You are on page 1of 3
PIPING/FLUID FLOW An improved experimental correlation for Darcy friction factor The pressure drop estimate may be on the unsafe side when using the traditional Moody chart G. 0. CORDERG, Techint Engineering & Construction, Buenos Aires, Argentina ‘cent research has shown that the traditional formulas smooth pipes by as much as 2.796 at intermediate Reyn- colds numbers and up to 2.9% at the highest Reynolds numbers measured. Deviations reaching ~2.8% are also expected in rough pipes. This small underestimace should be acceptable for most applica- tions, but engineering companies would certainly prefer a con- servative approach for al eitical lines when the contract includes hydraulic guarantees Moody's chart. Moody published his chart for finding Darcy friction factor as a function of Reynolds number and relative roughness in 1944." Through computerized approximations, this chart is tll the most widely used source in pipe low design. [tis based on the following formulas: + For laminar flow (Re < 2,000): Hagen (1839)-Poiseuille (1840) + For turbulent flow (Re> 4,000): Colebrook-White (1939)? (Ciencia cones Padi aed pipe lwand Ahad sgh pps lm providing conserve amar fot the transition: Smooth pipe lv: rand (1938) 1 251 | Fh = aog( te/7)-0.8 = -2log| 254 | vf (ed) nd so aig ton ealig-Iogaril eo profile with the friction factor and deriving the constants from Nikuradse experiments, Rough pipe ve Nikuradse (1933) 1 {Dp {_& \ ee? sy) +1 74—-2 — Feel ae) 21 7A= ale Sr Where isthe artificial sand roughness applied by Nikuads on the inner wall ofthe pipes used for his measurements. * For transition flow (2,000 < Re-< 4,000): Moody's chart pro- vides a shaded area, showing that laminar flow may extend beyond Re 2,00 and wurbulent How may devlop before aching Re = 4,000. Experimental results show a wide dispersion in this zone. Research in Princeton and Oregon universities. Z1- pa 425) rola’s research (1996) with the SuperPipe facility in Princeton Uni- ip "*\370 ‘ersty reached Reynolds numbers rom 31,300 up to 35,500,000 V Ref 9 ) P (beyond the maximum 3,400,000 covered by Nikuradse), show- oor 4 — ox isi on smooth 02a] ono toy 2=2828in, 5 [Wey =i, ke amiron Foon | __yeaborsin "| Fae re Pe eet 0.008. 0.020) bis “000,000 6 000,000 17,00q000 16,000000 21,000,000, Reynolds Cn res 10,000 20,000. 30,000 40000 $0,000 60,000 70,000 Reynolds of refiner rr HYDROCARBON PROCESSING suey 2008 | 97 PIPING/FLUID FLOW ko= itt Re ace Ae (0008 002 iaaee 1,990 10,000 100,600. 1,000,000 10,000,000 100,000,000, Reynolds number Function CBNIN(, 6, Re) "Dare ction factor combining Colebrook, Blsus, McKeon eta and Nikuradse "c= roughness, d= ise diameter, Re = Reynolds, kand din the same units Dim Rau, Fo, Ft, "Raux=ouxliny Reynolds, Ft = transmision factor = fDarey* 05 “Fto= previous guest for Ft, Fk= relative roughness factor = 3.7") If Re e= 2000 Then “Fr laminar flow (Reynolds <= 2000) Hagen Poisculle formula BUN = 64 Re. che HH Re <= 65964 Then "Blasus for Re<=B6964 IF Re < 3080 Then “for anstion flow (2000 = 3080) Raux= Re indit CBM =0.3164/Raux 025, Itk>0Then Jebrook White to account fr roughness contribution “Log = natural logarithm, Log'0 = common logarithm = Log? 302585083 (CBM = (2 * Log 7 *) + 10*(-1/(2* SplCBMIN)) 2302585083) ®-2 fndif Re-<3080 Then “near iterpolation for transition flow (2000 65964 ‘elatveroughnes factor Fk =k/G7*0) “nial value for transmission factor e712 Do Fo=Ft “Colebrook: White to account for roughness contribution Fra -2* LoglPk + 1855667 *(Fto/ Re) “0:965)/2.302585003 Loop Until Abs(to Ft 1) < 0.0000001 CBMN= FM? Endif nd if nd Function ing a clear trend fora frition factor slightly higher than predicted by Prandil’s smooth pipe law. Another team at Oregon University, working with a com: pletely different type of facility, investigated smooth pipe flow for Reynolds numbers from 11 up to 1,050,000. Both teams? published the comparison between their experi ‘mental results and the forrmula with new constants in Prandt!s law proposed by the Princeton group" to fi the SuperPipe measure ‘ments for Re> 31,300: McKeon, Zagarola and Smits: 1 $e tosita iF -0597 ‘The above formula is within £0.5% of Princeton's pub- lished results for Reynolds numbers between 400,000 and 24,000,000 and beeween ~1.78% and +0.84% for the whole Princeton set After refining the initial measurements of Zagarola, the Princeton group claims an experimental error of #1.19, The Oregon group acknowledges +2% to 4%. The region where both researches overlap shows differences within the exper mental error. More complex velocity profiles, including corrections for the region near the wall, showed no improvement compared ‘with the previous formula in approaching the experimental results! Another interesting conclusion from the same refer- ‘ence is that the exponential law represents velocity profile better than the logarithmic law for Reynolds numbers below approximately 98,000. Blasius’s law (related to the exponential velocity profile) is considered more accurate than Prandt!’s law in that region. Blasius (1911) 0.3164 "pees From a practical point of view, itis best co apply Blasiuss law up 10 Re = 66,964 where it coincides with McKeon’ thus providing a continuous friction factor. In that range, Blasius is between ~0.36% and +0.62% of Princeton's results. For all Reynolds numbers in the turbulent region below 66,964, Blasius remains between ~1.7% and +1.4% ‘of Oregon's results, well within the experimental error The only exception is an Oregon test point 6% below the Princeton points. In the transition regime, Oregon tests show decreasing n factor up to Re = 2,868 (f= 0.02804). Then the fric tion factor increases rapidly up 10 0.04285 at Re = 3,080, A review of Moody's chart for smooth pipe indicates shat ‘In the turbulene region, fiom Re = 4,300 up 10 Re = 66,964, Colebrook remains below Blasius with a maximum deviation of -2.7%. From Re = 66,964, Colebrook remains always below McKeor's formula with a maximum deviation of -2.9% at Re = 35,500,000. A review of Moody's chart for typical cases of rough pipe shows that: * For a 35.25-in. ID epoxy-lined pipe with 4-micron 28 | sur 2008 iocanson Roc pipelines), the Moody factor remains below the Princeton smooth pipe up to Re = 3,100,000. If Colebrook’s procedure is applied to McKeon’s formula to account for roughness, Moody falls to -2% at Re = 5,000,000 (Fig, 1). + Fora 10 in-ID bare sted pipe with 0.0018-in, sand rough- ness (typical of refinery piping), the Moody factor is 1.6% below Blasius smooth pipe at Re = 13,600 and 2.8% below Blasius corrected for roughness at Re = 17,000 (Fig. 2). ‘As a conclusion, Moody’ chart may underestimate pressure drop by up to almost 3%, both for smooth and commercial rough pipes. This margin should be acceptable for most indus- trial applications, but critical cases require a more accurate approach. A practical way to update Moody's chart. The recent experiments cited suggest the use of both Blasius’s and Prandtl’s laws with McKeon constants for representing (ur bulent smooth pipe flow up to the highest Reynolds number measured (35,500,000). McKeon’s formula for Re < 31,300 falls below Moody's friction factor and needs to be replaced by Blasius’s formula, which gives the est approach to experi- ‘mental results. For rough pipe, if Nikuradse’s equivalent sand rough- ness and Colebrook’s procedure for estimating the transition between smooth and rough regimes are to be kept, the follow- ing formulas result: For turbulent flow (Re> 3,080 to match the Oregon tests): + Re 66,964: ‘Smooth pipe: 03164 (os) 316 Ds atog 10-25 | wes Vp Rough pipe: Mca) Trae hs) + Re> 66,964 Smooth pip: 1.85567 1 pe = 1.930log| Re, |= 0.537 = =2k if og( keV 7) | Rough pipe: 1.85567 nap The basic program procedure fora spreadsheet in Fig, 4 easily makes available a continuous function for obtaining Darcy fric- tion factor for any equivalent sand roughness, inside diameter and Reynolds number, through the combined use of Colebrook, Blasius, McKeon and Nikuradse formulas: f= CBMN (,D,Re) PIPING/FLUID FLOW ‘This function covers laminar flow through the Hagen-Pe seule formula up to Re= 2,000 and transition flow through linear interpolation between the laminar friction factor at Re = 2,000 and the turbulene friction factor at Re = 3,080. 4 LITERATURE CITED * Moody, LF, “Fretion factors fo pipe flow” Thaw ASME 671, 1944 2 Colebrook, C. Ry “Turbulent low in pipes with particular reference 10 the transition region berween the smooth and the wugh pipe laws.” Jetson Ghul Engineer, 1939. ® McKeon, B. J, C.J. Swanson, M.V. Zagarola, RJ. Donnelly and A.J Smits “Breton factor for smooth pipe om”. Haid Mech, Val. 511, pp. I-44 2004. “ MeKeon, BJs MV. Zagato and AJ Smits,“ new ction ite ratoahip oe filly developed pipe ow J Fluid Mec, Vol. 538, pp. 29-443, 205. NOMENCLATURE CBMN’ Colebrook Blais McKeon NikuraseFancion for € ‘Inside pipe diameter Ff, Darcy fron fitoe 1 Equivalent sand roughness of pipe wall Re Reynolds umber Guillermo Cordero tis been with lechint Engineering & CConsructon for 38 years t Buenos Aes headquarters, where he Is the head ofthe Poces Deparment. is work covers basic and etaled engineering for pipelines, refineries and petrochemical ‘nis. Hei @ chemical engineer from the Univesity of Buenos ‘es, where he has bee a profesor of Unit Operations Select 172 at www HydrocarbonProcessing,com/RS °

You might also like