10 FLOW OF FLUIDS the flow of any fluid through a line containing no heat input or output may be considered isother- mal and adiabatic. One may therefore combine the First and Second Laws of Thermodynamics, using these assumptions to write an equation. 2 fvar+£ax+ 20) Ww, -W (10.1) Be Be : Where: V = volume of the fluid P = pressure of the fluid AX = change in elevation of the fluid ‘A(v?) = change in velocity squared of the fluid W; = work lost due to friction W = work done by the system g = gravitational force &_ = mass-force conversion constant Equation 10.1 is as far as one can go using thermodynamics alone, for it provides no way of solving for the term Wy, lost work. The total irreversible effects (W,) are usually attributed to friction. The work done in over- coming friction through a distance dL is proportional to the surface in contact with the fluid, approxi- mately proportional to the square of velocity and proportional to the fluid density. By including a proportionality constant, f, the equation may be written. 2 frictional resistance = coratna x or (10.2) Be where the definition of variables for Equations 10.2-10.4 may be found on page 321. The mass of fluid in the pipe is the length dL multiplied by the cross-sectional area and the fluid density. Any frictional work would be represented by the frictional resistance moved through dis- tance dL. Combining these concepts into Equation 10,2 then yields dW: CHAPTER 10 319FLOW OF FLUIDS corarna 5° Jona) dW, = (103) (Jou Saba 2fv'dL aw, 2 (104) which integrated between the limits of 0 to W, and 0 to L gives W; = (10.5) Equation 10.5 is called the Fanning friction factor equation. Other forms of this equation have been published which differ only in the value of the coefficient ahead of “f.” One of these is referred to as the Moody friction factor where the “2” is in the denominator rather than the numerator. The Moody friction factor is 4 times greater than Fanning. In this book the Fanning equation will be used in all correlations. Combination of Equations 10.1 and 10.5 then gives Bo?) __ dev? 28. Bed var +2 ax+ (10.6) Be which is the basic equation for flow often called the Bernoulli equation. If compressors and pumps are omitted from the system, no work is done in a line segment. So, W is eliminated from an equation like 10.6 to solve for the pressure drop ~ flow rate behavior, giving Equation 10.7: (10.7) 2 var O29, 20). mzitse 286 Bed EVALUATION OF FRICTION FACTOR The friction factor “f” must be evaluated empirically. One common correlation is the Moody plot shown in Figure 10.1.7) Factor “f? depends on pipe roughness and diameter, and the fluid involved. The dimensionless Reynolds number on the abscissa of Figure 10.1 may be written in various forms. 4 4 poe ap el (10.8) no xd dp 320 VOLUME 1: THE BASIC PRINCIPLESEVALUATION OF FRICTION FACTOR Values of (vd) for Water at 15.6°C [60°F] (velocity in ft/sec x diameter in inches) iQ a2 9808 1 gee to a om ao tam ant son oe - ToS Tare ooo pee tere LET Nimes TTT | ee oe | aa er rs | + it PE [ iP os i I i jaee LT os 0.06 Ht i | cot roe 7 & tn ae sle | CT Pie | 00s i z EE Core ale Tor o.c08 — il TCC one: pom | Se ALi eeu LUE RNET Th é | | I a: 4 | te ae WHEE | 30098 | oot é y He - - z | HS aaa 8 | | i ptt oie TNE LT TSdpowe| cos) ete rerumegreamoreaors (01 HNL | LLU LISS rears ‘obtained by dividing the values fo] fT TTE|E | LETTE | EE 0,000 008 | aud FH UTES meee ar aC CT CT CT RT vp a Figure 10.1 Friction Factors Using Moody Correlation”? (which combines the Colebrook transition between the smooth pipe equation and the fully turbulent zone) ‘Any units that make Re dimensionless may be used. For fluid flow applications the following units are most common. Re, Reynolds Number = : aH FPS Where: 4 = internal line diameter m ft v = velocity ms fuse p = density kg/m? Ibnvf?® = viscosity kg/ms Ibnvft-sec P = pressure Pa Ibf/f? B_ = mass/force conversion constant so kg =) aan Tbm-ft } Ns? Ibf -sec? & = gravitional force 9.81 m/s? 32.17 fusec? L = length of line m a AX = change of elevation m ft m = mass flow rate kgs Ibmvsec q = volumetric flow rate miss fisee f = dimensionless factor Lep = 0.001 kg/ms = 6.72 x 104 Ibm/ft-sec CHAPTER 10 321FLOW OF FLUIDS This type of correlation applies only to Newtonian fluids. Gas and almost all liquids of con- cem in this book are Newtonian in nature. Non-Newtonian Liquids One often assumes (erroneously) that the liquids being used are Newtonian fluids. In fact, many fluids do not truly fall within this category. There are several general classifications of liquids based on their rheological behavior: Newtonian, Bingham Plastic, Pseudoplastic and Dilatant. The general curves for these are shown in Figure 10.2. Of these, the Bingham plastic and Newtonian liquids are the most common models employed. For a Newtonian liquid, the relationship is expressed by the equation ro (Se) (10.9) g.\ dr Where: ~ (dvdr) absolute viscosity shear rate for laminar flow in a circular pipe 7 Rate of Shear ‘As shown in Figure 10.2, this equation pro- Figure 102 Shear Gurves of Newtonian and 4uCeS a straight line on cartesian coordinates starting Non-Newtonian Fluids at the origin, The slope of the line is w/g. ‘A Bingham plastic, on the other hand, is a straight line that does not pass through the origin. It is represented by the equation d tty tsa) (10.10) Be dr Where: «ty = the intercept hip = plastic viscosity, a value found from the line slope A Bingham plastic will not flow until the shear stress exceeds a value represented by ty, known as the yield point. Drilling fluids consisting of colloidal clay particles suspended in liquid ap- proximate Bingham plastic behavior. Pseudoplastic and dilatant fluids possess no yield point and exhibit nonlinear behavior. In ad- dition, there are fluids where behavior is time dependent. A thixotropic fluid will produce a double curve. When increasing shear rates are imposed it will break down with time. If shear rate is then re- duced no further, breakdown occurs. After exposure to high shear rates, this type of liquid will regain its original consistency after a period of time. This property is useful in special applications. A rheopectic liquid is one where at constant shear rate the shear stress increases with time, the opposite of a thixotropic liquid. 322 VOLUME 1: THE BASIC PRINCIPLESNEWTONIAN LIQUID FLOW It is common to characterize non-Newtonian liquids by an empirical power law formula (0.11) Where: K = an overall measure of the liquid viciousness W/ge for Newtonian fluid n = 1.0 for Newtonian fluid = 0.0-1.0 for pseudoplastic fluids * = 1.0 for dilatant fluids Equation 10.11 is a straight line on a log-log plot. The deviation of “n” from 1.0 (in either di- rection) is a measure of the degree of non-Newtonian behavior. The detailed calculation of the behavior of non-Newtonian liquids is beyond the scope of this book, but more details are available in References 10.3 and 10.4, and in many references covering rhe- ology of liquids. Most oils may be treated as Newtonian liquids, particularly those with a relative density below 0.9. However, waxy crude oils and condensates as well as some emulsions of oil and water may ex- hibit non-Newtonian characteristics, often pseudoplastic. NEWTONIAN LIQUID FLOW 2 The Moody chart (Figure 10.1) is a sound correlation if one can estimate roughness. Unfortu- nately, one never knows what it is'at any one time, and it changes with time. Initially the pipe may contain mill scale from its manufacture. After use this mill scale may be removed by liquid action and the pipe wall becomes smoother. On continued use erosion/corrosion, scale formation and the like may increase roughness. Figure 10.3 is based on field tests. The smooth curve is for 8 inch pipe or larger, or smooth tubing. As diameter increases, roughness has less effect. T - ll | “7 @=ppe danete, Ft ee linear velocity, f’sec Bt fee oe § 3 E 12 eo 7 fa fy ro Paes 7107 Reynolds Number, Re = “ Figure 10.3 Simplified Correlation for Fanning Friction Factor CHAPTER 10 323FLOW OF FLUIDS The equations of the three lines in Figure 10.3 are represented closely by the following equa- tions. Re<2000, f = 16/Re (10.12) Re>4000, f = 0,042/Re”!%* large pipe > 20 cm [8 in} (10.13) Re>4000, f = 0,042/Re®!7? small pipe < 20 cm [8 in} (10.14) Figures 10.1 and 10.3 have served for years as a basis in evaluation of friction factors. In the world of computers, equations are more convenient. Reference 10.5 provides an excellent comparison of several friction factor correlations. Churchill and Usagi have proposed a single equation to replace the f vs. Re plots shown. 7 oo r=2/[2]) +1. (10.15) Re (Cy +C2) fi I a ;o Tyna c= (222 | For the flow of an incompressible liquid in a section of line of fixed diameter where no work is done, Equation 10.7 may be written Where: 2fLv? AP =(P, -)-(285 *)-vaxcae) (10.16) = (-AP;)-pAX(e/g-) It often is more convenient to express flow rate in volumetric or mass terms rather than veloc- ity. The v? in Equation 10.16 may be substituted for as follows: 7 7 (10.17) As with Reynolds number, the substitution of these terms in the Fanning friction factor equa- tion to obtain AP; enables one to write Equation 10.16 in terms of volumetric or mass flow. 324 VOLUME 1; THE BASIC PRINCIPLESNEWTONIAN LIQUID FLOW Example 10.1: A 0.254 m [10 in] crude oil line handles 6400 m? [40 260 bbl] per day of a 0.83 Rel. p oil. The oil viscosity is 9.5 cp. The flow line is 10 km [6.15 mi] long and discharges to a 100 kPa [14.5 psi] storage tank. The tank battery is 100 m [328 ft] higher than the pump discharge. Determine YookPa [145psia] the necessary pump discharge pressure. The pipewall roughness is assumed to be 38 um [1500 pin). No attempt is made here to calculate losses caused by pipe fittings. Such losses may be rele- vant and in real world situations should be evalu- ated. Normally they are added as an equivalent length to the actual length of the pipeline. For this example (10 km pipeline) they will be neglible. Sk: d = 0.254m 6400 _ ~ Gas aoy(orssy(oasay“° 8 1 =(9.5)(0.001)=0.0095 ke/ms p=(0.83)(1000)= 830 kg/m? te = (0.254)(1.46)(830)_ _ 5 9 igs 6 £ _ 38x10" _ o.00015 ad 0.254 From Figure 10.1, f = 0.023 (Moody) = 0.0058 (Fanning) 2 Pr = 2)(0.0058)(10000)(1.46)" (830) _ 8.08 x10° N/m? =8.08x10°Pa g (1.0)(0.254) (p)(AX)(g/g.) = (830)(100)(9.81/1.0) = 814230 N/m? P, -P, = -AP; ~(/8.)(p)(AX) (100000 -P, )= -814000-808000 P, =1.72x10°Pa =1.72MPa FPS: d = 10in=0.8338 __(40260)(5.61) (86 400)(0.785)(0.833)° 4.79 ft/sec = (9.5(6.72x10" *)=0.00638 Ibm/ft-sec CHAPTER 10 325FLOW OF FLUIDS. Example 10.1 (Cont'd): p=(0.83)(62.4)= 51.8 Ibm/ft? From Figure 10.1, f = 0.023 (Moody) = 0.0058 (Fanning) (2)(0.0058)(6.15)(5280)(4.79)° (51.8) arr (32.2)(0.833) 16690 Ibf/ft? (pXAX)=(51.8)(328)= 16990 Ibf/f? (2088 —P, ) = -16990-16690 35678 Ibf/ft 248 psia In Example 10.1 AX is positive since the liquid is going uphill. If the liquid had been flowing downhill, AX would have been a negative number. Also, in this example no effort was made to spec- ify exact pipe ID, the nominal diameter was used. In reality, pipe ID varies from the values shown in tables, since some pipe specifications allow as much as 12.5% deviation in pipe wall thickness. As a practical matter, the use of exact diameters offers little increased precision in predicting line perfor- mance. Calculation of Pipe Diameter Situations often occur where it is desired to calculate the line diameter (d) needed for a given flow rate and pressure drop. The friction factor, f, and Reynolds number, Re, are a function of diame- ter and velocity. Therefore, the use of the correlation of f plotted against Re involves a trial-and-error solution if diameter is desired. A direct solution for diameter is possible by algebraic manipulation of the basic equations. Solving for “d” yields 02 d=1.265q° let. | (10.18) The friction factor in Equation 10.18 can be found by many methods. It is convenient to use Equations 10.13 and 10.14. Substituting these equations for “f” into Equation 10.18 gives 0207 L For small pipe, 4=0.647(q"” \(p° YW) ty | (10.19) 0208 L For large pipe, 4=0.647(q°?"" )(p™"* )(u°™" Nese 1 (10.20) where units are defined as those listed on page 321. 326 VOLUME 1: THE BASIC PRINCIPLESNEWTONIAN LIQUID FLOW Example 10.2: Determine the diameter for an oil flow rat e of 0.0416 m’/s [1.46 f'/sec], relative density of 0.79, a total pressure drop allowance of 500 kPa [72.5 psi] over 20 km [12.4 mi], and a viscosity of 10 cp. Assume large pipe (d 2 10 inches): 0.208 Sk: d= 0.6474 (0.0416)°37° [0.79(1000)]°" [10(0.001)]°™" Fe | = 0.255m 0208 FPS: d= 0.6474 (1.46)°3" [0.79(62.4)] *"* [10(0.000672)] pos eee Calculation of Line Capacity The calculation of the capacity of a line with a given pressure drop is trial-and-error when us- ing a plot of friction factor vs. Re. However, a direct solution is possible. Starting with the same basic equations as in the caiculation of the diameter, eliminating q instead of d, and using Equations 10.13 and 10.14 for “f,” . 3.12742 [(aP,)g. 1" For small pipe, a= Fy Gy aa tion 3.18042! [(AP,)e. |" For large pipe, q o 6 (57 ae (10.22) Calculation of Pressure Drop ‘A similar derivation approach for thg remaining variable normally required in field problems, APs, yields the following equations for normal pipe friction. io qi 90828 |, 0172 SmallPipe, = ——t. = Set (10.23) . a L 8.038g, d*#8 ” : AP, __q!6 p 0806 0198 Large Pipe, t= Ser ee (10.24 eae L 8.081g, 448 y Example 10.3: Determine the pressure drop for the flow conditions listed in Example 10.2. 0.0416 mis [1.46 ft/sec] B p= 790 kg/m* [49.3 Ibm/ft*] L d= 0.254 m [0.833 ft] 7 _ (0.0416)! #6 (790)°**[10(0.001)}"™ (20000) _ 7 8.081(1.0)(0.254)**°° Op 10 km [12.42 mi] SI: 510000Pa 1.46)"*5 (49,3)°**[10 (0.000672)]*** (12.42)(5280) FPS: ap, =f yi 493)" [o(@ a (02.42)(5280) _ 19 500 psf = 73 psi 8.081 (32.17)(0.833) CHAPTER 10 27FLOW OF FLUIDS Valves and Fittings Equations 10.19-10.24 adequately describe flow of Newtonian fluids in straight runs of com- mercial steel pipe under conditions of fully turbulent flow. The pressure drop through valves and fit tings must be addressed separately. ‘The most common method used to account for this additional pressure drop is the use of equivalent lengths (L,). Each fitting (e.g. valve, tee, el, etc.) is assigned an Lz based on its size and geometry. These are documented extensively in fluid flow literature. Refer- ence 10.2 provides an excellent review of the theory of this approach and Reference 10.6 tabulates L. values for the most common valves and fittings encountered in the hydrocarbon industry. Figure 10.4 is a handy nomograph for estimating L, for various types of fittings. Once the L, of a fitting has been determined it is simply added to the length of the straight pipe sections. ‘The pressure drop calculation is then performed on the total pipe length including the L, additions. In the planning mode, how does one estimate the pressure drop through a pipe section before the actual pipe layout has been determined? References 10.7 and 10.8 detail a method whereby pre sure drops can be estimated before detailed drawings have been completed. This is useful for predict- ing pumping and compression requirements. ‘The correlation accounts for a typical number of pipe fittings and valves installed in a piping system. L,/L=1+(0.347d"? +0.216) Fe (10.25) Where: L/L = Total equivalent lengths per length of straight pipe . d= Nominal pipe diameter, in F, = Complexity factor (see table below) Ln Very complex piping manifolds | 4 Manifold type piping * 2 Normal piping 1 Long reasonably straight runs 0.5 Utility supply lines, outside battery limits _| _ 0.25 Calculate the effective length of piping in the system shown below. Pipe size is 0.254 m [10 in]. 90° Std. Elbow (FD = 1) oS pe i m amor ee Std. Tee From Figure 10.4 10” Fully Open Gate Valve L, = 1.8m [6 fi] 10" 90° Std. elbow L, = 82m (27 ft] 10" tee (branch flow) L, = 18.0 m [60 ft] For AP calculation, the effective length of the system 50.0 m [197 ft] 328 VOLUME 1: THE BASIC PRINCIPLESNEWTONIAN LIQUID FLOW Globe Valve, Open mm + ao 42 —1 1000 n . oo gl Valve, Open 2 wo C9 24-1 600 2 20-1 500 0 tet 400 ‘Swing Check Valve, & | Pally Open é ae (o-\ £ 0 — 5 | 3 77 3 8 baw = (tose Rotum Bond Sus Enagement | & eaeeeoe ect 4 soeiaerens aa fF) wD- 12 2 e dD - 3/4 54 svar Teo a ‘Through Side fm. ° i ow be § =) ‘~ Ordinary Entrance: arte os 212 —L eo ore | fe sudden Contraction tmp oe aim 02 14-1 59 Medium Sweep Etbow or 40-34 lun of Tee Reduced 1/ E 1 : O19 wf? eee + Rh ERE IR ee Long Swoep Elbow or a un of Standard Tee Figure 10.4 Resistance of Valves and Fittings to Flow of Fiuids Economic Pipe Diameter Equations 10.19-10.24 provide the relationship between flow rate, pressure drop and pipe size. They do not provide line sizing criteria. In sizing a line one is always faced with a compromise of two factors. For a given flow rate of a given fluid, as pipe diameter is increased piping cost increases. But, pressure loss decreases, which reduces potential pumping or compression costs. The economic diameter will be the one which makes the sum of amortized capital cost plus op- erating cost a minimum. This total cost can be per unit time or per unit of production. One can corre- late this total cost (per unit time or production) versus diameter to determine a minimum. CHAPTER 10FLOW OF FLUIDS Equation 10.26 may be used to estimate this economic diameter. It is derived from cost data and must be considered an approximation only. 0.45 A)(m ae Yay" : (10.26) e ee Where d= economic diameter m in m = mass flow rate, in thousands of kghh Ibmvbr p = fluid density kg/m} Ibmift? A. = constant 0.189 22 Equation 10.26 is derived primarily for relatively short lengths of piping in processing and pro- duction installations where pressure drop is not particularly large. It is applicable to both vapor and liquid lines. Remember also that the most economic diameter may not be the one you should choose. There are mechanical considerations which are of concern. Many companies provide pipe sizing quidelines in terms of velocity or pressure drop per unit length. The following guidelines have proven useful for preliminary sizing of process piping. ___ Liquid Lines ‘Noncorrosive liquid hydrocarbon flow Water flow Corrosive liquid flow (sour condensate, glycol, amine) Centrifugal pump suction | Reciprocating (plunger) pump suction. | typiem apt ‘kpa/io0m | _psi/100 ft 0.33 to 0.66% 0.1 to 0.2 % of total pressure | of total pressure Natural gas ‘Vapor Lines _ Maximum Velocity - Some company specifications limit maximum continuous liquid velocity to vo «& YA ) (10.27) oe _FPS____ Where. v = velocity ms fuse p = density kg/m? Ibmv/ft? ‘A = conversion constant 1.23 1.0 kK 100 100 API 14E recommends three values of K; 100 for normal design calculation, 125 for intermit- tent service, 150 for the “never exceed” value. These values are, in general, based on erosion/corro- sion limitations in bends, elbows, and fittings. Metal loss is accelerated by high fluid velocities, presence of sand and corrosive components in the fluid. In solids-free, non-corrosive systems the VOLUME 1: THE BASIC PRINCIPLES.DESIGNING LOOP SYSTEMS values for K, quoted above, are conservative. Values up to 250 have been used successfully for inter- mittent service. While Equation 10.27 possesses the advantage of computational simplicity, it does not account for factors such as flow geometry, type of metal, sand size, Reynolds number, and acceptable corrosion rate. Reference 10.9 puts forward several alternative methods of estimating erosional velocities incor- porating sand particle size and fluid properties. In high sand production environments the erosional velocity for methane can be significantly less than for liquids. DESIGNING LOOP SYSTEMS ‘The need often arises for increasing the flow rate per unit pressure drop in a system. This is normally encountered where it is desired to increase the capacity of an existing system at a given pres- sure drop or decrease the’ pressure drop at a given flow. The former is a common need in expanding systems, while the latter usually stems from older systems wherein the pipe or pumping equipment may have to be derated for pressure. The usual and most economical solution to the problem is to place one or more lines parallel to the original, either partially or throughout the entire length. The figure at right shows a sche- matic view of a simple loop system that may be used to illustrate the principles involved. The loop and the line it parallels are usually equal in length throughout their common length, although conditions sometimes dictate a loop of different length. Loop Capacity It is often desirable to calculate the capacity of an existing looped system. The most straight- forward method of analyzing this problem is to convert the loop into a single line having the length of the looped section with a diameter (d,) which gives the line the same capacity as the looped section. ‘An alternative approach is to fix the diameter of the line and calculate the equivalent length L, of the line so that the line capacity is equal to that of the loop. This method can be derived from Equation 10.21 or 10.22. Equation 10.22 will be used as an example: 3.18043" da “e457 TL A similar equation can be written for gg. Now envision a line with a diameter d, and length Le which has the same capacity as the looped section (qq + da). — "4 (10.28) 3.18042! qat+dp= (9) (uy CHAPTER 10 331FLOW OF FLUIDS Remember that (AP), = (AP))p = (AP))c. Substituting Equation 10.22 into 10.28 for qa and qs. and canceling like terms gives 2661 q36st g3eel Lost = [sss * [oss (10.29) By an identical approach, for small pipe, 2641 gol gos d de A de (10.30) Coke Ae CBee Lose7 Laser * pasar Equation 10.29 can be used to convert complex looped systems into simple systems for capac- ity calculations. Calculate the capacity of the following system handling a 32°API gravity crude oil with a viscosity of 3 cp. Allowable pressure drop is 1000 kPa [145 psi]. Example 10. 10" fe ohm —-ele— 10m ——e}— 25km —>} Convert the first looped section into an equivalent length, L¢, of 14 in. pipe ~ 142651192661 472461 [ose = gusset pose + solveforL $.3km Similarly for the second looped section — 2.661 * 1 2.661 2.661 4 uo e solve for L. 22.8km, posse — 90st + 790s? Total effective length of system = 5.3 + 22.8 + 25 = 53.1 km [33.0 mi] Calculate q from Equation 10.22 - y = 141.5/(131.5 + 32) = 0.865 p = 865 kg/m? [54 Ibm/f] 1 = 0.003 kg/ms [2.02 x 10° Ibm/ft-sec] AP/L = (1.000 000)((53 100) = 18.83 Pa/m [633 (Ibf/f?)/ft] d= 0.356 m [1.17 ft] 3.180)(0.356)* a= we ae jor (18.83) = 0.094 m?/s [51 100 bpd] 332 VOLUME 1: THE BASIC PRINCIPLES.DESIGNING LOOP SYSTEMS Example 10.6: It is desired to increase the capacity of an existing pipeline system by 60%. The diameter of the existing line is 12 inches. What percentage of the pipeline system must be looped if the diameter of the loop is 14 inches? _1-(1.0/1.6) 1.806 X=— ps6 1) Tyner 142661 706 = 70.6% Loop Length Another common application of the loop calculation is to estimate the length of loop required to increase the capacity of the system a given amount. This calculation can be made by establishing a pressure balance for the system. ‘AP, (- *) en] (L-Lg) = (APP oat (1031) L A L Cc If we define X = L/L (fraction of the system looped) we can write the equation AP, APr\ yy _ ( APr_ (4 } co+(4 | x) ( 7 ). (10.32) Equations 10.23 and 10.24 can be used to calculate (APy/L), and (AP}/L)c. The right hand term (AP/L)a is fixed by allowable pressure drop. Equation 10.32 can be solved for X. Direct substitution of Equations 10.23 and 10.24 into Equation 10.32 yields the equations shown in Table 10.1. TABLE 10.1 Equations for Lines in Parallel Lines in Parallel - Small Pipe Equivalent ome One nee az! age ag Diameter Loss ~ posse * passe Lose? ~ 1950 * Tose 1.806 1.828 1-(a/a1) 1-(a/a1) ; Looping quo 08 qo pe Requirements* 1 a a Ae +p! 3 4a3 2.61 Entire Line aw |), (de a] 4 Looped* q dy 4 4 Diameter of a\™ : Original and x=1.40|1-|4 X= 1.392| 1-| Parallel Lines | a the Samet | Assumes that the length of all lines in the loop are the same length. (see following page CHAPTER 10 333FLOW OF FLUIDS TABLE 10.1 (Cont'd.) Equations for Lines in Parallel For Lines in Parallel: . Where: diameter of a single line that is equivalent to a group of parallel lines diameter of the original line before looping diameter of a single loop line (or equivalent diameter of a group of loop lines) length of equivalent single line corresponding to d. length of loop line dy Jength of loop line dg original flow rate before line is looped flow rate after looping line fraction of length of original line that is looped Complex Liquid Gathering Systems The situation often arises when it is convenient to use several diameters of pipe in series. Fluid principles also may be applied in this case after proper analysis of the conditions present. In the sketch at left, L and d represent the length and di- ameter, respectively, of two different size lines joined in series. Therefore: . 9A =4B =A and AP py + APra = AP rica However, no one equation may be written between Ly points (1) and (2) because of the varying diameter. It should be | possible to write an equation for section 1-3 and one for 3-2, and solve each for qi. P3 is found when q., “4.y- ‘A more flexible and convenient solution is afforded by determining a single-diameter line that is equivalent to the system shown. An equation in the form of 10.23 or 10.24 may be written for one section, say section A. Then the equation is written for a hypothetical section having the same diame- ter as section A but of such length that the APy is the same as section B. Then: AP; = Ot Pp =C| = (10.33) ax (#) where L, is defined as the length of line of diameter da that will give the same friction loss as length Lg of line dy. C is defined as all the terms of Equation 10.23 or 10.24 not related to length or diame- ter. If the physical properties of the fluid remain essentially unchanged throughout the line the C terms are equal. If the constants are canceled and Equation 10.33 rearranged: Sn Li-ta($9) (10.34) where n is the exponent in Reynolds number when the friction factor is expressed in the form of Equa- tion 10.13 or 10.14. 334 VOLUME 1: THE BASIC PRINCIPLESDESIGNING LOOP SYSTEMS If d, is defined as the diameter of line of length L, that will give the same friction loss as Ly feet of line dg, Equation 10.33 may be rearranged to yield: (10.35) Equation 10.34 is particularly useful, for the system shown then may be regarded as consisting of Le + La feet of line of diameter da. Once this equivalence is established, the problem may be solved in the conventional manner. Equation 10.32 is summarized for large pipe and small pipe in Table 10.2. TABLE 10.2 Equations for Lines in Series Lines in Series Large Pipe | Small Pipe Le Le” Equivalent Diameter | 4, ~an(E*) se=ta( 4] B a l dg Equivalent Length Example 10.7: Express the equivalent length of the following system in terms of 100 mm [4 in] pipe, assuming that the nominal diameter is the actual inside diameter. TT 200m 1000 m 2500 m 200 mm 4100 mm 150 mm . From Equation 10.34, Lioo = (2000)(100/200)**8 + (2500)(100/150)*** = 423 m Total equivalent length of 100 cm pipe = 423 + 1000 = 1423 m Therefore, the system shown in Example 10.7 may be considered as equivalent to one consist- ing of 1423 m of 100 mm [4 in] pipe. Capacity or pressure-drop calculations then may be made on the latter system using the correlations previously presented. Most gathering systems contain one or more branch lines. If the system is very complex, such as the one on the right, one has a large number of calculations to make. As a practical matter, these will be made on a computer. This essentially is a trial-and-error approach, although 7 some short cuts are possible. What one will do in principle is to use the fundamental equations discussed previously and combine these with the concepts of series and parallel flow. CHAPTER 10 335,FLOW OF FLUIDS & e A ‘The total system is made up of a series of branch lines coming SN together at a common point. In the sketch, on the left. cu = Gan + 4eD and Py is common to all three lines. In the usual case, the flow rate from each well or production fa~ cility is estimated from reservoir analysis and the design of the down-hole equipment. The problem is to establish line sizes and line arrangement to optimize initial cost and pressure loss. The proper approach is to plan the total gathering system at the time field development com- mences. Spacing and deliverability can be estimated within reason to set up an expected grid. Some modification will be needed usually, but the system will be more efficient than one which is expanded in a rather haphazard manner., Too often the resultant is an operating nightmare and is quite inefficient. The procedure used for system design may be varied. One approach may be illustrated using the simple sketch shown previously. 1. Establish qap, Quo. Pa and Pp. This fixes qcp- 2. Establish a reasonable pressure for Point C. 3. Guess a pressure for Point B, between pressures established at A, C and D. One method is to prorate on distance. Calculate d for each line, using methods outlined previously. : 5. Repeat the calculation for various combinations of pressures, flow rates and line configura- tions until a grid is found that is satisfactory. For preliminary studies we sometimes take a short cut by assuming a line velocity of 2-3 nvs [6-10 fi/sec] for each well flow rate to establish a diameter for its line. With these diameters and map lengths, a tentative grid may be drawn. AP is found for each leg and one can “work through” the grid to find the pressure distribution. In most cases, this simplified approach will check more complex methods rather well. In most liquid grids the oil properties are relatively constant, as are wellhead pressures. If this is the case . AP; & (d,v,L) One can set up a series of relative numbers for calculation of the system. If Pa and Pp are at the same pressure and at about the same elevation, lines A-B and B-D behave as loop lines and an equivalent di- ameter may be found. We can use the principles to algebraically develop many working equations that describe the specific system being studied. 336 VOLUME 1: THE BASIC PRINCIPLESGAS FLOW A Single Phase Flow Splitting A common problem also encountered is where the flow from a single line is split into two or more parts, as in the figure at the right. In c the rare case where the pressures and elevations at A and B are equal, this may be handled as a loop problem — in all others the calculation of flow is theoretically a trial-and-error process. However, this tedious ap- 8 proach usually may be eliminated by the following method. 1. Assume a pressure at Point C that is high. 2. Calculate AP; for lines CA and CB. 3. Calculate qr, for both lines. 4. Knowing the total flow rate at C and the relative qz's from above, calculate the actual flow rate for each line. 5. Using the actual flow rates, calculate AP; for lines CA and CB using a correlation of f vs. Re. 6. Calculate pressure at C for both lines. 7. If the two pressures at C check within a few percent, the actual flow rates calculated in (4) above may be used. If not, a new value must be assumed for the pressure at C and the cal- culation repeated. In most instances the above process is only necessary one time, particularly if the pressure at C is assumed high enough. Often it will work even if the assumed pressure and that calculated in Step 6 differ widely. As a rule-of-thumb, the assumed pressure should be about 15 times the maximum eleva- tion change, expressed in pressure terms. GAS FLOW Principles The fundamental thermodynamic equations used for liquid flow also may be used for gases. In order to conveniently handle Equation 10.6, several assumptions are often made: Aw? 2fLv? fvar += (ax)+ 409. __2fhv Be 28. Bed (10.36) That no external work is done by or on the system, i.e., W = 0. That the flow is isothermal. 1 2 3. That the changes in elevation on a long pipeline are negligible, i. 4, The flow is steady state. 5. Pressure loss due to kinetic energy changes are negligible. These three assumptions allow Equation 10.6 to be rewritten as: 2 fLv? aaa (10.37) Jvar CHAPTER 10 337FLOW OF FLUIDS To evaluate the integral term the following assumption is made oRT,, Z, =— 10.38) : (10:38) where Tyy and 2, represent the mean temperature and compressibility factor over the integration range. Substitution of Equation 10.38 in Equation 10.37 and integration gives the “Basic Equation” shown below: Pye |] £YL Tn 2m p? -P?)a5 | ak [5 (ri Pz) (10.39) All of the above assumptions are usually satisfactory on the typical long pipeline, However, those equations derived on this basis often contain an efficiency factor, E, to correct for mode! limita- tions (see Table 10.3). The assumption of isothermal flow has little effect on the final accuracy, which may be proven by assuming adiabatic conditions, the opposite extreme. The use of a mean compressibility factor zm is convenient. It may be determined at the aver- age pressure Pp, Which in tum is estimated. from the equation: =23 [c +P2)- (2, (10.40) where P; and P, are the inlet and outlet absolute pressures, respectively. The mean temperature can be found from the equation: (10.41) Where = temperature of pipe surroundings Standard Equations Table 10.3 summarizes several standard equations derived from the basic flow equation based on the above mean conditions. Equation 10.42 is the basic equation containing a friction factor “f.” The Weymouth and Panhandle A and B equations are merely the basic Equation 10.42, using the fric- tion factor correlations shown. The Weymouth equation assumes “f” depends only on pipe diameter; the Panhandle equations use different correlations for “f” as a function of gas flow rate, relative density and pipe diameter. The AGA equation uses the “f” values shown in Table 10.3. The drag factor, Fs, in the Partially Turbulent AGA Equation (Equation 10.46), accounts for several factors: pipe wall condition, welds, changes in line direction, isolation valves, etc. It is com- parable to the pipeline efficiency used in the Panhandle equations. The AGA"?! has developed a graph for Fy as a function of the “bend index” in degrees/mile, For bare steel or pig-burnished pipe walls, the drag factor varies from about 0.92 to 0.98 depending on the bend index. 338 VOLUME 1: THE BASIC PRINCIPLESGAS FLOW TABLE 10.3 ‘Summary of Gas Flow Equations Basic, Eq. 10.42 Te c= Weymouth, Eq. 10.43 ox[Ee ae-*[B Panhandle A, Eq. 10.44 Panhandle B, Eq. 10.45 e{_ (e-em 7" one] [amare AGA, Eq, 10.46 _ te _ (Rm) a a ole] [does wv yom (ne re)a’ El OME) ref { (RB pe ] [Saesea a Te 1.0788 (Re -w)ate poss wl] [areca 2 Z _ SI s () K = 5.746 x 10° K = 1.185 x 10° 321 x 10" 0.003 60 _ Gray | [LK = 5.746 x 10° Partially Turbulent €) = FeV Vice, - wf vol $5 }-os] FPS K = 38,774 K = 433.49 0.008 os K = 435.87 __ ais @rar* _ 0.003 59 Gy? K =38.774 Fully Turbulent 4 1og(3.74 d/e) Where: CHAPTER 10 . SI gas rate at Tye, Pye md absolute pressure kPa pressure, standard conditions kPa mean absolute temperature of line K temperature, standard conditions K ground temperature K inside diameter of pipe m absolute roughness m pipe length m viscosity Pas gas relative density S ‘mean compressibility factor Fanning friction factor - pipeline efficiency - Reynolds number > transmission factor (,//f) - drag factor (see text on page 338) - FPS seffday psia psia eR oR RK in in mi Ib/ft-sec 339FLOW OF FLUIDS The use of a transmission factor (1/f)°5 is common in gas transmission. It is a function of Re. Figure 10.5 shows the basic relationship for the flow conditions shown. The Weymouth equation is a horizontal line for fully turbulent flow. The smooth pipe law (no roughness) and Panhandle A paralle! the AGA equation for partially turbulent flow. The AGA equation for fully turbulent flow is a series of horizontal lines depending on pipe roughness (€). 108 std m2%/d 03 04 05 06 08 14 2 a4 art TTT T TT ze} : : = “Transition Point bre) ae = Peter en: i" = ee & ay Tarbert 5 3 _|- il 5 | & = 0.0256 mm [0.001000 in} T ‘AGA, Fully Turbulent [0.008 in) |_ «0.0468 mm 0.00180 in ” | | : 10 20 30 BBO 100 e006 a0 Bo B00 1000 Figure 10.5 Transmission Factor Comparison 40 VOLUME 1: THE BASIC PRINCIPLESGAS FLOW Although Figure 10.5 shows a sharp change between partially turbulent and turbulent flow, there really is a transition zone. If one wishes to use an AGA equation, use Equation 10.47 to deter- mine which one. Re=20.91 (2) (2 | (10.47) If the actual Re is less than that calculated from Equation 10.47, use the partially turbulent form; if greater, use the turbulent flow equation. For most large diameter natural gas transmission lines operat- ing dry in a non-corrosive environment, € = 15 — 33 jm [600 — 1300 j1 in] with 19 um [750 p in] a reasonable average. It should be noted however that in corroded systems e can be as high as 100 to 1000 1m [3940 to 39 400 p in}. . In Example 10.8 the range of answers obtained applies only to this example, but is indicative of the spread usually obtained. In the final analysis, the usefulness of a given equation will depend on experience with it in varying circumstances. The Weymouth equation is used quite often on gas gath- ering system design since it maximizes pipe diameter needs for a given flow and pressure drop. Determine flows using the Basic, Weymouth, Panhandle A and B, and AGA fully turbulent equations (Table 10.3) from the following information. i FPS _ Where: P, = absolute pressure 3448 kPa 500 psia P, = absolute pressure 3103 kPa 450 psia P,, = standard pressure 101.3 kPa 147 psia Ty. = standard temperature 288 K 520°R d= inside diameter of pipe 0.3048 m 12in 7 = gas relative density 07 07 L = pipe length 16 100 m 10 mi Tm = mean absolute temperature of line 300 K S41°R _ %q = mean compressibility factor 0.96 0.96 B= viscosity 0.01 ep 0.01 ep E = pipeline efficiency 0.92 0.92 € = roughness 19 um 750 pin Solutions: oe eeoee ee 10° std m3/d MMsefd Basic* 1.28 45.4 Weymouth 1.09 38.5 Panhandle A 1.38 48.9 Panhandle B 1.44 51.0 AGA (fully turbulent) 134 413 * friction factor obtained from Figure 10.1 is a Moody “f’. Must divide by 4 to obtain Fanning “f”. CHAPTER 10 341FLOW OF FLUIDS Maximum Velocity The maximum allowable velocity in a gas line is governed by noise, pressure loss and surge considerations. Equation 10.48 is a guideline for establishing any maximum velocity limits. Design velocity always should be less than this maximum velocity. A — (10.48) ()* ooo Se FPS. Where: =" velocity m/s fulsec p = density kg/m? Ibm/ft ‘A = constant : 146 120 At a pressure of about 7.0 MPa [1000 psia] Equation 10.48 will show a maximum velocity of about 17-18 m/s [56-59 fUsec]. At about 14.0 MPa the maximum velocity will be about two-thirds of this amount. Static Pressure (Head) in Flow Lines All equations like those shown in Table 10.3 are based on the assumption that the line is hori- zontal; i.e., there is no potential energy change affecting P, and P2. In actual practice the line may be going uphill or downhill. + For single phase gas flow a correction is sometimes made for the relative elevation between Points (1) and (2). Before using the horizontal equations some correction must be made to P; or P2, to convert the AP to what it would have been if the line were horizontal, i.e. adjust APioat for AP ctevation- Since the gas is compressible and tempera- ture varies, many models may be used. The simplest equation uses an average temperature and com- pressibility, ignores” any kinetic energy changes and assumes a constant friction factor. For these assumptions, from the basic energy equations, one may derive the equation (10.49) _ St FPS_ Where: Hi = head m f y = gas relative density =] - Ta = mean gas temperature K R 2m = mean gas compressibility . - Pq = pressure at bottom of static column (horizontal) MPa psia Py = pressure at top of static column (actual elevation) MPa psia A. = constant 29.28 53.34 342 VOLUME 1: THE BASIC PRINCIPLESGAS FLOW Equation 10.49 also can be written in the form Px =(Pa)(e*) (10.50) Where: e = 2.718 Hy B S = AT, 2m Subscripts A and B are used to show the pressure change due to head only. Do not confuse these with measured (or specified) pressures P; and P2. In the uphill case, P2 for use in the horizontal flow equations will be greater than the actual P._ In Equa- tion 10.50, actual P; will be inserted as Pp, and the calculated value of Pq will be the new P> for use in the horjzontal flow equations. For downhill flow the effective hori- zontal P, will be less than the actual P, since H in Equation 10.49 (and s in Equation 10.50) will be a negative number. ~ — Example 10.9: A gas line runs uphill. The following data apply: 49 H = 100 m (328 ft], Tm = 300 K [80°F], 2m = 0.90, ¥= 0.70, P) = 4.0 Mpa [580 psia], P; = 6.0 Mpa [870 psia] 1 oy 60 404 — St 7 FPS 100 m)(0.7 328 f)(0.7 0m) _ pooges 28 2)00.) __ ooogss (29.28)(300 K)(0.9) (53.34)(540°R)(0.9) P, =4 MPa e885 = 4.04 MPa Pa = 580.14 psia £988 = 585 psia (Which is equivalent P; for horizontal flow) —} (Which is equivalent P, for horizontal flow) If the liné had been horizontal, the total pressure drop would have been less and P) would have been higher. In Example 10.9, the total pressure drop of 2.0 MPa is allocated as follows: 1.96 MPa is due to friction and 0.04 MPa is due to elevation change. If the line goes downhill (H is negative), the equivalent horizontal P> would be less than the actual P. The basic rule — use as P2 in the pipeline equations the value that should occur if the line is horizontal. Also remember that length L should be established from the line profile and not merely the geographical distance between (1) and (2). The alternative is to incorporate the pressure and length corrections directly into the flow equa- tions, The method shown is simpler and more straight forward. For most actual gas pipelines the correction for elevation changes and length is very small. If in doubt, make it. CHAPTER 10 43FLOW OF FLUIDS Complex Gas Flow Systems The principles outlined for liquid flow also apply to steady-state gas flow. Table 10.4 summa- rizes the resulting equations. ‘The sketches below the table outline the nomenclature used. The sub- script (¢) denotes a single line equivalent in length or diameter to another line in series or looped lines. Examples 10.10-10.11 will serve to illustrate the application of the gas flow equations. For simplicity, the Weymouth equation, uncorrected for gas compressibility, will be used as an illustration of the principles involved. Example 10.10: A portion of a large gas-gathering system consists of a 15.41 em [6.067 in] line 15 km [9.32 mi] that is handling 227.9 std m°/d [8.07 MMscfd] with an average Rel. p of 0.64, The pressure at the upstream end of this section is 2586 kPa(g) [375 psig] and the average delivery pressure is 2.069 kPa(g) [300 psig]. The average temperature is 23°C [73°F]. %m = 0.9 Py Ps Pe << &§ —— + | + 15 000-, ——> Due to new well completion, it is desired to increase the capacity of this line 20% by looping with additional 15.41 cm [6.067 in] line. What length is required? Let La represent the length of the loop section. SI Solution: New flow rate, q) = (1.2)(227.9) = 273.5 x 10° std m¥/d The loop may be represented by a single loop with an equivalent diameter, d 4, =[(5.41)"" (san? ]* =19.98cm=0.1998m ‘The Weymouth equation for each section may now be written. Looped section T,(0.64)(0.9)(296) 101.3, > pr? PS 273 500 = 1assxio"( 28 J (2687) —P, | (0.1998)? Single line section 23 500 =1.185x10"( = J 05 —— (0.1541)* 67 1013 | (0.1841) Note, 101.3 kPa has been added to each gauge pressure. Since there are two equations and two ‘unknowns these equations may be solved algebraically. For this example L = 6.1 km. 34 VOLUME 1: THE BASIC PRINCIPLESGAS FLOW Example 10.10 (Cont’d.): Alternatively, one may use the equation from Table 10.4 xeaali-(i5) feoaor Lg =(0.407)(15)=6.1km FPS Solution: New flow rate, qy = (1.2)(8.07 MMscfil) = 9.68 MMscfa The loop may be represented by a single loop with an equivalent diameter, de. 4, =[(6.057)*? + (6.067) P* = 7.868 in The Weymouth equation for each section may now be written. Looped section os 520 (389.7)? -P2 2.667 9.68x10° = 433.49| “|| 7A | (7.868) Ee (23) fees 9(533)| (788) Single line section ape 2 a seenio <0na9( 29) Pa-G147) (6.067)? 14.7 )| (932-L ,)(0.64)(0.9)(533) Note, 14.7 psia has been added to each gauge pressure. Since there are two equations and two unknowns these equations may be solved algebraically. For this example La = 3.8 mi. Alternatively, one may use the equation from Table 10.4 x=4aft-(75) |-os7 L 4 = (0.407)(9.32)= 3.79 mi CHAPTER 10FLOW OF FLUIDS 4 4 | | padooy uonseg =x | uieg auf axe SOU] jereq pur feUISUC Jo sxoroueIG | padooy oury asnug [ois eP sivzP swt ae i | Ge oa APP) 1 | [tte | | (¥p) (Yh) | { T P Pht) _ ap te 7b) eel SP) St) ey padoo'] uonsesg = x wah 5) Axe, smoyg pue suoiaureic - sdooy | a4 4, v1; Ie 9: q wk, a TD, CTD CTD aia “te w9eP ae? oP oP seP | __~ wi8u07 40 roureig woqeamba iz - jeq Ul Sour] OM] 10g = /*p) a a (%y\ (8 | * - P 5) (®p eT | (a ie Ye | ve) sh”? | a} p) #7] ‘ypue7 quapeatnby a (aq) i ® (*1)4 = : ("1)q, | “Tap ou vel "1) | wl “1) | xf (5), Aaa yr \ 2p sojourig woeambg | Soudg uF SUPT OAL 104 ‘apuequed | ‘yynourkany, prong 1 —— 'SMO|4 SBE) Xxe/dwWOD 40} SUONENdS vol a7evL LL VOLUME 1: THE BASIC PRINCIPLESGAS FLOW [exemple 10.11 The system at left is to be designed for the gathering of gas. Well “A” is a flow- ing gas well that will deliver 84.7 x 10° std m’/d [3 MMscfd]. Production facili- ties at "B” handle high gas-oil ratio oil wells. It is necessary to determine what compressor discharge pressure is necessary in order to A D 1521mmj6.0in) C deliver 49.4 x 10° std m’/d [1.75 MMscfd] from point B. SI Solution: Line CD Line BD FPS Solution: Line CID Line BD CHAPTER 10 It is obvious that once the pressure at point D is found, Py may be found; furthermore, the average specific gravity in line CD is estimated. 76.1 km [10.0 miles] The pressure at the transmission line (point C) is 2.07 MPa(g) [300 psig]. Average flowing temperature will be 29°C [85°F]. The gravity of the gas well gas is 0.63 and that of the solution gas is 0.71. The line diameters m have been arbitrarily fixed because of pipe avail- 84.7(0.63) + 49.4(0.7. 134.1 P3 —(2171)7 (16100)(0.66)(302) 134.1510? =1.185 x107( 288 ) 101.3, os | (0.15217 Pp = 2426 kPa 288 | _ Pg - (2426) force (Patina Pp = 2512 kPa 494x1 05 ] (0.1024)? 05 4.75 x10° = 433. «(3 J PD-G147) | or" 14.7 }} (10)(0.66)(545) Pp = 351 psia 520)| Pg -(351)? Bear 1.75 x10° = 433.49| —— || 8 ‘4 j (Shores; 4% Pp = 364 psiaFLOW OF FLUIDS UNSTEADY STATE GAS FLOW All of the discussion to this point is based on steady state flow. Unsteady state flow is, how- ever, an important consideration. This cannot be analyzed thoroughly and conveniently by manual cal- culations but there are some simple approximations that may prove useful. Transient Pipe Line Flow Any time the rate of withdrawal of gas from a line differs from the input, unsteady state flow is occurring. During routing operations this occurs because the line is often used for storage to dampen out peak loads. During peak demand, output exceeds input and pressure is declining throughout the line. This is referred to as drafting and packing of the pipeline, Between peak loads input exceeds output and the pressure builds up again. To predict the behavior of the line, one must start with the basic continuity equations and then develop a solvable model that will approximate the transient behavior of the pipeline system. This is strictly a computer modeling game. Examine a model carefully to ascertain it if fits your specific sys- tem situation.” Blow Down and Purge Occasions occur when it is necessary to blow down and purge a gas line. This is a special case of unsteady state flow. One may estimate the blowdown time by some simple equations.1?!2101 s. For the critical flow case (when the ratio of the higher pressure to lower pressure is larger than two): BV os (P, t=|— 2) Inf (10.51) CaAy at Py A ee peccilesis Where t = blowdown time s sec constant 0.09 53 actual system volume mo a valve discharge coefficient - - blowdown valve area ‘gas specific gravity 7 - average gas compressibility factor S = erage gas temperature kK R initial system pressure kPa psia final system pressure kPa psia ‘The gas temperature and compressibility factor upstream of the blowdown valve change during the blowdown period. Thermodynamically the blowdown process falls somewhere between isentropic and isenthalpic. Estimating the average temperature and compressibility factor requires some know!- edge of the thermodynamic path which is not known exactly. For most blowdown cases use of the ini- tial temperature and compressibility factor is a reasonable assumption. MB VOLUME 1: THE BASIC PRINCIPLESUNSTEADY STATE GAS FLOW Equation 10.51 was developed for critical (sonic) flow. Once the system pressure falls below about 200-300 kPa [30-50 psia] the flow is probably subcritical. Fortunately the time required to com- pletely depressure a system to atmospheric docs not usually have to be calculated. When it does, Equation 10.51 has been found to give a reasonable estimate of total blowdown time if P2 is set to about 80% of atmospheric pressure. Pressure Surges on Closing a Valve When a valve is closed on a line a pressure surge occurs. The kinetic energy of the fluid is converted to internal energy when flow stops. A wave (surge) travels back through the line countercurrent to the fluid still flowing forward, but slowing down. Each section of the line behaves in a different manner. Singg flow does not stop upstream when stoppage occurs at some point down- stream, the pressure rise decreases as the wave front travels upstream. Some of the energy is absorbed by the compressibility of the fluid and the expansibility of the pipe. The more energy absorbed, the less the pressure rise. Thus, compressibility of the fluid is a pri- mary factor. Since water and stabilized crude oils and condensates are essentially incompressible, they represent the primary problem. Compressible liquids, such as NGLs pose less concern. Pressure surges are generally not an issue in gas pipelines. A thorough analysis definitely requires a computer solution. Reference 10.14 provides an ex- cellent summary of the problem and some important references. The maximum pressure surge is a function of two components: a) the instantaneous pressure increase which occurs at the point of total blockage, and b) the subsequent pressure rise which occurs due to the “line pack” effect. The instantaneous surge can be estimated from Joukowski’s equation !, AP, =(A)(p)(c)(Av) (10.52) St FPS Where: “AP, = surge pressure kPa psi p = liquid density kg/m? Ibm/f © = sonic velocity in liquid ms fusec ‘Av = velocity change ms fusee A. = constant : 0.001 0.000 216 “c” varies with the compressibility of the fluid, pipe diameter and pipewall thickness. For sta- bilized crude oil and condensates “c” is approximately 1000-1200 m/s [3300-3900 fi/sec]. The additional pressure rise which occurs due to the “line pack” effect can be 1.5 to 3 times the instantaneous pressure surge?! depending on the line segment length, sonic velocity and normal steady state pressure gradient. Surge pressures can be mitigated by slowing the valve closure time and/or installing a fast-re- sponse pressure relief system near the point of surge origin. Rupture disks have been used in this ap- plication. CHAPTER 10 ugFLOW OF FLUIDS Pressure Testing Gas pressure testing of a line is a form of unsteady flow in the sense that pressure must be held for sufficient time to assure that there is no leak. The time necessary for such testing may be estimated by the equation 2, =O) (10.53) P 7 St FPS Where: tm = time necessary, minimum h hr d= internal pipe diameter cm in L = length of pipe section km mi P, = initial test pressure MPa psig A = constant 0.002 3.0 When a line has been shut in for at least the above amount of time, it may be considered “tight” if the following pressure loss has not been exceeded. AP ax =D (10.54) (Aya) + st FPS Where: ‘AP pax = acceptable pressure loss kPa si t = shut-in time h hr 4. = pipe diameter cm in A = constant 0.372 949 P) = initial test pressure MPa psig TEMPERATURE CHANGES IN PIPING Most of the equations used commonly for pressure loss calculations require the use of a single value of temperature. This will be some type of an average temperature. In a numerical (computer) solution the line is divided into a series of constant temperature sections; an analytical (manual) solu- tion may treat the entire line as one section. However it is done, the temperature used should be repre- sentative of what is anticipated. The value used has a direct impact on the physical property values used. Prediction of the temperature distribution in the flowing system may be as important as the pressure profile. It affects many design considerations surrounding the line. Crude oil pour points, gas hydrates, vapor-liquid phase behavior and water content of gas are all temperature sensitive. A reliable temperature prediction is an early step in the calculation procedure. The temperature at any point is predictable through use of a first law thermodynamic energy balance. If potential and kinetic energy changes are ignored — and work equals zero in the pipe section involved ~ this first law balance for a steady state system reduces to AH = Q. 350 VOLUME 1: THE BASIC PRINCIPLESTEMPERATURE CHANGES IN PIPING The enthalpy of a gas increases with decreasing pressure; it is almost independent of pressure for a liquid. So, the enthalpy at any point in the line depends both on pressure at that point and the heat energy lost or gained through the pipe wall. From heat transfer principles, Q= UAAt. For a circular pipe this may be written as: Q=(U)(ndL)(AT,) (10.55) overall heat transfer coefficient 3.1416 pipe diameter pipe section length = Ly-L, log mean AT = (AT,-AT2)/In(AT\/AT3) heat lost in line section of length L T-T, h-T normal ground or water temperature at that point (far enough away from pipe to be undisturbed by it) Where: The general relationship is shown at right. If the fluid temperature is greater than T,, heat transfer will be negative as shown; if it is less than T,, the Q will be posi- tive and the curves will approach T, from below. | Line A is for the case where the effect of pressure 7 on enthalpy is ignored, as per the discussion following. — T, This is a simple heat transfer process and the fluid tempera- ture cannot drop below Ty. Line B reflects the additional change in temperature due to the effect of pressure on enthalpy. In commercial practice this line normally would fee not drop over 2-3°C below T, for ordinary pressure drop situations. ‘Ground Temperature, Ty If no phase changes occur in the line section, the following equation applies. Ah= ‘fmcyar + ily -T(av/ar), JaP (10.56) i, ®, This equation relates enthalpy to heat capacity (C,) and P, V and T. The second term is zero, or essen- tially so, for an ideal gas and a liquid. If the pressure drop in the line segment (dP) is small compared to the absolute pressure, the second term of the equation is small compared to the first term. For a gas line at a pressure above 3500 kPa [500 psi], ignoring this second term may not be important. There are thus three basic solu- tions for the Ah equation. 1. Use an enthalpy-PVT equation of state program. 2. Approximate the second term by assuming a series of Joule-Thomson expansions. 3. Ignore the second term completely. Method (1) is a routine computer solution. Method (3) is an easy manual solution. The practical accu- racy is as good as for (1) if a gas line is calculated in segments so that the pressure drop per segment CHAPTER 10FLOW OF FLUIDS does not exceed about 20% of the initial pressure. Method (2) is a carryover from pre-computer days as a manual method to calculate the effect of pressure on temperature (A, B). It could be programmed but is not as convenient as Method (1). Both Methods (1) and (2) require iterative solutions. Method (2) can be approximated by translating the change of enthalpy with pressure into a temperature change per unit length. If this is done, Equation 10.57 results. hn (ee (1, re aL (1, +L2)-(T, + Ya) ane ——__St_|__FPs_ Where: heat capacity IkgeC Btu/lb-°F distance from initial point m ft distance from initial point m f le-Li m ft temperature at Ly °c F temperature at Lp °c F gas sp gr (air = 1.0) = - ‘ground or water temperature °c °F Joule-Thomson coefficient °Chm orf = outside pipe diameter m ft = overall heat transfer coefficient ‘Wim??C Brwhr-f?-°F = gas mass flow rate kgs Ibe = dUmC,) Because of the Joule-Thomson effect, it is possible for the gas temperature to be less than that of the ground. This is only likely to occur with long lines possessing a large pressure drop or at regu- lator stations. For lean, pipeline-quality natural gases the cooling due to Joule-Thomson expansion is approximately 0.004 to 0.005°C/kPa [0.05-0.06°F/psi] at pipelines operating near 6900 kPa [1000 psia). With Method (3) the term “J” in Equation 10.54 is zero and Equation 10.58 results. a T,-T,) 4 inf 227s.) 20k (10.58) T,-T,} mC, The above equations normally are solved for T. It is obtained by taking the antilog of the left-hand side, This means that any errors in the data are amplified. Overall Heat Transfer Coefficient (U). This is the single number that represents all of the resistances in series. In a buried pipe, all of the following resistances to heat flow can occur: 1. Film coefficient between fluid and pipe wall Inner pipe wall conductivity (tubing) Annular space between inside and outside pipe Pipe wall Pipe coating Insulation Concrete layer or bond NAY EYR 352 VOLUME 1: THE BASIC PRINCIPLESTEMPERATURE CHANGES IN PIPING 8. Sand backfill . 9. Native soil near pipe at temperature above normal due to heat gain from pipe A line buried in soil normally would have resistances 1, 4, 5, 8 and 9; buried on the sea bottom 1,4, 5 and 7 may apply. In a wellbore with fluid flowing inside the tubing 1, 2, 3, 4, 7 and 9 would be in series. The use of insulation can be justified only in very special situations. Resistance 9 is necessary to correct for the temperature rise around the pipe above T,, the tem- perature of the bulk of the ground at that depth. Temperature surveys show it is common for the ground temperature to be higher than T, up to 5 m [15 fi] from the line. This is an effective resistance reducing heat flow. As with a heat exchanger, the most reliable values of “U” are found by test. A study of unin- sulated gathering and transmission lines shows values of U from 0.4-3.5 W/mK [0.07-0.6 Btwhr-ft?-°F]. Most of the values measured are in the lower half of this range. The soil resistance is limiting. The lower end of the range applies to dry loosely-packed soil and the higher end to tightly-packed soil with high moisture content. Pipelines laid on the seabed or river crossings will exhibit higher heat transfer coefficients, par- ticularly if a strong current is present. For exposed, concrete coated lines, the heat transfer coefficients can be as high as 7.0 W/m?K [1.25 Brwhr-f-°F]. For buried or partially buried lines the overall coef- ficient will be nearer the values quoted in the paragraph above. A given line may cross several different heat loss environments. The most practical approach is to recognize that a calculation based on a single value of “U” is basically invalid. Pick a likely range of values from test data to arrive at a range of answers. Then design for the worst case. Temperature T,. This is not a constant quantity but depends on air temperature to some de- gree. Ata burial depth below the “frost line” the soil temperature generally will vary from 2-16°C [35-60°F] seasonally. Maximum ground temperature will lag air temperature by a month or two. About the same temperature range will be encountered with burial in water over 30 m (100 ft] deep. In a body of water with no pronounced currents, the fluctuation will be less. Some lakes will have a bottom water variation of no more than 3°C [5°F] year round, Temperature surveys should be a part of planning if the temperature of line contents is a design factor other than pressure loss calculations. For very preliminary considerations the minimum temper- ature seldom wilt be less than 0-1°C [32-34°F]. Because of changes in T,, the value of “U” will vary seasonally also. In a temperate climate with discernible seasons, the “U” may be 50-60% higher in winter than in summer. In theory this should not occur, “U” being independent of T,, but it has been noted in tests. One must recognize that the models only approximate the real world. Above Ground Lines. The above equations do not apply for above ground lines; no T, is appli- cable. The maximum temperature of fluid in the line is the result of gain of daytime heat from the sun by radiation and the corresponding loss by convection to the air. The relative rate of these would de- termine the rate of heating with distance. The temperature depends on + Time of day + Wind velocity + Atmospheric conditions _* Color and character of pipe surface + Air temperature (T,) + Flow rate and properties of fluid in pipe CHAPTER 10 353FLOW OF FLUIDS There are no absolutely rigorous performance prediction methods, but data are available from line tests. ‘The maximum temperature for an above ground line may be estimated by the equation!” R T=|——~ |+T, (10.59) (nh,) esi __FPS Where: T = maximum temperature of fluid ec °F R= solar radiation absorbed Wim? Btw /hr/ft? hy ir film coeeficient for convection ‘Wim? Btu/hr-ft?-°F n= 31416 T, = ambient air temperature °C F Values of R and h, may be estimated from the table below. _R___|__ Wind Velocity | ky | __ Pipe Surface si FPS ms | mph | si | Highly oxidized steel 950 300 0 0 14 Oxidized 728 230 09 2 148 Normal 370 180 22 5 199 . Bright 348 110 45 10 256 | Aluminum Paint 285 90 67 15 28.4 | White Paint 238 15, 89 20 | 295 Equation 10.59 shows the maximum temperature possible with the sun directly overhead, a clear sky and a long enough line to achieve thermal equilibrium between radiation energy gain and convection loss. In a typical case, the rise in temperature above ambient will not exceed 10-16°C [18-30°F]. The actual temperature probably will be lower than the maximum unless the exposed line is very long. Fluid properties, velocity, pipe length, season of the year (sun position) and geographical position all influencé the actual radiation effect. Since not all of these are constant with time, neither is performance. At night, even in a tropical climate, radiation losses to a clear, dark sky may be enormous. So ... fapid temperature changes can occur. The best one can hope to do is estimate the magnitude of such changes for decision purposes. Insulation. ‘The purpose of insulation is to reduce the rate of gain or loss of energy, to or from a pipe, in a cost-effective manner. With a group of resistances in series, the one with the lowest ther- mal conductivity controls total heat loss. Subsea developments, particularly in deep water have spurred new interest in insulated flowlines. Deposition of hydrates and paraffins can be prevented by the use of inhibitors, or by main- taining the flowing temperature above the hydrate and/or pour point. The second option required insu- lated lines. Bundled flowlines are in widespread use in the Gulf of Mexico and North Sea. These lines use an insulated conductor pipe which contains the production flowline, inhibitor line, heating medium, 354 VOLUME 1: THE BASIC PRINCIPLESMULTIPHASE FLOW control line, etc. The insulation is typically polyurethane. Overall heat transfer coefficients vary from 0.26 W/mK [0.046 Btwhr-ft?-°F] for 54 mm [2-1/8 in] insulation thickness”1® to 10.6 W/m?K [1.9 Btu/hr-ft?-°F] for 11 mm [7/16 in] thickness.“°!? Wellbore Temperatures Prediction of wellbore temperatures parallels that for pipelines. In this case, though, ground temperatures around the pipe may vary substantially with depth. A series of models has been proposed for injection, production and geothermal wells. The prediction of surface flowing temperatures is critical to surface facility design. None of the models is completely adequate because of the inability to define all variables quantitatively. The work of Ramey is most often used as the basic equation. Actual wellbore then is used to adjust the constants. Effec- tive values of “U” may vary from 1.4-11 W/m?K [0.25-2.0 Btu/hr-ft?-°F]. They tend to be higher than for pipelines because of higher fluid velocities and film coefficients, The models used for calculation of well-bore tem- peratures are designed for computer usage. The accuracy of the output is contingent on the amount of hard data available for similar wells in the same area. In the absence of such data, the accuracy of such prediction may be no better than 15-25%. One of the common mistakes in sur- face facilities design is not allowing for a range of well- head temperatures and designing for the worst condition teeerterreregergs ‘Temperature depends on time. It takes finite time for the temperature profile to stabilize. This is true particularly in wellbore and gathering lines as wells are continually shut in and opened or flow rates change. With steam injection, geothermal wells and wells in permafrost or ice islands, insulation may be required to stabilize the temperature-time relationship. Reference 10.20 provides a good summary of this topic. : MULTIPHASE FLOW The use of multiphase flow systems is common practice in the oil and gas industry. Multiphase flow is encountered in the well tubing, flow lines and gathering systems. Furthermore, for transport of oil and gas to downstream processing facilities the preference, for economic reasons, is for a single pipeline in which both phases are transported simultaneously. Even in gas pipelines where the gas enters the line as a single phase fluid, condensation of liquids can occur due to pressure and tem- perature reduction along the line. As we have seen, the calculation of pressure drop in a single phase line is relatively straightfor- ward. This is not true for multiphase systems. In a multiphase line the gas and liquid can behave in a number of ways. The liquid may run along the bottom of the pipe (stratified or wavy flow) it may be entrained as tiny droplets in the gas (mist flow) or it may form slugs (slug flow). CHAPTER 10 355FLOW OF FLUIDS Figure 10.6 shows some of the common flow regimes observed in horizontal multiphase flow. Wavy Figure 10.6 Various Two-Phase Horizontal Flow Regimes The basic behavior of these two-phase systems depends on whether or not both the liquid and gas phases are present as a continuous phase; i.c., a continuous channel for flow exists for the phase. In bubble flow one has a series of bubbles of gas in the liquid. Gas is not a continuous phase. At the other end of the scale one sees spray or mist flow — small particles of liquid suspended in the gas. Liquid is now the discontinuous phase, gas the continuous phase. In both cases the discontinuous phase may be considered to be merely altering the flow behav- ior of the continuous phase. One very simple approach is merely to modify the properties and/or fric- tion factor of the continuous phase and then treat the system as single-phase flow. In bubble flow a liquid, single-phase calculation would be made. In spray or mist flow a gas correlation is used after same correction, This often is done by default when correlating data on existing systems. Most crude oil lines some gas and most gas lines contain at least entrained liquid. This is why we need separators and scrubbers. In some of the regimes shown in Figure 10.6 both phases are continuous. This is a more com- plicated system because of the interactions between discrete phases. The gas moves more rapidly than the liquid, There is a shear force at the phase boundary. The character of flow depends on such vari- ables as phase densities and viscosities, the velocity of each phase, gas-liquid ratio and properties like surface tension In a given length of line several flow regimes might occur because of varying forces and gas-liquid ratios. The latter changes as liquid condenses from gas or gas is formed from liquid, as dic- tated by phase behavior. Consideration of variables like these is a necessary part of developing a cor- relation. Figure 10.7 shows four basic regimes that occur in vertical flow. A fifth regime, not shown, may be called froth flow. As noted later, this occurs at high liquid throughputs when the gas bubbles in the liquid are dispersed in an ever-increasing number of small bubbles as gas throughput increases 356 VOLUME 1: THE BASIC PRINCIPLESMULTIPHASE FLOW ed 5 ali ola Gilizeaeadl: biucit--na | nna Direction of Flow (a) Bubble (b) Plug (©) Sug (@) Mist Figure 10.7 Vertical Two-Phase Regimes In vert gles to it). This results in slippage. Consequently, than horizontal flow and may be more complicated. flow the force of gravity opposes the dynamic forces (instead of being at right an- vertical flow exhibits some different characteristics Consider friction loss for a moment. The friction factor is a measure of all irreversible energy changes occurring in the pipe. In single-phase flow the friction factor primarily is a representation of the loss between fluid and the pipe wall and is correlated as a function of fluid properties, fluid veloc- ity, pipe diameter and pipe roughness. In two-phase flow there are two phases, with different properties, interacting with the pipe wall as well as with each other. A friction factor, therefore, depends on more, and complex, variables. The difference between correlations depends to a large extent on how these variables are treated in order to fit the available data. In addition to pressure drop a second parameter important to design and operation of multiphase systems is liquid holdup. The liquid holdup represents the fraction of the pipe cross s tional area which is occupied by liquid at any point along the line. Liquid holdup affects the area available for gas flow and is also important in estimating the size of liquid handling facilities (slug catchers) at the terminus of the line. It also sets the fluid density in-situ which has a signifi on pressure drop in in the line, particularly in vertical and inclined flow systems. Each flow pattern (or regime) requires a different mathematical model to calculate pressure drop and liquid holdup. All models are bas cussed previously. As with d on the same thermodynamic principles and fluid flow principles dis- ingle-phase flow, a prediction of the “lost work” (friction drop) term is CHAPTER 10 357FLOW OF FLUIDS essentially empirical. Thus, any one method will be most reliable in the range of pipe sizes, fluid char- acteristics, flow conditions and geometry, and gas-liquid ratios used to develop the correlation. There is no single best correlation for universal use. When using a given model, examine the Kinds of systems on which it is based. This is why data on existing systems is vitally important. Is the ‘model and its supporting data physically compatible with your proposed system? Like all semi-empiri- cal equations, extrapolation beyond the range of data used to develop the correlation can lead to seri- ous errors. ‘The primary concems of the engineer working with multiphase systems are: + pressure drop/line sizing + liquid holdup + slugging and its effect on equipment sizing Each of these will be discussed in the context of flowlines/pipeline systems. Horizontal flow will be emphasized. Vertical flow in well bores will not be covered. There are numerous references to this topic in the petroleum engineering literature. Vertical flow in risers, particularly as it applies to slugging, will be discussed. Important Equations and Dimensionless Groups Several equations and relationships are used in two-phase models. am _ 4m VLAD AT oo A pLA xpd? eseeeeresrrerrereraetEE A +4g Vat + Msg — __FPS_ Where: . actual volumetric gas rate m/s sec actual volumetric liquid rate m/s fP/sec superficial gas velocity mis fusec superficial liquid velocity ms fsec ‘gas mass flow kg/s Ibm/sec liquid mass flow ke/s Ibm/see inside cross-sectional area of pipe m fe p, = gas density (at flowing conditions) kg/m? Tom/ft liquid density (at flowing conditions) kg/m? Tbm/ft} side pipe diameter m a liquid fraction (liquid holdup) dimensionless gas fraction dimensionless 358 VOLUME 1: THE BASIC PRINCIPLESHORIZONTAL MULTIPHASE FLOW In addition to Reynolds number, several dimensionless groups are also used to characterize fluid behavior. Froude Number: This dimensionless group represents the ratio of inertial force to gravity force. It can be used to characterize flow whenever gravity force influences fluid flow. 2 Fr=— yor Fr=~ (ed) ad The form at left is the basic equation. The right-hand expression is oftentimes more convenient fo use and is often referred to as a Froude No. even though it is actually (Fr)? Weber Number: The dimensionless group is defined as the ratio of inertial force to surface force. It is equal to twice the ratio of kinetic energy to surface energy of a given volume of fluid. It may be written as v vipd e=— or = We= PS (s/pd) ° Once again the right-hand form often is used because of its convenience. This dimensionless group is a logical correlation parameter where one phase is discon- tinuous because interfacial tension (@) affects drop size. We Euler Number: This dimensionless group is the ratio of applied pressure force to inertial force. It is a suitable correlating function in a flow system where pressure force is a control- ling force, particularly if one of the fluids is compressible. It is written as v _pv? Eu=— - or Eu= (2aP, ip)* 2AP In the above three dimensionless groups the nomenclature is as follows: Where. Vv = effective fluid velocity p = fluid density g = acceleration of gravity © = interfacial (surface tension) d= diameter of pipe P = pressure HORIZONTAL MULTIPHASE FLOW All multiphase correlations are some form of a momentum balance and continuity equation containing terms for the head, friction and acceleration effects. ota eset (sr) me ee ee Equation 10.60 includes several assumptions: 1) steady state, 2) no pumps or compressors in the system, and 3) adiabatic. With these limitations, this provides the basis of most fluid flow calcula- tions. CHAPTER 10 359FLOW OF FLUIDS The general approach has been to develop empirical relationships to describe the two-phase friction factor, fip and liquid holdup (which determines phase velocities and densities). In many cases these empirical relationships were applicable to only one flow regime, so several different relationships (often by different researches) could be used in the modeling of pipelines as the flow regime changed. Today, with the advances of computing speed, the approach has been to use one model (oft termed a mechanistic model) to describe the system over a wide range of flow conditions. The various correlations now available can be placed in three general classes. Class 1 a. Liquid holdup is not considered in density. Liquid holdup and wall friction losses are incorporated into a friction factor. ©. No distinction is made between flow regimes. Class 2 : a. Liquid holdup is considered in density. b. Liquid holdup may be correlated separately. ¢. Friction factors are based on composite properties. 4. No distinction is made between flow regimes. Class 3 a. Density is adjusted for liquid holdup. b. Liquid holdup is estimated using some concept of slip velocity (difference Between gas and liquid flow velocities). c. The continuous fluid phase(s) determine(s) wall friction losses. d. Flow regimes are considered. ‘Most of the correlations in the past twenty years are in Class 3 and are only applicable as computer so- lutions. Flow Regime Maps Superficial Gas Velocity, vsg, m/s Flow regime can affect model selec- tion in evaluating multiphase flow systems. Figure 10.872!) is a simple flow re- gime map for two-phase flow in horizontal pipes. It was generated using air and water at low pressure in small diameter lines but it is useful for visualizing how flow regimes change depending on gas and liquid flow rates. Superficial Liquid Velocity, vg, m/s Figure 10.97°22) shows another t io 0 widely used flow regime map for horizontal Superficial Gas Velocity, vq, fUsec or near-horizontal pipes. Relatively small - ——— ~—.--1 deviations from horizontal flow can signifi- Figure 10.8 Flow-Pattem Map of Mandhane et al.""°"” for cantly change the flow regime boundaries Horizontal Two-Phase Flow in Pipes on a flow regime map. As little as a 5% VOLUME 1: THE BASIC PRINCIPLESHORIZONTAL MULTIPHASE FLOW inclination can increase the slug flow region dramatically. The following nomenclature is to be used with the equations in Figure 10.9. peceeeee eu eect | Etec eS eee Where: (APYL) = pressure drop per unit length Palm psflft p = fluid density kg/m? Tbnvft} 8 = gravitational acceleration 9.81 m/s? 32.18 fUsec” d= inside pipe diameter m ft © = angle of inclination (horizontal = 0°) degrees degrees subscripts g and L = refer to gas and liquid phase, respectively For horizontal gas pipelines with low liquid loading operating at velocities of 3-5 m/s [10-16.4 f/sec] most horizontal pipelines operate in the stratified wavy regime. 101 r PAAR ‘Annular Dispersed Liqu(AD) te Uy i 108} ete 3 7 © 107+ 10 i | K tool Lit Sl | 10° 1 10 1 10 10" x 05 AP, 0s fe LL sL (PL -P,)gcos0 os Ps 0.5 Tr, — K=(F)(Re), Gatos 7 Figure 10.9 Flow Regime Map for Horizontal Two-Phase Flow CHAPTER 10 361FLOW OF FLUIDS Estimate the flow regime of a horizontal gas pipeline operating at 6900 kPa [1000 psia] and 15°C [59°F]. The gas density is 73 kg/m’ [4.6 Ibm/ft’] and the liquid density is 700 kg/m? [43.7 Ibm/ft’], and the liquid viscosity is 0.5 cp. The gas rate is 5 x 10° std m/d [177 MMscf]. The gas MW is 18 and the gas viscosity is 0.015 cp. The inside pipe diameter is 0.432 m [17 in]. The condensate loading is 56 m°/10° std m? [10 bbl/MMscf]. SI: Vat = 0.022 mis (Re)g. = 13 300 (AP/L), = 0.01 Pam Vog = 4.1 m/s (Resp = 8.6 x 10° (AP/L), = 15.8 Pam 0s (7) 7 05 ses (3) =0025 ( Ei ; 158 5 05 p B iz lg cal “se -lpocmeeneann | 4.=0.68 From Figure 10.9, the-flow regime is stratified wavy. FPS: Va. = 0.073 ft/sec (Re)g = 13 300 (APIL), = 4.4 x 10° psi/100 ft Veg = 13.4 f/sec (Re)sg = 8.6 108 (APL), = 0.07 psi/100 ft 05 2), [ea ( ar) 007 L Jeg 0s i p, 4.6 : le. ~p,)aacos i Vag = [az amin CO 0s } = 0025 From Figure 10.9, the flow regime is stratified wavy. Pressure Drop No line buried in the ground or under water is truly horizontal. The word “horizontal” simply means that the length is a much larger number than any elevation change. In such lines two-phase flow occurs uphill and downhill as well as horizon- tally, Some aspects of vertical flow are occurring simultaneously with said horizontal flow. The figure at left is a view of a near-hori- zontal line where the magnitude of the uphill and downhill portions has been obtained by survey. Unlike single-phase flow, each change in elevation between (1) and (2) affects pressure drop. 362 VOLUME 1: THE BASIC PRINCIPLESHORIZONTAL MULTIPHASE FLOW If the energy of the liquid entering a low spot is insufficient to carry it over the next “hill,” lig- uid will collect in the low spot as shown next. As this liquid builds up, the cross sectional area available for gas flow decreases. At some point there is sufficient pressure force to start the liquid moving. A portion is "gas-lifted" over the hill and the remainder slips back. The liquid flowing on contributes to liquid buildup in the next low spot. Then, in time, it surges. This process is repeated successively as liquid moves down the line. This represents an inefficient use of pressure energy. It takes energy to overcome liquid inertia and produce the momentum needed. Furthermore, the lifting process is relatively inefficient. Both contribute to excess pressure loss. For this reason, it is desirable to maintain sufficient flow rates to minimize liquid holdup. In addition, excessive liquid holdup results in larger liquid slugs at the outlet of the line. Both higher than expected pressure drop and slugs are noted in lines operating well below design capacity. As rate increases, both decrease. This is illustrated by the general figure at right. Pressure loss goes through a minimum and then in- creases as flow rate increases. The reasons include the liquid segregation discussed above. For the typical offshore installation where the initial and end-of-field-life flow rates may be low, “slug catchers” and other equipment must be designed for this worst case. The line must, of course, be sized for the maximum anticipated rate, but some consideration must Rea paaneae be given for lower rates so that the system is satisfactory under all conditions expected. Several correlations used for horizontal flow are shown below. The usual, intelligent approach is to make the calculation by sev- eral methods to ¢stablish a likely performance range. No one method is superior to all others for gen- eral usage. Pressure Drop One of these methods (Flanigan’s) is suitable for manual calculations and illustrates the relative role of some key variables. Modified Flanigan Correlation!”0-23) This work was done on field systems and has proven useful even though it is relatively simple. The relationship between gas flow rate, diameter and pressure drop is represented by the Panhandle A equation (Table 10.1). ‘Two corrections are made for two-phase flow. 1. The value of outlet pressure (P2) is adjusted for the pressure loss due to uphill and down- hill flow of two phases, including the effect of holdups. 2. The efficiency term (B) is correlated to reflect measured system performance based on gas velocity and liquid-gas ratio. CHAPTER 10 363FLOW OF FLUIDS As thus modified the Panhandle A equation becomes 1.07887 2 270-5394 de «(2 [F san) nar coon _ SL FPS Where: Gee = gas rate at Ty, Pye md scfiday P = absolute pressure kPa psia Py = pressure, standard conditions kPa psia Ty =" mean absolute temperature of line kK R T,; = temperature, standard conditions K R d= inside diameter of pipe m in L = pipe length m mi 7 = gas relative density E 7 Zm = mean compressibility factor S = Ey = two-phase efficiency K = 1.364 x 107 435.87 The term (P2 + AP2) is the outlet pressure of a truly horizontal line equivalent to that of the actual P>. The pressure correction for inclined flow is calculated from Equation 10.62 . AP, = A(pL Ey ZH, —Py ZHy) (10.62) ease FPS Where: AP, = additive correction to Panhandle A P, kPa psi A = constant 0.009 81 0.0069 pL = liquid density kg/m? Ibi P, = gas density kg/m? Ibe H = head m ft E, = empirical head factor (Figure 10.10) - - 4i, = uphill heads m Hg = downhill heads m ft The head terms are found from the line profile as shown previously. Figure 10.10 is a correlation to find the liquid head factor in Equation 10.62, At low velocity the head loss is high. Liquid collects in the low spots and is lifted periodically. This is inefficient. As velocity increases above 2 mis [6.6 ft/sec] this head loss decreases rapidly. Above about 8 m/s (26.2 fusec] there is little effect of velocity on head pressure loss, and in fact the flow regime will likely be entering the mist or annular mist region. 364 VOLUME 1: THE BASIC PRINCIPLES