XS. £ voll ize falc ss Mai iat Applying known laws of conservation | _ of encrgy gives good equations fox cal- alating gas flow distribution in mani- folds and branched piping systems Harbor, Long Island, New York branched piping systems, or rather of sizing piping ar- angoments to yield a desired distribution, particularly where low overall pressure drop is important, is one { }-5 2 frequently encountered in engineering design, Methods "(of correlating pressure losses in converging and diverging 2 flow through variously oriented T's, Y's, and elbows vary quite broadly and few appear to be fully aware of the Dasic studies which have nlrendy boon made or 10 appro- clate the significance of the related classic “blowing” and * “sucking” manifold phenomena, This field has received most attention in the mechanical enginecring literature, which it all too frequently forgotten or overlooked by those of other disciplines, Elbows. Applying the law of conservation of enargy for an incompresible uid flowing between two poinis in a + piping system (on éither side of an elbow) wherein pres- sure and velocity are constant across the pipe erass sec- tion one may write: ]- [az] \where total loss equals frictional loss in the inlet and out- let piping of the elbow (as though it were straightened nut as a continuous pipe) plus a so-called “mixing lois." ~For the elbow itself (particularly a short radius elbow), its aightoned Jengih is usually negligible with respect to the mixing loss so that H may be replaced by ph where ‘his the “mixing” loss. ; In considering an elbow 3 yy Rte ‘Tota bow, Hee [ ‘FIGURE 1—Vassonyi considered an elbow Tike this and wrote 4 momentum equation in the direction of the leaving fluid suining the prewure: 6a the wall—ab—equals the inital Frederick A. Zenx, Squires Intemational, Inc, Roslyn | ‘Tne rroaem of determining. flow distributions in ~ ich ‘as thown in Figure 1,” Prossure Drop itt a +! -Maszonyi wrote a momentum equation in the direetion 5 of the Jeaving Auld (assuming the pressure on tho wall ~ & th equate the inital pressure By and that the luld enters Eat an average angle 6) which is of the form: ee Py LEMS ora, PUEAL oy Sinco Vis Vary the term VitAy can be written VaViAs OW, VeAy #0 that equation (2) simplifies to: np Ne = oe ‘s) ort : ae ‘ : ae wee a ivan (ay From equation (1): F , maa oF oa he “ Equating (4) and (5), dividing through by p, and multi. plying each term by 2g yields: 2gh== V+ VP —2V, Vy 0018 © Because of the various assumptions made in deriving equation (6) one might consider adding an experimental correction factor giving 2ghe G, (V,8++ Vet —2V,¥, coro} When the angie « approaches zero the considered to act as a diffuscr with Tit resenting the-efficieney of represented by turbulence oc- in the passage. In such an instance, with B rep- the diffuser, the loss ean be Behe (IE) Ve) @ which may be combined with equation (7) 10 yield 2 re- lation applicable over the entire range of « from 0° to 90° (with Cy appropriately conforming), Adding. equations (7) and (8)? Bah = Gy (VyEE Vy = 2VVy cone) ch (LE) (WP —V Letting (1—E) = Gy, equation (9) simplifies to 2Gh= (C,+ GV (Gy— Gy) VP — 2GVV core 2gh=GVii++ 2O—O,)V,I—2CV,¥, cor@ (19) Applying equation (10) to experimental data for flow Ghrough circular elbows yiclds C= 06 and Cy = 1.05 for rectangular ducts C = 0.75 with Co again equaling unity, Vaszonyi illustrates the agrocment with experimen tal data covering values of « from 15° to90% Branches, Equation (10) may be empirically applied to pipe branches considered as two concentric elbows rotated 380° (ez, 2 oncxlimensional variably-angled y configu ation) by replacing « with an effective angle designated a’. By shutting off one branch of the y and allowing all the fluid to discharge through the other branch an em pineal function for G, may be determined experimet tally, ‘Then allowing’ # and V to vary, » trial and errorMINIMIZE MANIFOLD PRESSURE DROP» « - procedure may be followed to. arrive at a least equates Ft of C and a? as “smooth” functions of a, Using values so obtained for the ease of a 90° Tee of constant cross section with the straight ‘through branch shut off (c. fs Tee acting as an elbow) one finds that this Is equivar Tent to = 0.716 (sith G, = 1.0), as compared to G = 0.6 (and Cy’ = 1.0) determined for true elbows. By a procedure similar to that followed in deriving equation (10) Vaszonyi arrived at equation (11) for the oi 10} cs ost yo og! tn =e 04% ~ i: | 025 20 40 60 80 100 Lot x : rR 08 eS = Fa ov # o f Pie i : 20 40° 6080100 120140 160 z ole 2 a | oe | ost - 04 a r IN ‘0 2040 60 80 100 120. 140 160 FIGURE 2—Shows the branching flow coefficients used ja Frjuations (10) and (11). FIGURE 3—There are the definitions used in a “blowing” fold since all branchce discharge (oa common pressure Ps represents a fictions preaore Ios inthe hear Fepreseats 2 mining Tose (due 10 the flow di Toren auey Oa lin of the ld poe {agra hem he Uranch n) ples Say prenre Pin Panas re ecovery duc 10 the drop in’ velocity’ in the FEST IS fd eal a eee tes cata fe ef ames ee ccc pay ee eesere recovery of Jos case of converging or joining flow (eg, two straight pips ischarging into a third pipe). Wee vena, (% O 2% ayy ais & ‘The empirical relationships. for C, Oy, C’, a’, B%, and are shown in Figure 2, if Since by far the most common pipe branches in prac tice consist of 90° Tees and clbows, the appropriate con stants from Figure 2 have been combined with equation {1}, (10) und (11) to yield the working formulas sum- inatized in “Table J. ‘These formulas contain 8 multiply ing fav tor of 1.25 t0 allow for sutditional turbulence dic to entrance and exit effects when the lengths of lea Jipe are whutively short, This is based on recommendee tice for computing. straight piping pressure dep since the conventional Fanning Srietion facios are base en long Iengths where end effects are negligible, Jt is common practice to treat the loss in fittings it terms of either equivalent length of pipe or some numbe. of velocity heads. When the equivalent Jength is repories, asa function of pipe diameter then, aswining the Fan ping frietion factor as cewatially a constant, velocity hear and cquivaient Jengtis are divectly related by equatie (14). ‘Fhe Fanning equation mzy be expressed in term of velocity head as oP. cos Bh shy a= ae 05 X 10-4 pV 5,08 x 10-¢ avs (= Joss similanly as and fittin AP =K X 1.00 X 10-4 pV? an where K represents the number of velocity heads. Equat ing (12) and (18) yields the relation iL, K us where Lis now the cauivalent length of straight pip (having K velocity heads of pressure loss). Manifeids, The manifold problem is a speed cise ¢ al problem of predicting, pressure iw at flor distribution in branched pipiny systems, Because it repre tents a regular piping pattern and is a frequently ree ring problem, it has received gecater theoretical and imental study. Of particular interest are the so-calle Mylowing” manifold, the “sucking” manifolg, and the the een TADLE T—tquotlons for Prosture Loss In 90" Ply 3nnehes ary =e LAYS bra BENS Vy 0388 YN) argo 0029 HAHVY) OLY =O VV ] satan, eins PSpet ataemeemseis as) pecans soning enema (or frvrmannem egal 24 (soy, 48combination as frequently represented by panel oF harp ++ ype heat exchange surfaces. : "Bore § represent chemical the socalled "blow- ing” manifold, Since all branches discharge to a coinmon pressure Pp it should be possible to derive the flow diss tribution from each branch’ by’ caleslating the pressures at all points from equation (10) and considering charge Hows proportional to (Pp—Pp)™. As an. illus tration, consider a3 branch manifold with zero frictional loss in which branches and header are all of the same diameter. If V represents, the fluid velocity at the inlet to the header and a, by and ¢ represent the volume frac~ £2" sions ‘of the total fuid quantity’ leaving through the $ branches, then from equation (10) or Table J, using the nomenclature of Figure 8, the following equations may be written to define the pressures at all points: PYLPy = 185% 10-4p (1.82) F063 V (eV) : sy Pym Py 1.39 X 10-4 p [1.36 (V—avy— 0.68V8— 0.72 V(V—a¥)) i) (BP, 195 104» [L.8(WV)? —OEBBV(V— AVY] iss Ct an, weeny PY PY = 1.35 X 10-4 (1.86 (CV)? — : \ DIV — AV)? — 0.72 (Va) CV) «ay : Py 135 X 10+ p [1,8 (eV) #— 0,868 (2) 2) «sy With the discharge fow from each branch taken as proportional to the square root of the pressure drop: av) Py (PP) — Pp aie () [release Avy? Py (PP) Pp (@) sae cai} Substitution of equations (15) through (19) into (20) and (21) yields two equations of the form: = 0.000135 A («) (2) Taos Aeteey |) 2 = 0.000135 A# (a) oe where #() indigates a function only of the quantities fa parene thet and since a-Fb-Fe= 10» eH equations (22), (23),'and (24) can be solved simulta neously for a, b, ahd’ ¢ corresponding to various values, ‘of the constant parameter A, where: T= 0.000185 A# (ae ‘The results for the frictionless S-branch manifold are plotted in Figure 4. Figure 5 shows similar results for a frictionless 2-branch manifold under “sucking” as well fas “blowing” conditions. The equations for the sucking manifold were similarly derived from the basic equations of Figure 2 and Table 1. ores 4 and 5 illustrate the general observations of * flow distribution from untapered manifolds having litle resistance at each of the branches (as well as negligible resistance along the header), This distribution pattem 4s shown schematically in Figure 6, Tt emphasizes‘ the Peer ae eee on Flow of Totot Erection FIGURE 4~Shows the ealenlted fow dlitcbutlon from frictionless Shole blowing manifold FRACTION OF TOTAL FLOW o4 23 sq : 03-—a3| 3743 i 02 Lx 4138 \ ou 4320aot P(A) Blowing FIGURE G—These are dists atl Manifold terns for low pressure frie and coupled manifolds, to yield more uniform floiv branches connecting inlet and discharge headers) than combination (C) despite the fact that in (C) each path from x to y is of equal length. The numerical solutions ribution, (through the ranches ber for manifolds with increasing numbers of i lor comes extremely tedious and if friction along the hes $e alto taken into account the calculations become essen- tially hopeless without the aid of an electronic computer. ‘One approach has been to assume a distribution, calcu- te the resulting pressures, use these pressures to adjust the assumed flow, and then repeat the calculations itera- tively until hopefully after the third or fourth iteration convergence approaching the desired result is finally obtained, By considering a manifold model more ideally suited to analytical treatment, Acrivos, Babcock, and Pikford were able to obtain design curves similar (0 those of Fig tires 4 and 5 which in. addition accounted for header Friction. ‘This was possible by simpler approximations, than one would obtain from Varzonyis analysis, for the Josses at the branch ports, The mixing Joss in the header, fas the flow passes a take off port, was incorporated into the total pressure difference as a fraction of the pressure, recovery £0 that referring to Figure 3 the presure dif ferences Pray =~ Prayer were expressed a¢ Pa ke », V4 SZ [(Vau) Vion ‘Similarly the mean pressuée in the header at each port {was used in conjunction ‘with the orifice equation to de fine the pressure versus low relation through each port. ‘The mixing loss in the turn was thereby combined into the port orifice coefficient, ©,, in the expression 1) header area) aia) OK (Pont Panes p, Dm ‘The friction loses in the straight portions of the header ‘were based on the Fanning equation so that i 2p Vi)* @ ale Page Pa = where A} represents the distance between side branch ee eee the friction factor, f'., by @ relation- “()"s" Shoved Distribution Grid (26) 2a)! 1 | (Kl J (0) "U" Sheped Distribution Gri ship of the usual form K/Re®, relating all point veloc | ties as fractions of the header inlet velocity, and rearran; | Jing the variables, pemmitted solution of (36), (27), a1 (28) in terms of 3 variables I, Al, and (g/kA) where is defined by equation (25) and k Ty eqjutan (26). fucther srmpliiration of Che manifai: meiei was 60 caved. by considering the = =7ing herwerrs the brane Connecti¢ns as very small (approaching zero) and Gruner of braneh conn rtians in any increment of Ia Temgith as very large, This i equivalent to considesing 1 header as having, @ fongcitwdinal slot of constant width rather of simply being ans tube, Though this anit nay be questionable for the case of a header nuinber of side branches it has the: virtue of sim the three precious equations into a single dll czquation in terms of F and (g/kA) only, (since al» ‘As the mumber of side branches increases the fractior flow leaving at ench branch decreases so that the sits) fieatian in equations (26) and (27) approaches the m Figorons formulas of Vaszonyi. Equating (26) and (! to the first and second equations of Table 1 and solv for k and Cy, at various vatios of V’zy/V'ay-s yieked values shown in Table 2 in comparison to the const D6 values used by Actives, Babrock, and Pisford. Leting With Vessenyl's Tquetiene and the Court APDr AADtt 2—Comperlsen Betwean Sida Port Loss Coetclents Teettens Used by Revives ot #l ‘The computer solutions of the differential eque for the continuous slot or porous “header are shaw Figures 7 and @ for “blowing” and “sucking” ew respectively. in comparison to the previous caleult hummarized in Figures 4 and 5, ‘The results are pl with curves of the ratio of part arca to header’* fuperimposed; these are all based .en the pores header model and should only be used as a guide in r fold design since solution of the more tigorous e405 dicates that a header with discrete ports appre: the continuous or porous tube model only when the10 Tacegtactat Tapia TTT] Los | zg, og > yer le oe a x i oe PO" | naoset £22 — re é Sate . rss a OE 4": fo 3” eet 4- 3 LB 3 4 BL Be | Lae 4 ag y | Le? 14 ee eee —_ log Fes/isec: - S|, gy ‘ahere f= fanning friction [|e factor bored on fe ct el tet Ny booder entronce bad | & Ly ee ndt/4pi Y | 5 oe 3 10 10® ot s |] wae Contimous Slot road : Approximation of | e? : hcnves. ol ol ' + & s = Based on Voszonyt’s | [.& & Equotions (See Fiwres 5B ; 4 ond 5) a a & dt. o ze & port donor] pessoas 2 3 |5- oe = ara & ‘i a, 3S 2” _ a 8 & Hott ot 2: v a, oe tt J "eta Heater af Garton Diometer,D . ieee fan esbeEe La her of poris reaches 10 er more, Though Figure 7 sppeas to indicate that the 3 port case already approaches the ‘continuous model, it must be realized that there is a dif- ference between the rigorous (Vaszonyi) and equivalent (06) mixing loss coefficients usod in calculating the dashed and solid curves of Figures 7 and 8 Semple Calcvietion, To illustrate the use of Figures 7 oF 8 in estimating manifold flow distribution or cone { versely in specifying suitable dimensions, consider the dis- tribution of 39 clm of air entering a 7 foot long 3 inch 3D header provided with 7 equally spaced 1 inch ID ports ere era y ie 3 from a Fanning friction factor plots £ = 0.0076 nd? TARY 4 (3712y eee . o.076 ~ JF 3re * “rae Koons 708! Port ‘Area/Header Area = + na? i pre“AINIMIZE MANIFOLD PRESSURE DROP . 10 T T wv Vaz Figure 7. into First Port “typ ee eee For Nomenclature See Maximum Flow Always Ol ae yh —— CONTINUOUS SLOT 4 APPROXIMATION OF ~ " bE AGRIVOS et ai 4 ve eoP —==BASED ON VASZONY!'S 4 EQUATIONS (See Fig 5) PV/(P\-Po) i ool bo cease Eee peace psa 100 1000 10000 FIGURE 8—Use this chart for ealslting the flow disrbution Sato subking manifolds. : eve ‘or 10.5 perecht more flow out ‘of the seventh hole thar + from Figure 7: —p po = 2250 ‘out of the first hole. . “The results of similar caloulations, varying the mum 0.076,“ 30 X-4.X 144 ber and size of the ports as well as the volumetric air rat’ BP peg (ARS 0.00852 pal and header diameter, are summarized in Table 3+to illus also ste norte] eager spe] si eee egg pee geY ge Estreive Poriay ‘Seren trate the relative influences of these variables, Units contintent with Tables and Figureu rt diameter, Inches Neader anaing 1585 der Jongih between end pos change premure, pri A presure, pai assure drop, pal Jolumeltic Flow rate, egy CFM. ty, flsee Padeechy B. Dy Figlerd, Ru Lo, Chemica Englcesing Seine NOMENCLATURE meter, Inches jon factor feet Besfeuste DIBLIOGRAPHY heat ttataa- ant ceeis tt SS II HEAT TRANSMISSION TT HT! Tort i ae i itr iI TIT UII nT py Ten | { ia! TT TAT HOT LUI rer TEA ET TU HT‘o1r9 Tay uy umoys ox0 9 pee “4 "ayo wontva ou, ‘oamersad aryouosva wai a a WATS TH SEE OOR Bak ssoid spay dooms yorsson pu ce rmranjout 930.009 Su, SUL "$1000 = 2/(cz00'0+ s209"0) -n200 4 50 4 6¢ Bd 9943 “S,081 1 TH1CO90'O PW JOOS 9 $6 10 Ayso0e.4 Sy ipsaday 9 wos “@r-9) (96-9) “