130 1708320 “18-14 12202 426 % 575 1401284 "38 28 12203423 7 857 957840 32 "20 e204 ait 7 et7 938778 24-10 12208401 7% 338 astm 48-02 12302 518 0 238 1395187 14 “12 12303535, ct 509 1961892 20-02 12904 526 4 505 670d “20 “08 12702378 7 2o1 gauge “08-04 12703384 2 505814055 16 08 12704 405 a7 1077 70271 “30 “18 12705 285 9 sor? 489842 “20 “1.0 12706 387 8 1080 28022 “14 “02 12707387 99 867 26111 12-02 12708427 9 864 926655 24 “12 12708 395, 8 864 659008 20 “08 12002 635, st 128 1494170 “10-08 12903 Git 20 421837799 1204 1290 605 8 515 703698 “16 02 12005571 B 0s 947422 02 12008 546 80 703985678 02 12907533, 80 703542509 08 22001412 nm 381 1007089, 08 2002 391 cd 3o1 raoz7a 12 “07 05 22003 385 80 331 $8047 0B -05, “04 22008 347 84 331 291484 “10 “03 “02 22008 308 88 331 251860 -08 -03 “02 22008 291 a7 3s1 22048 “07 “02 “04 22007 289, 87 331 149908 “07 “02 04 22008 286 88 31113662 08-02 “04 22000 285, 83 gar 7143 0B “02 “02 “01 2a010 284 9 a5 7148 09 “03 01-21 “04 Boor) 288 90 415 280048 07 “02 o1 “24 “or 2012 290 a7 415365469 10-03 01 227 “02 22502 431 2 515 319009 10-03 01 28 “01 22503 432 7 515 280475 11 “04 01 “28 “02 22508430 R 512 218097 “13-03 01 24 205 22505 430, 6 512 159057 “13 “oa 03 21 “08 22508 428, 78 515 95000 -13 “02 04 4B “10 22507 429, 8 eos 81808 15 “0.1 05 0B “19 22508 430, 9 eo¢ 175835 13-03 01 "09 20 22509 431 2 60s 230541 “11 “02 2 41 48 22510432 S 604 21048 13-03 2 3 43 22511435. 6 604 928608 13 “04 Of Ba 22512 439 98 73 311348 “13 “07 or 19 419 22513440 100 703 259689 “114-02 or 7 9 2514440 100 70317755613 “08 02 43. “22 22515499 101 703110171 “05 “03 22-14 22518 440 97 805 107612 “08 3 25-10 517 439 102 805167485, “08 03 45 “2 22518 405, fa 805 242856, “07 05 18 “22 22519 406 cy 805201122 07 “21 05 23 “22 ‘2002558 75 2331768601 “14 1g 415 “50 43 2008511 fa 2331319003, 08 17 07 “36 “07 32004 470, 85 23351544 02 “15 “01 "20 “03 32005477 a7 130 606329 “02 0 “03 “15 “04 32008489 2 131 1363708 10-05 “02 “29 “03 ‘32007 488 ot 1311779816 “13-22 10 “50 “15 32008435 97 131 3aate. 303 “09 “or “4 “05 32000 433, 98 131 262084 01 “09 “or ia “93 B2010 435, 102 131 219786 “03-07 “02 “14 “04 Be011 420, 107 131173545 01 14 "02 215 03 32012 425 108 131118403 06 10 “98 “16 “04 32013421 108 8115796 07 “04 “14 “07 “is “04 32014 431 108 8 171138 03 “01 "16 “06 “14 “02 32015450, 7 8% 246550 -03 -01 “04 06 12 “04 32018485 102 8 980687 -03 “01 “03 -08 “12 “04 S217 491 100 4 130383 “05 “08 “03 "12 “24a 42 32018510 100 4 1049565 -08 10 “08 "15 “33 41 JOURNAL OF PETROLEUM TECHNOLOGY ‘TABLE 4 — CUMULATIVE FREQUENCY DISTRIBUTION OF PERCENT ERRORS IN PRESSURE DROP (TOTAL TEST SECTION) Beggs and Bril (modified by Robinson) Beggs and Brill (neglect pressure recovery) Beggs and Brill rough pipe fiction factor) Dukier (HL), Dukier (FF), Flanigan (EF) Eaton HL), Eaton (FF), Flanigan (EF) Guzhov ef a pressure loss at least 20% for 2-in. pipe. A similar analysis for the Eaton-Dukler-Flanigan combination gives a 99% confidence interval of 1.16 to 10.80%, Table 4 gives the percent error distribu- tion of the other methods used to compare calculated pressure drop with the measured pressure drop. Holdup Analysis Liquid holdup was measured in three uphill and three downhill sections for each experiment. Calculated values of liquid holdup corresponding to the measured values were determined from the Beggs and Brill,” Flanigan,’ and Guzhov et al? correlations. The results of this analysis are provided in Table 5 and in Figs. 5 through 10, Examination of the individual percent errors for the uphill sections indicates that the correlations predict the holdup more accurately in some sections than others. This fluctuation could result from the presence of slugs in some parts of the test section. As can be seen in Table 5, the Beggs and Brill correlation was the most accurate in predicting the liquid holdup in the uphill sections. In downhill flow, the fluctuation of the holdup percent error in individual sections is similar to that found in the uphill sections. This also could be we 0 0 4 6: 87 eo m9 . Oo 33 33 50 87 93 10 | 0 2 30 6 73 8 99 100 0 0 0 11 31 84 97 100 0 10 20 44 64 73 96 100 7 84 8 88 88 9 % 100 caused by slugging. The Guzhov ef al. correlation gave the lowest average percent error of the three correlations. Using the Guzhov ef al. criteria for stratified flow, liquid holdup in downhill flow is ted most accurately by Guzhov er af. in stratified flow and by Beggs and Brill elsewhere. Statistics for this combination also are shown in Table 5 ‘The Beggs and Brill correlation was used in a modified form to see if the downhill holdup could be predicted more accurately. The correlation was forced to a segregated flow condition for those experiments where stratified flow was indicated, according to the Guzhov et al. criteria. This modified form proved to be less accurate than the original Beggs and Brill correlation for downhill flow. The consistently large positive percent errors for liquid holdup in the downhill section indicates that the correlation is overpredicting the downhill holdup. Table 5 shows that the Flanigan correlation was the least accurate of the three correlations. AS Flanigan stated, the correlation was a crude ap- proximation, but this was one of the first steps taken to predict liquid holdup. Flanigan’s correlation was developed for uphill flow, and holdup in downhill flow was ignored, Fig. Measured uphill noldup vs calculated uphill holdup using the Beggs and Brill method. SEPTEMBER 1979 Fig. 6-Measured uphill_ holdup ve calculated uphill holdup using the Flanigan method. 203 Pressure Loss Analysis ‘The calculation procedure used to determine pressure loss in the test section is given in Ref. 9. Six different correlations or combinations of correlations were compared with the experimental data. Table 6 and Figs. 11 through 18 give the results of this com- parison. The statistics in Table 6 are based on the total pressure drop for the entire test section. A. Positive value of average percent error indicates an overprediction of pressure loss. The Guzhov et al. correlation was significantly worse than the others. This was attributed to the downhill holdup being equal to the uphill holdup in plug flow. This results in complete recovery of the elevation term in downhill flow. The majority of the data are in this flow regime, according to the Guzhov Fig, 7-Measured uphill holdup vs calculated uphill holdup using the Guzhov etal. method. TS Fig, 8~ Measured downhill holdup vs calculated downhill holdup using the Beggs and Brill method 204 ‘TABLE 5- STATISTICS FROM HOLDUP "ANALYSIS Uphill Downhill Method WE S MES BoggsandBril 92 150 610 808 Flanigan. cr re Guznoveral. 143«198 368 87.4 Bogge-Guznovy 79 487 “tal covtaion odorant ony : et al. criteria, The: acceleration term in this corre- lation was neglected. The combination of Dukler holdup, Dukler friction factor, and the Flanigan elevation factor did not predict pressure loss as well as the Eaton holdup, Measured downhill holdup vs calculated downhil holdup using the Guzhov ef al. method, m+ FI . 10 ~ Measured downhill holdup vs calculated downhill holdup using the Beggs and BrilliGuzhov ef al method. JOURNAL OF PETROLEUM TECHNOLOGY TABLE 6 ~ STATISTICS FROM PRESSURE-LOSS ANALYSIS ethos cS ‘Begoé and Gril 4953 634 5585 Boggs and Brill (rocitied by Robinson) 284 2621 028 4008 448 88.55 Beogsand Brill (coplect pressure reaver) -062 7621 398 4352 8.03 S708 Boggs and Brill ‘tough pipe Frctlontastor) 188) 2098 18.43 51.95 2558 5805, Eaton (HL, Duo (, Flanigan(@F) 1089 4274 1261 6173 1420 5985, Dukier HU, Dukier (Fr, Flanigan(@) 2247 41.79 22.19 63.28 2540 52.09, Eaton HL) Eaton (FF) Flanigan(EF) 11.98 47.79 1223 ett 1492 55.70 Gumhovers, 1198 5880 -11.86 5621 2205 «2.88 Dukler friction factor, and Flanigan elevation factor combination. This must be a direct result of the Dukler holdup correlation. This agrees with a previous evaluation of Dukler’s holdup correlation by Marcano® wheré the Dukler correlation. wi derpredicted by 33%. Underprediction of liquid holdup would result in overprediction of friction loss. ‘The combination using Eaton’s friction factor was, fairly accurate since the correlation was developed with data taken in 2-in. pipe. The original Beggs and Brill correlation and the version modified by Robinson* gave essentially the Fig. 11 Measured vs calculated pressure drop (total test section) using the Beggs and Brill correlation, SEPTEMBER 1979 Downhit Total se0.2 00.4 sec. Test Section aE We sss “jaso7 1707 6820 1082 res | 1EO58 2452 1268 “12192 17295-6885 11809-19192 10289 -2575 1445 1695 20084 4216 22119 9085 T2384 -145 1501 “9167 2626-2420 16240-9277 W74A1 408 21.79 2187 25981 5255 22459 9509 12228 5:98 1529 4056 70035 7968 25068 6297 14655 10.43 1485, 20.18 20780 46.15 2281 AIA 12595 G87 2084 17892 161.44 14008 24005-2205 288-4868 41.43 same results for these data, The Beggs and Brill correlation next is analyzed in detail. ‘Beggs and Brill Method When analyzing the holdup data, we found that the Beggs and Brill correlation overpredicted holdup in downhill flow. This would result in too much pressure recovery and a corresponding under- prediction in overall pressure loss. In view of this, the Beggs and Brill correlation was evaluated with the elevation term set at zero in downhill flow. This resulted in considerably better performance. The average percent error was reduced to ~1.45% with a Fig. 12 Measured vs calculated pressure drop (total test ‘section) using the Beggs and Bril correlation as modified by Robinson. ros 99% confidence interval of -6.49 to 3.59%. Fig. 13 shows the improved performance of the correlation, However, to conclude that the downhill holdup is the only error source in this correlation could be an oversimplification. Possibly, some error exists in the friction com- ponent of the total pressure drop. The Beggs and Brill correlation was developed in plastic pipe, and roughness was not a parameter in the friction factor. In the original correlation, the Drew e al."* equation was used to calculate a smooth-pipe friction factor ‘multiplier. This was replaced with another equation,"* which is a function of relative roughness and which resulted in an average percent error for the total test section of ~4.38M% with a 99% confidence interval of -11.28 to 2.52%, Fig. 14 shows the im- proved performance of the correlation. Table 5 gives the uphill and downhill statistics for this variation of the Beggs and Brill method. It seems that using a rough-pipe friction factor provided too much friction loss. However, the overprediction of pressure re- covery compensates for this. Note that the roughness used was that of commercial steel or wrought iron (0.00015 ft). This value resulted in good agreement Fig. 19—Measured vs calculated pressure drop (total test Section) using the Beggs and Brill correlation With no pressure recovery. Fig. 15~Measured vs calculated pressure drop (tota! test section) using the Eaton-Dukler-Flanigan ‘combination Fig. 14—Measured vs calculated pressure drop (total test ‘section) using the Beggs and Brill correlation ‘with a rough pipe fiction factor multiplier. 1206 Fig, 16 Measured vs calculated pressure drop (total test section) using DuklerDukler-Flanigan com Bination, JOURNAL OF PETROLEUM TECHNOLOGY between calculated and measured pressure drops in single-phase gas flow. ‘The Beggs and Brill correlation seems to be af- fected primarily by an overprediction of holdup in downhill flow. This results in too much pressure recovery. It is possible that the friction loss is not as ‘great as it should be, However, neglecting pressure recovery and using a rough-pipe friction factor would result in an excessive pressure loss. An in- teresting analysis would be to use the Beggs and Brill/Guzhov er al. combination discussed previously for downhill holdup with a rough-pipe friction factor, We did not do this because our paper is a combination of two independent research projects (see Refs. 8 and 9). Results of the holdup analysis ‘were not available before completion of the pressure loss analysis, Conclusions Several conclusions can be drawn from the ex- perimental data analyzed in this study. 1. The Beggs and Brill correlation accurately predicts holdup in uphill flow. 2. The Beggs and Brill correlation overpredicts liquid holdup in downhill flow. 3. Liquid holdup prediction in downhill flow can be improved by using the Guzhov et al, correlation when stratified flow is indicated and by using the Beggs and Brill correlation at all other times. 4, The combination of Eaton holdup, Dukler friction factor, and the Flanigan elevation factor accurately predicted pressure loss. 5. The Beggs and Brill correlation underpredicted pressure loss by nearly 25%. This is primarily because of an overprediction of liquid holdup in downhill flow. 6. Neglecting pressure recovery in downhill flow to Fig. 17—Measured vs calculated pressure drop (total test section) using the EatonEaton-Flanigan com bination, SEPTEMBER 1979 compensate for the overprediction of holdup with the Beggs and Brill correlation gave excellent results. Acknowledgment ‘The sponsorship of these projects by the U. of Tulsa Fluid Flow Projects is gratefully acknowledged. Nomenclature SE = percent error TE = average percent error number of data points deviation about the average percent error variate ofthe f distribution c= confidence coefficient References 1, Flanigan, O.: “fest of Uphill Flow on Pressure Drop in Design of Two-Phase Gathering Systems," Oil and Gas J. (March 10,1988) $6,132 2. Gushov, A.L., Mamayev, V.A.y and Odishariva, GE: "A ‘Study of Transportation in Gas Systems,” paper 1GV/C19-67 presented atthe 10th International Gas Union Conference, Hamburg, June 610, 1967 3. Begss, HLD. and Brill, J.P “A Study of Two-Phase Flow in Inclined Pipes," J. Pet, Tech. (May 1973) 607-619; Trans., AIME, 285. 4, Robinson, 1.R.: “Development of a Two-Phase Well Data Bank and an Evaluation Study of Pressure Loss Methods Applied to Directional Oil Wels," MS thesis, U-of Tusa, OK a7), 5. Baker, O.: ““GasLiquid Flow in Pipelines, Hl. Design Manual,” APL-AGA Project NX-28, New York City (Ox. 1970). 6. Dukier, A.B: “as-Liguid Flow in Pipelines, Part 1, Research Results," APLAGA Project NX-28, New York City (May 1969), 7. Eaton, BA “The Prediction of How Patterns, Liquid Holdup and Pressure Losses Occurring During Continuous Two-Phase Flow in Horizontal Pipelines,” PAD dissertation, U-of Texas, Austin (1966), Palmer, CM. “Evaluation of Inclined Pipe Two-Phase Fig. 18—Measured vs calculated pressure drop (total test section) using the Guzhov ef al. correlation 07 Liguid Holdup Correlations Using Experimental Data," MS thesis, U.of Tulsa, OK (1975). 9. Payne, GA.: “Experimental Evaluation of Two-Phase Pressure Loss Correlations for Inclined Pipe,” MS thesis, U of Tusa, OK (1975) 10. Beal, C: “Viscosity of Air, Water, Natural Gas, Crude Oi and ts Associated Gases at OiLFeld. Temperatures and Pressures," Trans. AIME (1946) 168, 94-115, 11, Bill .P! and Beggs, H.D.: Two-Phase Flow in Pipes, U. of ‘Tuba Press, Tulsa, OK (1975). 12, Lee, AL, Gonzales, M.H,, and Eakin B ‘of Natural Gas,” J. Pet, Tech, (Avg, 1966) 997 ‘AIME, 237 13, Mareano, “NLL: “Comparisons of Liquid Holdup Correlations for Gas Liquid Flow in Horizontal Pipes," MS thesis, U-of Tulsa, OK (973), 14, Drew, T-B., Koo, E.C., and McAdams, W.H.: “The Friction Facor for Clean Round Pipes,” Trans. AICRE (1932) 2,56. 18. J Inst, Chol Engrs, London (1938) 1,133 2 The Viseosiy 000; Trans. 1208 SI Metric Conversion Factors B/D x 1.589873 E-O1 dyne/em x 1.000" E+00 °F CE32/1.8 ft x 3.048" E-01 in, x 2.540" E-02 Ibm/euft x 1.601846 E+01 sef/D X 2:863640 -E-02 = std m?/m* 120i Paper eceatd tor blention sen 1a a7. Revenant [Sept Fa ache! Confrence and Exh, eI Care Oc 2, JOURNAL OF PETROLEUM TECHNOLOGY Discussion of Evaluation of Inclined-Pipe, Two-Phase Liquid Holdup and Pressure-Loss Correlations Using Experimental Data Ali Danesh, SPE-AIME, Abadan Inst of Technology Introduction The following comments pertain to two methods of holdup prediction discussed in “Evaluation of In- clined-Pipe, Two-Phase Liquid Holdup and Pressure-Loss Correlations Using Experimental Data” by Payne et al. (Sept. 1979 JPT, Pages 1198- 1208). Beggs and Brill Correlation Payne ef a! concluded that the Beggs and Brill correlation accurately predicts holdup in uphill flow and overpredicts holdup in downhill flow, but remains the most accurate method for all’ flow regimes with the exception of stratified flow. Use of the Beggs and Brill correlation also has been recommended by others (such as Gould and Ram- sey?) for gas-condensate offshore pipelines. The Beggs and Brill correlation was used at the Abadan Inst. of Technology’ to predict the per- formance of a 24.5-in, (622-mm) ID hilly-terrain gas- condensate pipeline 30 miles (48 km) long at 2,500- psi (17 236-kPa) inlet pressure. When the predicted holdup results were studied, many unrealistic values ‘were noticed. The method predicted negative values and values greater than one. It also_ predicted horizontal holdup may decrease in the uphill section of the pipe. Careful examination of the calculations revealed most of these incorrect results originated from use of the inclination correction factor. In the Beggs and Brill method, the two parameters of input liquid content and Froude number are sufficient to predict only the horizontal flow regime and the horizontal holdup, and they are independent of the physical properties of the fluids. But the in clination correction factor, which adjusts the horizontal holdup for the inclination, depends on the JANUARY 1980 liguid velocity number, N; y, defined by Niv=Vst (01/0), where Vsy is the superfacial liquid velocity, pz is the liquid density, and o is the interfacial tension of the two phases, The slip between the pipeline phases basically is caused by the difference in properties of the two phases. The most important physical properties are density, viscosity, and interfacial tension, When the pressure of a gas-condensate mixture increases and approaches the critical condition, the difference in the properties of the two phases is reduced and the holdup for such a gas-condensate mixture at high pressure is definitely lower than the air/water system used in the Beggs and Brill experiment at the same flow rates of gas and liquid. Hence, it can be con- cluded that the method, which does not consider the effects of physical properties, overpredicts the horizontal holdup for a high-pressure gas-condensate pipeline. For inclined flow where the effects of physical properties are considered, the inclination correction factor for horizontal segregated flow shows that the crease of (pz /0) increases the uphill holdup. While for the gas-condensate line, the increase of (pz /0), Which is the pressure increase, should reduce the holdup. For example, at the critical condition the flow becomes homogeneous and the slip is zero, while the correction factor approaches infinity. For horizontal intermittent flow, the correlation shows that the holdup decreases with the inerease of (pz /0). This further magnifies problems associated with the discontinuity that occurs from methods giving different correlations for different flow regimes. 3 Gregory* showed that the effect of the angle of inclination on holdup is small for angles less than 10°. Therefore, for uphill flow it is recommended 169 that the inclination correction factor be ignored for a high-pressure gas-condensate hilly-terrain pipeline and the horizontal holdup be used. This recom- mendation may reduce the accuracy of the holdup prediction at some conditions but ensures that large errors are not introduced in the evaluation of the holdup. It is worth mentioning that setting con- straints in the calculation procedures (such as 0=H, <1) and other constraints given in the revised method of Beggs and Brill is not a reliable solution to the problem. We discarded only the obviously erroneous results. Proper constraints should be in the form of limits on flow properties and conditions, such as physical properties, diameter, Reynolds number, ete The final comment on the Beggs and Brill method, which can be generalized to most of the available correlations for two-phase flow, is that these correlations are based on some forms of dimen- sionless groups or factors introduced by simply combining the parameters or exactly duplicating the forms used in single-phase flow. To have a reliable correlation, the dimensionless groups should be developed by rigorous analysis of the multiphase flow and a thorough study of the pertinent forces. Otherwise, these correlations should be used only within the tested range of the variables. The ap- plication of these correlations becomes further limited when parameters are employed that were not varied at all (e.g., 0, and o in the Beggs and Brill correlation), Flanigan Correlation Flanigan defined the liquid head factor By, as the fraction of the elevation pressure drop that would ‘exist if all uphill sections of the pipe were filled with the liguid. Flanigan did not measure the liquid holdup but evaluated the elevation pressure drop by subiracting the calculated friction pressure drop from the total measured pressure drop, assuming no downhill pressure recovery. Therefore, the values of Ey, evaluated by Flanigan depend on the pressure revovery in the dowakill section of the pipe and the calculated friction pressure drop, which are not Pertinent to the liquid holdup in uphill section of the pipe. Neglecting these effects, rin! manuscript SPE AT seed n Solty of Parl Engines 170 APp=Eye.e DZ* = [oct tec-Mi)Je DZ". which gives AL Ey-eG!01)/(1—9G/0L) where APe is elevation pressure drop, py, and pg are liquid and gas density, respectively, £2 is the sum of uphill rises, H, is liquid holdup, and g is gravitational acceleration. ‘The equation’ shows thet Hy, is always smaller than Ey as Ey,<1, and Ey given by Flanigan can be taken as Hy When pg/oz 18 very small, The correlation of Ey was developed using data of Flanigan as well as other investigators Flanigan’s data were taken from a high-pressure gas- condensate line, and the density ratio of the two Phases should not be ignored, especially at low holdup. The properties of the two phases are not given by Flanigan, but the density ratio may be estimated to be about 0.1, which results in the following values. Liquid head factor 0.3 0.4 0.5 0.6 0.7 Liquid holdup 0.22 0.33 0.44 0.56 0.67 Brrorpereent -35—i201B BS Flanigan used his data to develop the correlation for Ey>0.35, and for Ey<0.35 he used low- pressure gas-liquid data of others. Hence, the density correction for Ey <0.35 is not required. Fig. 6 in Ref. 1 shows that the predicted liquid holdup values by the Flanigan method are higher than the measured values for Ej, >0.35; therefore, the density correction definitely improves the results. References 1. Payne, G.A., Palmer, G.M., Bail, J.P. and Beges, HD. “valuation of inclined-Pipe, Two-Phase Liquid Holdup and Pressure-Loss Correlations Using Experimental Daa,” J. Pt Tech. (Sept. 1979) 1198-1208; Trans. AIME, 267. Gould, T-L. and Ramsey, EL “Design'of Offshore Cas Pipelines Accounting for Two-Phase Flow,” J. Pet Tech (March 1975) 366-374; Trans, AIME, 289, 3. Danesh, A. and Noghtekar, G.R': "Hilly Country Gas Condensate Pipelines,” Oil Service Co, of Iran Report No. 30621976) 4, Grepory, G.A.: “Comparison of Methods for the Prediction of| Liguid” Hoidyp for Upward. Gas Liquid. How in Inclined Pipes," Cam. Cherm Bn. (1975) 384388, eT JOURNAL OF PETROLEUM TECHNOLOGY

Вам также может понравиться