5,PIP!NG
HANDBOOK
PIPING HANDBOOK
TABLE
OF
CONTENTSSTRESS ANALYSIS
A Simplified Computer Program For Pipe
Piping Design Method Beats Computer
Symmetrical Piping Arrangement Solves
Expansion Loop Design
Two-Phase
Page No.
Flow Distribution Problems
LAYOUT
Plant Layout And Piping Design For Minimum Cost Systems
What lnformation ls Essential For Good Piping Design
How To Design Yard
Piping
.
.Locate Tower Nozzles Quickly
Piping Of Pressure Relieving Devices
MATERIALS
U.S. vs British And
Which Material For
.
.139 European Piping SpecificationsProcess Plant Piping?
4 5 10 20 24 25 34 39 49 59 66 67 73 80 81 88 94 96 EXPANSION
Find Best Pipe Expansion Loop Quickly
Expansion Joints: How To Select And Maintain Them
Spring Pipe Hanger Design Simplified
Piping Tierod Design Made Simple
THERMOWELLS
Procedures For The Piping Designer lnstallation And Soecifications
.,
. . lnstallation And Specifications.,
. . .Selection Of Thermowell lnsertion Lengths
PRESSURE DROP AND VIBRATION Simplified Utility Loop Balancing . .
Piping Design Stops Pulsation Flow
Find Line Pressure Drop By Nomograph
New Approach To Pipe Reactions
STEAM TRACING .
New Guide To Steam Tracing Design
100 101 105 111 120 121 125 131 133 138
A
Simplified
Compufer
Program
for
Pipe
Expansion
Loop
Design
Using
a
single
input
card
with
terms
and
measurements
common
to
any drafting
room,
computer
computation and
analysis
has
5:1
time
advantage over manual
methods
W. W. Shul!, and G.
N.
Bogel, Jr.The Dow Chemical Co.. Ilouston
A
conputnn pRoGRAM has been written thatwill
de-sign piping expansion loops with less cost and with less
elapsed time than ,existing rnanual methods.
It
requires asingle
input
card containing measurements common to any drafting room.Mqnuol
l,lethod.
A
previous
articlel recornurends amanual method for piping loop design except for critical lines (those with high or low operating temperatures and pressures andfor force sensitive connected equipment)
and except for piping which conveys hazardous or flam-nrable materials. These exclusions eliminate application of the author's semi-graphical methodl from most of the
piping in a petrochemical plant and require that the pip-ing designer both understand and have access to process
data. Even
if
process data were available, the hoursre-quired
to
segregate the hazardous senice and critical piping for solution by a stress consultant or specialist andthe queue hours for the attention of the stress consultant
or
specialist areadditional
factors which the authors omitted.The authors' make the claim that,
for
non-excludedpiping, their graphical approximations are obtained
in
afraction of the time
it
takes to prepare the data for com-puter analysis. Readers will have to guess at theengineer-ing and clock times used by the authors
in
arriving attheir conclusions.
In
ourtrial
runs, assemblingof
datatook the same time and computation and analysis of re-sults from computer and hand calculations had a five to
one time advantage in favor of the computer, besides hav-ing the ability to better understand the effect of the vari-ous alternates.
It
is probable that the computer programs available to the authorst did have a complicated input for solution to simple problems; and considering only over-alltime lapse, their claim is often true for an answer to any
one single problem which happens to fit a manual method
which the designer uses with sufficient frequency to main-tain his speed and accuracy.
Effective use of computer programs also requires user
familiarity with their input requirements and their output capability. This learning eflort is less than the eflort of
learning any equivalent manual method. Moreover. one
rnan can examine pipe layouts for a plant, input the loops
on a computer program input form, and get answers for
one
to
several hundred problemswithin
24 hours, andin some offices within the hour. Well planned piping
de-sign jobs seldom need answers faster than 24 hours; but,
under planned conditions, the computer results can be
rnade available within minutes.
Resislonce ?o Computer Use.
It
has been ourexperi-ence that people who resist using the computer as a tool
for the routine as well as the difficult tasks in design may
be described
in
two groups. The first group thinks that computer answers are either beyond questionor
are not subject to parametric study of input variables ormodifi-cation based
on
engineering judgment. This group hasA
SIMPLIFIED COMPUTER PROGRAM FOR. .
.S AND B ENGINEERTNG SEPVICES PIPE LOOP OESIGN PROCRAX {AINLINE PIPE LOOP DESIO -_sINGLE PLANE'2 TO 6 PIPE IEXAERS
rITH I TO 5 ELAOiS OR BENOS OF SAXE BENO RADIUS IRITTEN BY T.I.SHULL ANO G.N.AOGEL' HOUSTON TEIASI ilAY T967' PRL]GRA{ USFS FORNULAS ANO T BLES PPES€ilTEO IN TIPIPING OESIGN AND
ENGINEERINGI' THE GRINNELL COXPANYIINC't PFOVTO€NCEI RHOOE ISLANDT usA. SECOND EOtTION. 196f,.
ALL SUBS FILOI PROCEOUPE AS DETAIL€O ON PAGES 52t5f OF CRINNELL INPUT STAFTING IITH NY XEXBER AND €NDING IITH ANY f,EiAER' EACH XEXAEF HAS T'O JOINIS (TELD€O) IHICH ARE ANALYZED FOR IHE LO.EST ANO HIGHESI JOINI NUXBERS ARE CONSIDEREO IO tsE THE LOCATION OF IHE TNNER{OST PAIR OF GUIDES OR THE LOCAION OF
ANCHORS IF IHE lNPUT LENCIH TOUISIDE OF GUIOES' I 5 ZEKO. iHE EXCEPTTON IO THIS RULE TS FOR TIO XETBEF ISINGLE ELBOI' JOINIS FOR IBICH THE PROGRAI TILL INCREIENT ONE f,EMBERS LENCTH TO PROVIOE THE NECESSARY FLEXTBILTTY'
c.rf,xoN o(6) t6(6).vl 6),x( r2 );Y( r2l.xL( r2r,r( l2l,rxl l2ltrY(12,,R(12) r.s( r2t.Rxt t2),sT( t2), AAI 20 )
COTXON SIPTPI. TLA IXKTSAETAI XBARtYtsAR! XLP!YLPTPXYTP I XTPI Yt IMAX ICT' IFX! FY t CONSIX!CONSIYTOA'THX !FST{AXTDEMT ILEN!ALTsI
COHXON TLNCTXOFY.OLNG
coxf, oN tEx(5, t TRY( 5),BXl 5l.AY( 5),OXl 5l rAST (5' rBR'srtAXrKBD COIXON KPFI TKSTI(FNDIKFSI!KLOOPt J!K'N.KXAi
OOUBLE PRECISION PIPEIO
t.JI = OST TO.ENTROIO X OIRECTION Y(J) = DST to CENTROID Y DIRECTIoN II IJI: LFNGIB OF SEGXENI R(J) = X DIST FROY CENTFOIO 5(J) = Y OISI FROB CENTROIO PXY = PROO OF INIER I A
PIX = COU OF IdTERIA ABOUI X
PtY = xof, oF INIERIA ARoUT Y
CT = ExPANSION AT TEHP
XLGTYLG = t ANO Y DIST FND FOINT TO END POINT PI = P1PE IOXENI OF TNTERIA
Sf,P = PIPF SECT TON XODULUS
CENTROIO OF STFAIGHT LtNE IN PLANE OF PROJECTION - COLN A = LENGTH B= x OR Y DISI EoN r
COLNlarB, = A*A
( CENIRDTO OF BEND XK=FLEx.FACT'R=RADIUS OF aEND!X=DI5T TO AXISE0NS c coaNo(xKrRrxt= t.57axxrx aR
C PROD. OF INIEFIA LINE PARALLEL TO AN AXIS EON6 XL=LENGTH C PLNXY(xL!x.Yr= xL t r: Y
< PROO OF INTENIA AEND NEG IHEN AXIS I RADIAL OR ARC
( PENON(xKrR.x!Yl= xK +.lf7rFaRrR. xKI r.57 +R +x' Y
C PROD OF INTERIA EEND POS iHEN AXIS '
RADTAL AND AFC ( pBNOp(XK!R,r,y)= XXr.r37*R+R*R + iKal.57tR*XrY
C MOMENI OF INTERIA ST L'NE tN PLANE OF PROJ PARALLEL TO AXIS X
( PLMILNIxLTY) = XL aY:Y
C MOU OF INI SI LTNE PERPENDICULAR TO AXTS Y
( ppxtLN(iL,xl = iL*xllxL/I2.. XL+X1X C IDB OF tNT BEND FROM AItS Y
< p!taN (xxrRrxt = xK +to.r49aRrR+Rl. l.s7! xK tR tx.x C PIPE SECIION NOOULUs DA = ACTUAL OTA. THK= THICXNESS IALL
s!( oA. rHKr sxx )=0. 25loarsMx*( DA+oa_2.4( DA.THK T+2.a(rHK.THKl ) . LAMBAOA XLAX F = RADIUS OF PIPE AEND
iL^{l THK. R. DA)=THK+R/l ( cA_rHK I *(DA-THxl +.2s) C FLFX. FA'TDA XLA = LAMNADA
FF ( xL A l= I .65/ xL A
C STRESS INIENSIFICAIION FACTORT AETAS' FOF IELDED ELBOIS OR BENDS' EEIAs( xLA l=0. 9/xLA*t O. 6667
C OIPE CROSS 5ECTION!L ICTAL AR-A PAREAI DATIH('=3. I4 I 6IIHK:(OA_TBK I c plpE ioH oF lNr
PMINTI OA'THK I=3. L 4I6*I DA-IHK)'+O3TIHK/4. C XBAR DISTANCE
C D l5 DISI v= vAFIAELE B=tsEND RAOIUS
C XPRI--A NON-ZEEN ENIPY IN CDL 7A PRINTS VALUES TN CO{{ON C LPRT--NON-ZERO ENIRY IN CI]L. 79 SHOYS LAST OF PIPE LODP5 USING C HEADER CAFD ANO PR'GRAil TILL EXPECT A NEt HEAOER CARD' C NPPI_-PFINIS STRESS A5 PtOF IEilBER LENGTH IS VARIED FOR OESIGN C PROALEqS. ENIER IN COL. AO DIGIT 1 FOP STNESS EACB.5 FEFIT
C DIGIT 2 FOR STRE5S EACH I. FEEII
C DIGTT 3 FOR SIREsS EACH I.5 FEEI! ETC'
C 6UIOE INOICAIES OI5IAN'E IN X OR Y DIRECTION OUISIDE THE SECIION A AO
C OEFINED FY Ol-D6 . o=x l=Y DIHECTIoNS x=Ol 'D3'D5
C T€5I DAIA FROM GFINNELL SIJCOlO EDITION PAGE 52. 1CT FROM PAGE 9 A 72 c A6 F5.O Fs.O Fs.O F5.O Fs.O Fs.O F5.O F5.O F5.O F5.O F5.O rtF4.IIF4-IlF4. JI1 C tD DIA TH( ALiST CT tsND D DT D2 03 D5 06 GUIOEVARYLVARYA PRT cP.52R1O.75.50 17675 996.50, {0. 24. 12. 14.
cP.5201o.75.50 17675 996.50.40. lO. 12. A. 2 lO.l4oOO CP.52RGRINNELL ANsiER5: ANCHoRS x=2795rY:lA67LB5. END HOMFNIS=24f64'-23970 FT-LB C FOR PROBLE{ P.52: MAi.HE'TD STRESS EET. JOINI 2
' 3 = 9O3OPS] t IOA'=29672 C {ANUAL-GRAPHICAL: 6REATESI STRAIcHT PtPE tS AI JOINI 2 = a7l4 Pst
c INEPIIA oF PIPE=212. !BAR=3a.63!YBAR=9.6oFT.
C sEcTloN MOOULUS=39.43 lxY=lO4s7. lx=941s. lY=21i69'
C
'OHPUTER RESULTS: XAX.BEND STRESS RET.JOINTS 2 A 3 = 9IA2 PSI C FOR PAOBLEM P.52: GREATEST JOINI SIREsS IS AT JOTNT 2 = A723 PSI
c INERTIA OF PIPE=2\?. XBAR=34.62'YBAR=9.59 Fr.
C sEaTloN HoDULUS=19.43 tXY=I0446. lX=9395. lY=2135'.
c 105 c I lo ttt at2 ttJ tr5 C I l6 c rlt I to too lor to! lo:l to2z IOal I O2a I 0?5 lo2, t 0t I O7 l06 ]L 32 a, t20 121 t22 c 23 2a 26 3t J2 ll fa 4a 5l 52 55 57 5A 65
little
knowledgeof
the difficultyof
progran'ringa
com-puter to satisfy every possible need or whim of the user.
Its members reject the computer as a tool when the pro-gram's logic requires flexibility
in
the
user's notion ofinput and output content.
The second group has some experience rvith either
pro-graming
or
using the computer; yet, its membersover-6
tla
l4i
tro
estimate the
difficultyof
communication
r.vith the computer. Both groups deter the economical emplo1'msn1of
computers and thus restrain the developmentof
the engineering sciences from the precomputer art ofapply-ing empirical relationships
to
approximate solutions for-nvhich exact theoretical solutions exist. Fortunateln mem-bership in either of these groups need not be perrnanent!
(ONT RL:2 rRITE (f,r r r5) lor RE-aD (rtl29, aA
5J
to2 icITE (r, r30l NPACETAA DO tOr J=rt 132 oo tor J=r53r2ll lOa AlJl=0.0
READ I tr lf I I PIPEIO,DA!IHKTALIST.CT tAFAOtD!kFSr!FSTMAx.<SNDtsNDilAx I TKDIRI OLNGT(PRIILPRTI NPRT
YFIIE (3! 1f2 I
IRITE (f, I IfJ} PIPEIDIOA'THXTALISTTCTTAFAOTDTKFSI'FSITAXtXSND.
Fig.
l-Pipe
expansion loop program. Note: Subroutines not included in the program listing solve step-by-step thE manual method illustrated in Lit. Cited 2, pages 52 and 53, using the functions duplicated as commentT
Pf,OBLE X I D El,lTlF lC AT lOil .?a,aat+.cnt, ,tF€ ,1/tC, ,4 , t , , t LOOP lDEl'lTlFlCA-ilot I 6 ACTITAL 0wsl0E PIPE PIPE
rlU_ ALLOY-IELE sTncss (PSt, fr 2 c FACIOff AT OEB. E D ,t Itotu3 OF PIPE BEilD FIBST LOOP sE00ilD r.ooP TAIND LOP SEEIETT LEXGrll ITEEI) g, at FOUNil LOOP FIFTII LUn strTltLooP .= i= Er a-*E t ltax. L DE lBu I ax- IETL. IETGTII IFEET} ara LEUOTN IFEETI P al IErt6TE (FEETI t 5 LEXGTH (FEEN o tt LEI'IGIH IFEEN o A BF /0|'lD iIJIDE DITUETER lrilcHEsl 7 Ir[cr{Esl 2 1 ltnurEJ, llIERs-01 t7 !i e
TYPICAL LOOP LAYOUT START AT ANY MEMBER,AND STOP
AT ANY MEMBER TO DEFINE LOOP
GEOMETRY - EETWEEN ANCHORS OR GUIOES
Fig. 2-lnput card with data for problem shown on Fig. 5.
The Progrom.
A
computer Programin
FORTRANIV
was written (Fig. 1), just to see
if it
could be done, tosolve expansion Ioop design based on a published manual method.2 The program does
not
require manualinter-'-'i1"
allow
solve
upon consultants.
If
the loop as input has too great ananchor force or too great a stress, the program provides the required flexibitity by increasing the length of a
desig-nated member of the loop within user input limits. The Iocations
of
guidesnormally
arenot
altered becauseguides usually are mounted on supports whose locations are subject to controls other than piping flexibility.
Input Data. The entries for an individual piping loop
r"qri."
a single card (Fig.2-80
lettersor
digitsmaxi-mum) containing the following information:
.
PiPe loop identification name or number.o
Outside diameter of the pipe and the wall thicknessl;:n"r)
. Thereare no limitations on diameter or
thick-o
An allowable stress range at bend and terminal joints based upon codeor
connected
equipment limitation(pounds/square inch).
o
A value of C obtained from the following equation:^
Expansion, in.,/100ft.
(E")r-@
or from Literature Cited 2, page 9. This value combines
the lineal therrnal expansion
in
inches from 70o F to theoperating temperature with the tensile modrrlus
of
elas-ticity at 70o
F
(pounds/inch).o
Radiiof
bends(inches)-a single
bend radius isapplied to all bends.
o
Lengths between corners of the loop members within the innermostpair
of
guides (feet).A
"corner"is
de-fined as the intersection
of
the center line extensions of straight pipe members (however short) which connect toa 90" bend or square elbow. The input length must be
not less than the sum of bend
radii
(expressedin
feet) incorporatedin
the particular member.o
The number of a member whose length the program allel the
ex-eto
hold anchors and move guides.
o
The maximum permitted anchor force (pounds) andthe
X
orY
direction of this limitation.o
The piping length (most pair of guides and
axis
of
this
additionalcontributes force because of thermal expansion.
An
additional header card precedes a grouPof
pipeloop entries. This card contains any desired alphabetic or numeric descriptive data desired
in
the output printout(job title, user's name, accounting information, etc.).
Whol
The Progrom Solves. The program is capableof expansion to handle many possibilities
of
piping loop design but implemented for PurPoses of thisdocumelta-tion
for
the following frequently used single-plane looptypes which have no external loads or restraints between anchors except guides:
o
U-shaped loop with unequalor
equal legs plus upto
trrzo tangent membersof
unequal or equal
length.(Terms correspond
with
Literature Cited2
terms forU-shaped examples.)
o U-bend expansion loops similar
to
above. IJse our term "legs" instead of the Literature Cited 2 term "tan-gents" for the expansion U-bend examplesin
LiteratureCited 2.
o
Simple two member loop with one elbow.o
Z-shaped expansion loop.o
Hooked Z-shaped with upto
one tangent rnember.Progrom Tesis. The program has been tested with book
problems' (Fig.
3)
andwith
the
examplesfrom
the7
PIPE LDOP FlPE IALL ALLOI FACTOR C RADIUS LENGTHS OF ME{AERS (TO CENTER OF CORNERSI D- IHICH ANCHOR FORCE LIXIT , rnFNt- n D THtak- qTRFsS a nFG-F oF FFNn Dl n2 Df nd n< nA xaY vAeV .=Nn- I -Y ,=v nlo
rNcHEs l-lellEs -_e5!_
--__--- !!!tsES- -EEEI- -EEEI- -EEqf- -EEEI- -EEEI- -EEE.M: EEE!
qlqE PouNDs FoRcE
75O-..._o <nn -E6b> SE qI_rnrNT . )>
--E-6330. A020. 63A0, 962- 7426.
JOINT 2 HAS GREATEST STRESS ANcHoR F0RCE IN X-DIREcTION = 2a03. ANcHoR FoRCF IN Y-DIRFCTION= 147O. LENGTH OF sUM DF MEABERS
--
F' rrrn qtrH--Dl=D6 LENG.IHS
-FEET FEFT FEET AX I S
COORDINATE OFFSET FEET MAXIMUM STRESS IS IN BEND LENGTH OUTSIOE AXIS DISTANCE .ll-uI_1,f.-_THRU .r I AErEF{ rpiHiB
' ^Hp r I O,|I X_AXIS Y-AXIS IN POUNDS/SO. INCH
MOMENT AT JOINT I = DIAGRAM OF PTPE LOOP MO{ENT AT JoINT a =
L4LOz.-_J5_{EXBEF "I_J6_ rl J7 ll - --L-Lq-ottrrrr -.+ | o-ooo ffEL MFMBER D2 MEMEER O4 | _J9__MEMRER D5__J r o_ | LOOP O. K. FROM STRESS STANDPOINT
Fig. 3-Test problem based on book2 input data.
PIPE LOOP PIPE TALL I|.OI FACTOK C PAOIUS LENGTHS OF MEMBERS (TO CFNTER OF CORNERS) D- IHICH ANCHOR FORCE LIf,IT
lrlclEs rNe,BES __PS!_ !NS!ES- -EEE!- -EEEI- -EEEI- -EEEI- -EEEI- -EEE! !S! EEEI qgOE eQ!.Nqs FoRcE j.o9B lm lMo-a-@lo4Lp4
7t96. l6la7. l30aa. 2154f,. 24642. I tq6. o- o.
JOINT 5 HA5 GREAIEST STRESS ANCHOR FORCE IN X_DIRECTI ON - I2N6. ANCHOR FORCE IN Y-DIRECTION= LENGTH OF SUM OF MEMBERS COORDINATE OFFSET FEET MAXTMUM STRESS TS IN EEND LENGTH OUISIOE AXIS OISTANCE
FtLF PATH. -D.!=s. LENGTHS
-JONI J+-THPU J I
FEET X_AXIS Y_AXT5
1NO 7 !@
IN POUNDS/SO.INCH FEET AX I S
MOMENT AT JOINI. DtaGRAM 0F PIPE LotP MOMENT AT JOINT IO =
+ leooo rEL l. ro.o@ Er MEMtsER D2 VEMEEi D4 / l2.ooo FEET J3 I - - - -- -A€G-INNIN5-:]Jl---::4EUBER-oJ:=:-:r+=l- -- . . l--cuIDE = | J' I o.o FEET EtBU N-E I I F#T- -- JrL=-I J9____MEMBER D5___JrO_l THIS I OIIP OVERSTRESSEO
Fig.
4-1"r,
problem based on Haque/Starczewskil input data.PIPE LOOP PIPE TALL ALLOI FACTOR C RAOIUS LENGTHS OF MEMBERS (TO CENTER OF CoRNERSI D- THICH ANcHoR FoRcE LIHTT
J!-C!ES I-NCHE5 --eS.!.- --- !!g!ES- -E-EE.L- -ECtrI- -EEEI- LEEEI- -EEEJ- -EEEL lsi EEEI ggqE EgUNgS-FoRcE
t!Pu' OgL6.6fs O.'r0 -'5Oo. 35s.0 d.ooa r2.OO A.^0 a.OO a.OO 12.^O O.O 2 4.5O t e2OO.e rtrH w^ovrNc xFMqFa I Fil.TH = a-6 FFEY. eToFss I< r.:Ao- I aq/ea IU - aM.Hno FnorFc y-6ro = ,^A9.. w-DIp - -O IITH VARYING HE{BER LENGTH = a.s FEET' STRESS IS 30O79. LBS/SO.IN.r ANCHOR FORCES X-DIR.= la3l.. y-DlR.=
-O. YIT{ vAqvINA {EMBS I ENGrts = o O FFFr- STREq( rq 'a^^1- LA<'sO.IN.- ^Ncpno FnR.FS y_DIor_
'5ra.,. v_DtF._ _o. IITH VARYING MEMBER LENGTB = 9.5 FEETr STRESS IS 26276. LAS/SO.IN.. ANCHOR FORCES X-DIR.= 1434.. Y-DIR.= -O. !u+wYl!r: l.MHFo LE.Mg - rO- s| 3lS2-
' pq"qr- : 12a6.€
IITH VARYING xEHtsER LENGTH = lO.s FEET. STRESS IS 23251. LBS/so.IN.r ANCHoR FORCES X-DIR.= lr57.r y-DIR.= -O.
-u Jl! JP 6473. 1456t. 12044. 19443. 21960, 21960. 1944a. 12044. 14561. 6473. o. o.
JOINT 6 HAS GREATEST STRESS ANCHOR FORCE IN X-OIRECTION = I045. ANCHOR FORCE IN Y-DIRECTIoN= LENGTH OF SUM OF MEMDERS COORDINAIE OFFSET FEET MAXTMUN STRESS IS IN BENO LENGTH OUTSIOE AXIS DISTANCE EL'rlo ptIB ot_DA til.-rs< Jot{ItL t THRU sl!'s 6 N0 , LOOp
FEET X-AXIS Y-AXTS IN POUNDS/SO. INCH FEET FEET AX T S
,rotu__
-^r.oo
UOVENT AT JOINT I - DIAGRAM OF PIPE LOOP MOMENT AT
aE. FOOr-POTJMIS Ar5- JOINT IO =
I I JI I ! l.ooo Ef MEMBER D4 ,l a tz.ooo FEEr J3 I ar u[Ytrac_:_I__---8ff8tre ^t )2_l CUTDE I I o. o FEET I
tt
l2.Qio FFEr Jll | _J9_____MEMBER D5___J I O_ | LOOP O. K. FRO! 5TRESS STANDPOTNTFig. S-Test problem based on Haque/Starczewskil input data.
About Ihe qa.rthors
'Wrr,lr,lrr \1r. Snur-r, is u, r,iril cnginecr utith tlte Cor.poro,ta Enginer:.ring antl
C onstruc tion Sertices of The Dou
Cltenticol Co., HotLstott. He cletelops
qualitA tnctltods of ck:sign problem
sctlu-tions f or ci.-il, ntcclta.nicrLl, an.cl :--esscl
disciplines. Mr. Sltull holtls a, B.S.
cle-gree in nkc/l.anisa1 engineering front Lottisionct Stote Uni.uersitu rtnd, Itcts
don.e gradu.ote toork at Tulane, tlLe Lini-uersity of Ho.ttston, and Georgle
lVaslt-ingtotr Uniuersity. He lm,s pS ,LJ(.al s
c:t:-G. N. Bocrr,, Jn. is a seuiol" engineer
in the Corporate Engincering and
Con-struction Seruices of The Dous Chemicol Co., Houston. His utork inrolr-es
chenti-cal engineering, systems dnalAsis f or
effecti"-e use of coinpllters in design
engineering, liaison f or lrcat erclTan!.ler tlesign practices and metltods witlt Heo,t Transf er Researclt, Inc. Mr. Bogel ltolcls
a B.S. degree in chetnical enoineeri,ng
L
-J|I
from Teras A.&M. Uni"*ersity. He h,nswollced Lil.ptocess det:eloptnent, pilot ancl production plant supert:ision, pyocess and mccltanicctl design, and, project
en-gineering. He is a metnber of AIChE, ACS, ACM, TSPE, NSPE, Process Heat Exchu,nger Society, rntd the National
S ociety f or O ceo,nogro,pltu.
Haquc/Starcze$.slii article (Figs.
,l
ancl5). In
general,the agreement rvith answers gir-en b,v our program asree
u.ith the problems and examples
\rithin +
2
perccnt.Extrerle care \,vas necessary
to
make sure the problemsr'r'ere the same and that the results u,ere identifiable. For
the exarnple given
at
the botton'rof
Page 204in
the I{aqLrc,/Starczervski article,our
stress using long radiusbencls and con-rputed intensification factor is in agreement,
but
onr anchor force is lorver than tlte square cornersolution bv
a
factorof
tu o. The anchor forcefor
our square corner solutionof
the sartre probleln agrees r.viththe force from the computer program mentioned
in
theHaque/Starczeu,ski article and is nine percent lo',ver than
their graphical method.
Square Corner Technique. The use by inclustry of the
square corner technique
with its
conselvative answers(sometines by
a
factorof
twoor
more) is difficult tounderstand because
it
tends to indicate problems rvherethere are none and results in tvasting investment and
oper-ating capital for unnecessary expansion loops.
Note: The mainline program (Fig.
1)
with its sanple output (Fig.3)
represents original u,.ork and is in no caseendorsed by our emplo),er or based on any program cllr-rently
in
usein
our ofTice.1Ha.gu9,
)'-n",Jll?"r",t*n r\,rethod Bcarscom-prrrer.'' FI; 16. No. -3. March l9ti7.
- 2 "Pipinq serond edirion, 1963. Grinncll Co., Providence,
Index!.ng Terms: Coefficients-6, Computers-10, Design-8, Diameter-6.
Expan-sions-6, L^engrh-6, Loops-9, Pipine-9. program,-i0, Radiui-6, Sr.*.F.-i, ,:l
"(
perience in tlrc Dotu Engirtcering Dr:pttttltt.emt in ci.-il ancl mcchanicctl engineering dr:si11tt. Hc is q, nttutttr:r of ASCE NSPZ', and ACI.
Piping
Design
Method
Beats
Computer
Symmetrical,
U-type
PiPing
looPs
can
be
analyzed
for
flexibilitY
using
this
new graPhical
method.
lt
can be done
faster
than
the
time
required
to
prePare
the data
for
computer
analysis
M.
S. Hoque, Engineering Consultant, EMMCON,London and
J.
Storczewski, Woodall Duckham, Ltd',Crawley, England
ple easy
to
follow method is used and no extr'aordinary mathematical knowledge is required.In
structural design, the imposed force on the systemis specified and the
deflection
pipingdesign, the deflection is
given
by theimposed deflection is
determi
rigor-ous methods available
which
accu-racy, but their involved computational intricacies demand
the attention of an engineering mathematician or an
ex-perienced pipe stress analYst.
A
less time consuming, easy to use method is the only answer. One that would even win over the total time in-volved to solve a problem on a comPuter, and at the sariretime produce acceptable results and also satisfy the re-quirements
of
the Codefor
Pressure Piping ASA 831.1or British Standard Specification 3351-1961.
The authors present a method which can be used for
to
the solution'of
'lJ'
type symmetrical expansion loops.It
procluces results reasonably accuratein
almost a fractionof the time
it
takes to Prepare thedata
for computeranalysis.
The total deflection in a piping system is usually known' For example,
if
the pipe length of a symmetrical loop is50 ft., i.e.
'U'
length betrveen anchor points, the thermalexpansion
4
inch per 100ft.
of
lincar length, then thetotal deflection
:
50/100(4):
2 inches. The height andwidth of the loop are generally detennined by the space
available. Oncc the shape
of
the loop is decided by thelayout engineer, forces, moments, and stresses can easily
be found using the graphs presented in this article.
It
is suggested that the useof
precise and analytical methods and also comPuter analysis shor.rld be limited toonly critical and hazardous lines. Critical lines are those
involved
with
highor
low operating temperatures andpressures and/or the type of equiprnent to which they are Connected. Hazardous lines are those concerned l'r'ith the
nature
of
fluid being conveyed,highlf
inflammable etc.Therefore, designers should segregate all critical and
haz-ardous lines rvhich demand
the
attentionof a
pipingstress
consultant or
specialist. For simpler, noncritical lines an approxirnate calculation method is permitted bythe Code for Pressure Piping ASA B31'1. The
Starczew-ski/Haque stored energl'method is an approrimate
solu-tion and the analvsis produces safe results.
Allowqble
Stress.It
is
recommendedthat the
stressobtained bv this method should be compared by the code allor,r'able stress range S.1; rvhere
S,r
:
I ( 1.25 ,Sc + 0.25 Sr, )S"
:
allorvable stress (S value ) in cold condition S,,:
allorvable stress (S value) in the hot conditionS" and S,, are to be taken from tables in the applicable
sec-tions of the code
f :
stress-range reduction factor for cyclic conditions. IJsea value
of
1.0 for one cycle pcr day or less. Consult2.0
1.8
r.61.4
t.2 r.0 0.8
0.6
0.4
0.2 SCALE PWeight and other sustained external loadings shall not
exceed Sa.
Pipe supports qnd restrqints are not considered in the
flexibility calculation.
It
is
assumed that the supportswhich have not been considered
in
the analysis shouldbe located and designed so as not
to
interfere with theflexibility of the system. The reactions computed by this method shall not exceed the limits which the attached equipment can safely sustain. Equipment such as. pumps, hrrbines and similar strain sensitive machines should re-ceive the manufacturer's approval; and the piping system
should be designed flexible enough to comply with their
recommendations.
SAMPLE PROBTE'VIS
The following data apply to all sample problems,
un-less otherwise stated.
PipeSize:3in.
Schedule
:
40Operating Temperature
:
8600 FE
:
Young's Modulus (cold):
27.9 (10") psi.Thermal Expansion
:
7.37 in. per 100 ft. 1:
Moment of Inertia:
3.02 ina.Z
:
Section Modulus:
1.724 trf .i :
Stress intensification factor:
1.78Material
:
Carbon SteelCode
:
Power PipingSe
:
Allowable Stress:
16,800 psi.Solution: Use Figures L and 2.
Fig. l-Force graph (above) and moment graph (right) for U-typ" pipe expansion loops.
Somple Problem
l.
Fora
simple'IJ'
type expansionloop as shown
in
Fig. 3, check maximum stressin
theloop, and
if
the calculated stress is much less than the allowable, suggest a loop which will produce a maximumstress equal to or near Sa. Space does not permit the mod-ification of 11, but G can be modified.
Given: 3 inch carbon steel line, Sch. 40, thermal
ex-pansion 2 in. per 100 ft., temp. 325oF.
Data:,
I :
3.02 h.n;Z
:
1.724 in.3;i :
1.78. Allow-able StressSr
:
18,000 psi (Power Piping).Solution:
w10
B-
:_:1
'H10
?o 30 40 50 t00 200 300 400500
a
5
rp00 2p00 3p00 4,000 5p00 rop00ll
2.0 t.8
t.61.2
r.0
0.8 0.6 0.4
0.2 SCALE BG10
o:7.-ro-1
z. T 50.u00,)
d = too,ooo E ut = o =6o
-
Total thermalexpansion:+
(100):2
inches.Note:
B
and a and other nomenclature have no con-nections with similar symbols used in piping design books.These are used
in
this article just as symbols.Step 1. Determine Force Fe using Fig. 1 (Force graph). Entei scale
p
withF
: t
then move vertically upwardsto the curye
a :
1, and now move horizontallyto
theright to the line 8o
:
2 and then vertically move downtJthe
lineFI
:
10ft.
and horizontally from this point to scale FalI
and read the value FafI :
38-The Force Fa
:
38(I)
:
38(3.02):
114.76. SaY Fa:
115 lb.Step 2. Determine maximum moment in the bend using
Fig. 1 (Moment graph).
Enter B scale with 13
: l.
Move to curved
:
1 andthen horizontally on right to
H :
70ft.
and vertically down to line F1:
115 lb. From this point gohorizon-tally
to
the moment scaleto
rcadBM,
(BendingMo-ment)
:
6,300 lb.-in.Step 3. Calculate the maximum stress
in
the bend. .S,:
Expansion stress: U{ ,,
:
63oo,r.rr)
-
6.52opsi.r.72+',
,S,
:
8,000 psi. S,(
S.500,000
The system is acceptable. The calculated Stress is less
than the allowable stress.
Step 4. Determine the flexible loop, rvhich
will
pro-duce a stress of 8,000 psi. Use Fig. 2..s.
18.000 Stress ratio1_l_- _-2.i'6
sE
b,5zuFind, y
:
0.475 from Fig. 2 when a:
1 and F:
1'New y
:0.475
(stress ratio):0.475
(2.76):1.312.
Enter Fig.
2 with
newy :
1.312, move to 13: l,
then travel vertically dorrnrvards to read ne\vd :
0.25.G Since a
:
PIPING DESIGN METHOD BEATS COMPUTER
l'-n
t23456789
a=G/H
Fig. 2-Variation of bending mometrt and stress with beta and alpha.
p:Y:
t,b/-
lo, tr-
to..H
The suggested loop will have the following dimensions: H
-
l0 f.t.,W-
10ft., G-
2.5 ft.,andU-
100 ft.This new loop will produce a stress
:
18,000 psi.Check stress:
Calculated
BM:
6300 lb.-in. Stress ratio-
2.76BM (Stress ratio)
-
6,300 (2.76)-
17,400 lb.-in.New stress
-
BM,,
17'4ooZ ,):
,12a
(1'78):lB,000Psi.Somple Problem
2.
Calculation of forces at the nozzles, see Fig. 4.,:(+)
-4/to_
0.4/c
\
a:l_
I
:8,/10-0.8
\H/
t-al U,:
100 (55) :4.05 in. (Thermal expansion of 'U') From Fig.
I
findFalI
:
l2O."'
Fe-
120(I):
120(3.02):
362lbs.Computer result
:
252 lbs.Kellogg graphical
:
368 lbs.Fa
:
Forceof
pipeon
anchoror
nozzle, caused bythermal expansion.
Somple Problem
3.
See Fig. 5.:0.4
:0
7.37 uo
:
100
(55):
4.05 in.From Fig. 1, by interpolation,
FalI
:270.
:.
F.q.:
270(I) :
270 (3.02):
815 tbs.By computer Fe:6421bs.
Somple Problem
4.
See Fig .6.-
0.4:0
l-it
oo:
100 (4) :0.295 in. Find force from Fig.
l,
FalI
:
20.:'
Fe-
20(I) :20
(3.02):
60.40lbs.By computet
Ft
:
46.0 lbs.From the above three cases,
it
is obvious that the resultby this method compared with computer analysis indicate
safe and reasonably accurate values; and for these shapes,
this method wins over computer analysis including data
preparation time.
Somple Problem
5.
See Fig. 7.G10
a:-
- --:1
H
10 7 .313:
-r55,-4.05in.
u-
looFind from Fig. 1,
F^lI :
78. F.E:
78(I) :
7A (3.02):
236 ]bs.M 1
-
12,000 lbs, in.Stress in the bend.
So-
M;!- qil-- ':::?
z ''
7.12+(t.78)
:
12,400 psi.S,i
:
16,800psi.
Sr(
S,rThe
reactions and stresses arewithin the
allou,ableIimits. The svstem is therefore adequately flexible.
Economic Loop Design.
If
the anchor forces are not thefactors which dictate the design, then the loop can be
quickly determined by the use of this method which will produce maximum moments and stresses equal to its al-lowable limits. No
trial
and error method is needed. Byusing Fig. 2 a great deal of labor can be saved. This also
eliminates a large amount of mathematical computations
and reduces the chances of errors to a negligible degree.
The graph is self-explanatory and results are sufficiently accurate
for
most engineering purposes.The
following example illustrates how Fig.2
can be used efficiently.oW4
'
H
10GO
A:-:-H10
T
HI
oW4 p:-:-'_H10
GO
H10
O: G/H B=w/tt t.0 0.9E
0.8P
0.7s06
u)
0.5 at)H
0.4 F <n6
0.3 F z. trJ=
0.?=
>-w10
6:_:
:
IH
r0 0.09 0.08 0.07 0.06 0.05t3
If
the calculation, based on either the Starczewski/Haquemethod
or
an
analytical
method, indicatesthat
themoments and stresses exceed the allowable limits, then
Fig.
2
can be usedto
predict the guide distance "G" which would enable the shape to become flexible enough and to yield moments and stresses equal to the allowable limits.Sompte Probtem
6.
A
loop which has the ratio a:
GIH
:3
andp
: IrylH:
5, and the solution indicatesthe Bending Moment
:
50,000 lbs.in. and
ExpansionStress
:
30,000 psi. This exceeds allowable limits.Allowable BM :25,000Ibs. in.
Allowable Sr
:
15,000 psi'Determine, U-tyP" symmetrical expansion loop to yield
the maximum moment and stress equal to the allowable
limits.
Solution, Step 1. Determine from Fig. 2 when a
:
3and
p
:
5. FolLw the arrows and read value of 1:
O.l7which is the moment and stress factor on the left hand
vertical scale.
Since
the
allowable stress Sr:
15,000 psi. and the calculated expansion stress SB:
30,000 psi' then theratio
:
15,000/30,000:
0.5.Therefore, corrected y
:
0.17 (0.S;:
g.g8t.It
is assumed that F:
Wlu :
5 remains unchanged'Step 2. Determine new d.
Enter in Fig. 2 the corrected y
:
0.085 on the vertical scale from theleft
hand sideof
the graph, and movehorizontally to the right to the curve F
:
5
then movevertically down to the e scale and read the new ot
:
5.7.Now summarize the new values as follows:
a:
5.7B:5
Expansion stress Sa
:
15,000 Psi.Bending Moment BM
:50,000
(0.5):
25,000, equalsthe allowable moment.
Step 3. Determine the distance of the guide 'G'.
New a
:5.7 :
GIH
when 11 remains as before.Therefore, the distance
of
the guide G:
5.7 (H).
II
the
'G'
is increased to 5.7(H)
then the new shape will yield a stressof
15,000 psi. which is equalto
allowablestress.
Bosis For Method. This method assumes that the energy
is stored
in
a system because of bending, which is causedby
the deflection dueto
the thermal conditionof
thepiping material.
Energy stored
: tY!
(1)J2
Again:
It
is
acceptedthat
whena
piping system-is subjectedto
deflectionit
stores energy,and
that
the stored energy must be equal to the work done upon the piPe'The work done by a force upon a piping system
: y?
(2)t4
Fig. 3-Simple U-type expansion loop for Example
l.
Fig. 4-Calculation of forces on nozzle, Sample Problem 2'
G:0
oo'
U
Fig.'A-Figure for Sample Problem 3.
Fig. 6-Figure for Sample Problem 4.
.rc'J
l*-u=,0'U=55i
Fig. 7-Figure for Sample Problem 5.
where, F
:
Force andX
:
Distance travelledby
theforce in the direction of the force 'F'.
Also: Interndl energl' stored
by
the system is causedby'
e
Compressive stresseso
Tensile stressesPIPING DESIGN METHOD BEATS COMPUTER .
Fig. &-Basis for the method starts with this piping layout.
'i-1
Fig. 9-The loop deforms from the cold to the hot position.
l.--G6---*j
l.-Gf ---i
Fig. 10-The forces and moments acting on the Fig.
I
loopare shown.
If
the energy stored by compressive and tensile stressesare neglected then the total energy stored
will
be due to bending moment only.See Figs. B, 9 and 10.
Energy due to Bending Moment
:
BM:
I+
(3)M
But d0
:
EI
d*
(4)Therefore, the equation will be transformed as follows:
ur,:l*!ru:l+W)
fMzdx
:Jzn
(s)Deflection taken by the shape between
(a)
and(Gr)
i.
equal to 8o/2
Fa6o/2
f
U, a*I
G"2 :l-Znr)n
(5A)Thermal stresses
in
symmetrical expansion loops arederived as follows: E,
f-i
ts-Lq lzl---t
R1Deflections 81, 82 and 8s are produced
by
elementsLab, Rae and. cd respectively.
The system is confined between
2
anchor points.Le:
Teft hand side anchorRa
:
right hand side anchor.The following conditions apply
in
derivation:o
That the whole system lies in one plane.That the system is treated with square corners
inter-sections.
That
the system is composedof
straight elements of pipe of uniform size and thickness.o
That the thermal expansion of a given element isab-sorbed by the elements orientated perpendicular to the direction of the deflection.
o
That the effect of dead weight, wind etc. are neglected.o
That the clearance between guides and pipe is nil.o
That the compressive stresses within the element areneglected.
o
That
the flLxibilityof
the elbow caused by an ovalshape is neglected.
The three deflections
in
the horizontal plane are:81
+
8,*
8,: 8o:
total deflectionThe total deflection 66 must be absorbed by the system
contained between the points a and fo which is tJle flexible portion of the system, see Fig. 9.
The loop
will
deform from a cold condition to a hot condition, when the system is subjected to the operating temperature, as shownin
Fig. 9.Fr
:
thrust acting on the loopat
the point,a,,
seeFig. 10.
Mo:
bend'ng moment acting on the loop at the point 'a', see Figure 10.From Fig. 9
it
is evident that the slope at the point Gomust be horizontal.
IIence the change
in
the slope between pointa
andpoint Gs is equal
to
zero.i.e
ll
o'1"":o
F o: Fo and Go- G1 Fr+
(6)Mo:
M6Bending moment, BM at point X
-
X-
Me-
FEU) e)
Slope and Bending Moment. The general relationship between slope and BM, is as follows:
dt: M,!:
(8)EI
Therefore the total slope change between point a and
point Go is:
ll+l'"
X
:
Pipe Iength in general,(s) Go I q. I -l I [,--1
f--0.---!
.:llr#):".
llr-#,
*,),
*
ll*+*
*)*,"
But the changein
slope betweena
and.Go:
Q*{*"*o)
+M,(H)
-L#L
*
Y* -
elnw)l:o
f
w1 fu,
r+twtl
*"1.
+H+
T):r"l;*-)
f H,+H(wt I
T
H2+H(w\
I
M,:Fo
Lzrdffit
):'"tdc
+TiTT)
M-- F,'"
"2(G+H)+w
H*W
Now refer Equation 1 to 5a,
Further, (10) (II)
oruo:
4 EI 3o 2Fo EI 3o 2 FoH2 Er 60w
:
r,
{,,
+ w),
\,
I'
I- W'2
H 3HW
-3-2
4(G+H)+2tr
I
(13) t+2
p + p,+al4l2P
H2 1 -)-Hl-L2 _ tl/ -Lc
- rl
H(#)'
:"[-f,
*
Force, Moment, and Stress. The follou'ing units applyin
the formula below:Modulus of elasticity
:
E
(psi.) Resisting force:
F,
:
F"
(lbs.) Moment of inertia of the pipe:
1 (in.n)Total thermal expansion between anchor points is in
inches.
Height of the loop
: Il
(ft.) Width of the loop:
W
(ft.)Distance of first grride
:
G (ft.)Section modulus of pipe
: Z
(in.')Stress intensification factor for the bend
: i :
0.9lh
where
h
:
tRlr'
See code for pressure piping ASA 831.1.
Then the force is
,
+2
+
'
(#)'
-(+)
W LLL,) 't-HI
r+28+p'
--3 , 2
I4aa4l29
where F - I,Y/H and, a : G/H.
I
,,-,
2B+
++2p
4
1 4 B 2'oouo
:
;*{ll
-;
o""f:'
+ll
w.-
rot)
*7'"*ll
,*"-'outo'f*""|
#
{rr
",
(r-ffi *r)'
(""
*
"
*[)
-
FoHr#T,
(FoH, + FaHW)*r,,
(F
*
L)\
I c+H+w/z
l
l, H,
zz\
ttr@+h+wF -
T(G+h +w-l
-
3
-
Tl
:
,o*
rr'{
212(G+H)+w)
2(G+H)+W
(H+
14/)2 t1/ H ++)
.
r*'(+
w Ht6
(12)PIPING DESIGN METHOD BEATS COMPUTER The maimum stress can therefore be in a system either
at bend b or at the bend c.
It
is suggested that the moment at point b and point c be calculated, then take the largest moment of the two to calculate the expansion stress SE in the loop.M
s,,
-
_
(i)Fig. l2-Loop designations for round corners.
Fig.
ll-For
round corners, ttewidth, height, and guide distances must be modified.
Fig. l3-Symmetrical loop
with guide G distance from
bend.
Fig. l4-Symmetrical loop
with guide very near to the bend.
( 1e)
Loop Restroinls qnd Supporls. Design engineers should
make certain that the loop between the
two
guides ismade to function without any environmental obstructions
or restraints. Also that the system is fully supPorted and
no branch connections are made within the flexible
por-tion
of
the loop.It
is not recommendedto
induce anyexternal loading upon the loop.
The
designers shouldavoid locating
rigid
sections, such as large valves etc.,within the loop.
A
good practice is to locate valves and other rigid sections near the guide or between the anchorand the guide.
Line Size Limits. This method provides engineers with
reactions and stresses
that
are reasonably accurate for pipeup to 6
inchesin
diameter. Lines above6
inchesin diameter can be safely analyzed by this method. How-ever, the authqrs would like to point out that the results
thus obtained will be on the conservative side. Therefore, where the system is dictated by the space, reactions and
stresses or economic limitations, a precise analysis should be made.
Round Corners. Solutions to loops having round corners
can be solved as follows:
o
If
the
square corner solution gives reactions andstresses, which exceed acceptable limits,
o
If
the radius of the bend is more than 1.5 pipediam-eters,
o
If
the line is above six inchesin
diameter,.
Refer Figr,rres 11 and 12. Use Equations 20, 21,22, and 23, to modify the width (W),
height(I1),
and guide distance(G)
respectively.Use Figs.
I
and 2 to determine forces and moments.Assumption:
R
:
radius of all the bends whichI{:
height of the loop which mustsides.
G
:
distanceof
the guides which both sides.must be the same.
be equal on both
must be equal on
Fig.
l5-A
two-plane loop configuration.Er
s^
l-n'-
" t_
^ 3456H3 I 1 IIs
l
isp
L+2tt+t2
, Z 4a*4*2F
the following forroula
(lbs. ) (15) derived. ( 16) (17) ( 1B) From equation (10) Mo: FoH
H+W
2(G+H) +W
W:u+1.57R(K/3)
H-h+t.s7R(K/3)
H-h+1.57R(K/6)
C:e*1.57R(K/6)
K:1.65/h -flexibility factor, K) 1, rf R)rvhere h - ' ..' ; i: thickness of pipe; r: mean
r2
pipe: R: radius of the bend,
1-rB ' Fo
It
T; )-,
+-O (lbs.-ft. t Note that Fo:
F^ (20) (21) (22) I 2q\ (24) radius ol M": Mo- Fr(H) (lbs.-ft.)Note: For force calculation use modified Height (H) of
loop. For moment calculation use original Height (II) of
ioop'. the the
ln
general: Stress:
M /Zt7
For Figs. 13, t4, and 15,
u:
W-
2R;h:
H
-
2R;g:G_R.
Symmetrical Loop. Refer to Figs. 12 and 13. Modify W,
H,
and G as follows:For W, use Equation 20.
Eor
H,
use Equation 21.For G, use Equation 23.
Symmetrical Loop. Refer
to
Fig.
14. Notethat
theguide is very near the bend.
For W, use Equation 20.
For H, use Equation 21.
Since the guide is at or near the bend,
G:0.
Two-Plane Loop. Refer to Fig. 15.For W, use Equation 20.
For H, use Equation 21.
For G, use Equation 23.
U-Loop With Equal Legs. Refer to Fig' 16.
u:W-2R.
h:H-R.
For W, use Equation 20.
For 11, use Equation 22.
G
:0.
Two-Pliane Loo,p
With
Y-Leg Longer Than Fitting to Fitting.w:W-2R
hr:
Ht
-
2Rhr:
H,
-
2Rg:G_R
For W, use Equation 20.
H:.:
Ht
*
Hz. For G, use Equation 23.Two-Plane Loop With Equal Legs.
u:
IU-
2R.hr:
H,
-
R.Hr:
H,
-
2R.For W, use Equation 20.
For
IIr,
use Equation 22.For H2, use Equation 21.
H:
Hr]-
Hr.G
:0.
U-Loop
with
Equal Legs and Single Tangent. Referto Fig. 19 and solve the same as Fig. 16.
Grophicol vs. Computer Solufions. The following
ex-amples are given to illustrate the results obtained by this method and by the computer.
Example. Given:
4
in.
pipe, schedule 40; radius ofbend
:
0.5ft.;
operating temperature:
3500 F.;mate-rial
:
ASTM
106 GR. B;thermal
expansion:
2.26in./100 ft.; allowable stress
:
22,500 psi.; code:
PowerPiping; moment
of
inertia: I :
7.23 (in.a); sectionmodulus
:
Z'
:
3.22(n.3);
stress intensificationtac-tor:i:1.95.
t8
Fig. 17-A twb-plane loop configuration with the Y leg longer than fitting to fitting.
Fig.
lLA
U.loop configurationwith equal legs.
Fig. 18-A two-plane loop with equal legs.
Fig.
19-A
UJoop corfiguration with equal legs and single tangent.a:GfH :
12110:
1.2;B:
WIH
:
B/10:
0.8; 8o:
(2.261100) (180):
4.06 in.From Fig. 1, Force Graph,
FII : r :
86Force
: r(I)
:86(7.23)
:
620From Fig. 1, Moment Graph,
BM
:
27,100 lb.lin.stress
: +#
(1.95):
16,450 psi.Fwi
PIPING DESIGN METHOD BEATS COMPUTER
.
. ized staff members to spend their valuable time only onthe analysis
of
critical and hazardous lines. Alsqit
will help site engineers design and incorporate a loop in a rackpiping system, or in a long transmission line, without
seek-ing help from the design office.
Furthe4 the method
will
help project engineersesti-mate piping flexibility in the proposal stage of the project. Symbols used:
Allowable Stress (psi.) Modulus of Elasticity (psi.)
Moment of Inertia (in.u) Section Modulus (in.3)
Force (lbs.)
Force component in the direction of axis.
Moment (lb./ft.)
Stress intensification factor. ,s/ E
I
Z F FxM
i
Fig. 20-An example configuration calculated by the graphical and computer methods.
Fig.
21-A
square corner configuration calculated by graphical and computer methods.Method
Force, lbs.
Moment Max. stress
at guide at bi:nd, Remarks
Ib./in. psi.
About the quthors
M. S. HIQUB is an engineering
consul-tant as s o ciated.with E M M C O N, L ondon.
He specializes
in
pipe stress analysi,s,piping lo,gout, and fleribilitg analgsi,s i,n th,e lal1out stage. Mr. Haque receiued
a diploma
in
mechanical and electricalengineet"ing from Dehri Technical
In-stitute
of
India.He
is
an
o,ssociatenzember of ASME, Institute of Engi-neering Designers, Institute
of
Plant Engineers, Associate Fellou; of the In-stitute of Petroleum, and Associate of the Institute of Fuel.He has had 16 gears enperience as a senior designer and
piping analyst usith such, firms as Wellman Smith Ou;en
Eng. Co., Mattheu Hall & Co., Ltd., McKee Head
Wright-son, Ltd., and Constructors Joh.n Brou.tn, Ltd., all in London.
He also has fi,ae years fi,eld eupet'ience on construction jobs
in Ind:ia.
J. St-q.nczowsKr 'is an engineet'wi,th Woodall-Duckham, Ltd,, Crawley, Eng-land. He o,ttended the Polislt Techruical
College
in
England and completed a B.Sc, mecltanical course at LondonUrvi-oersity. He has done graduate work in
fluid dgnamics, mathematics, and
nuclear energA.
Mr.
Starczeu.tski has had, enperiencein
heat enchangerde-sign, pressure aessel design, u-telding equipment dea elopment, special pu,t'pose
machine design and, other process equi,pment design. He h,as
uorked, u.titlt, such firms as Constructors John Brown, Ltd.,
and, Caird & Raynet", Ltd,. both in London, Graphical 620 27,100 16,450 square corner
solution Computer 2 7,000
t6,47s
.t;?.?;,:to"*Example. Given: 6 in. pipe, schedule 40.
All other data as in example above.
8o
:
4.06 in.,I :
28.1,Z
:
8.50,i
:
2.27From Fig. 1, Force Graph,
r
:
86,r(I) :
86(28.1):
2,410 lbs.
From Fig. 1, Moment Graph
BM
:
90,200 lb. in.Stress
:
90,200 18.5(2.27):
24,550Force, Moment Stress,
lbs. lb./in. psi.
Graphical 2,4tO 90,200 21,550 sq. corner soln.
Computer 2,179 90,500 20,122 sq. corner soln
In
conclusion, the authors feel that the introduction of this time saving piping flexibility analysis method will helppiping design engineers solve most of the simpler config-urations. This
will
allow consultants and highlyspecial-6t7
Symmetrical Piping
Arrangement
Solves
Two-Phase
Flow
Distribution
Problems
The
secret
to two-phase
distribution
in branched
piping
systems is
strict
adherence to symmetrical
piping
and
an
evenly dispersed
liquid
flow pattern
John L. Greene
The Fluor Corp., Ltd., Houston
A
rnequrNT ENcrNEEnrNcproblem
is
designingbranched piping systems
for
flow distribution, mist ordispersed flow, and an over-all low pressure drop.
Con-trolling flow patterns, liquid distribution, flow
distribu-tion, and optimizing pressure drop need to be considered.
Recognizing
flow
polterns
in
two-phase flowis
thefirst part of the problem.1,3
In
a two-phase system, when gas flowsat
various rates, demonstrative typesof
flow patterns are developed.In
general, these flows arede-scribed as bubble, plug, stratified, wavy, slug, annular, and spray, mist, or dispersed. Slug formation, plug flow, wavy flow, and stratified flow are shown in Figures
l,
2,3. and
4,
respectively. For this discussion, slug flow isdefined as a mixture of liquid and gas that has a varying density with respect to time. Therefore, the term slug flow
will
also include plug flow andwill
border on annular and bubble flow.fn
engineering design the flow pattern must be deter-mined in every two-phase application.fn
a service where pressure fluctuations cannot be tolerated, there can be no slug formation. For example, slug flow downstream of adistillation column
will
cause pressure fluctuations and unstable operation,or
downstreamof
catalytic reactorsit
can cause catalyst attrition.Liquid Distribution. When two phases flow through the
same pipe, the gas flows faster than the liquid.
In
a20
Fig. l-Shows slug flow; G:0.0085 lblsec;
L:
0.38 lb,/sec.*Fig. 2-Shows plug flow; G:0.00421 lb,/sec; L:0.38 lblsec.*
Fig. 4-Shows stratffied flow; G
:
0.0081 lblsec; L:0.38
lb/sec.'
* Water and air at atmospheric conditions and in a 2-inch OD horizontal pipe.'
ll
,'=T,'
(I) ELBOW PERPENDICULAR (2) ELBOW PARALLEL (3) TEE 8 CAP
TO HEADER TO HEADER TO HEADER
(GOOD) (POOR) (ACoEPTABLE)
Fig. S-Shows liquid distribution into a header:
(1)
elbowperpendicular to header (good); (2) elbow parallel to header
(poor); (3) tee and cap to header (acceptable).
The problem of two-phase flow distribution
in
manifoldpiping arrangements is frequently encountered
in
large plants, particularly around air coolers, parallel exchang-ers, etc.The simplest solution to flow distribution is to provide a block valve
in
each branch line. From the standpointsof valve and of pressure drop costs this is often
unattrac-tive. Therefore, the pressure drops through the system
must be depended upon to distribute the flow. As is shown
in Figure 6, if valves are not provided in each branch line of two-phase flow, then the layout should be symmetrical.
For comparison the preferred layouts
for
single-phaseflow are shown
in
Figure 7. The selection depends uponthe importance and duty of each service. Friction loss in the fittings was determined by using Bernoulli's Theorem
(Velocity head method) and velocity head coefficients
from the literature.5,6 To determine the pressure drop in two-phase flow when there
is
less gasby
weight than Iiquid, two-phase flow correlations should be used.2Applicotions. Savings can be realized in optimum
over-design of heaters, heat exchangers, etc.
In
large systemshorsepower usage from pumps and compressors can be
reduced.
With this knowledge of how to minimize pressure losses
in manifolds, plot plans can be laid out more efficiently.
The
preferred piping layout creates fewerplot
planchanges and shorter pipe runs.
Slug flow causes pressure fluctuations
in
the system.Elimination of slug flow helps stabilize the unit. Slug flow can also cause major problems such as catalyst attrition.
Therefore, catalyst life can be increased. Minimizing
pres-sure drops and equalizing
liquid
and flow distributionwill increase yields and decrease capital and operating cost.
Exomple Problem. Distribution into and out
of
a
16-section air cooler with 5.3 pounds per square inch pres-sure drop (16-6' nozzles) and
with
the following flow:LIQUID
Flow lb/hr.
Specific Gravity @ Tem-perature & Pressure
Temperature oF
Viscositl, Cp.
VAPOR
Fis.
pass
&-Shows symmetrical piping and (2) two pass.
(2) TWO PASS
in two-phase flow: (1) three
(I) GOOD DISTRIBUTION (2) FAIR DISTRIEUTION (3) POOR DISTRIBUTION
FiS.
Iayouts for single-pass flow intwo
d distribution, (2) fairdistri-buti
smooth turn the iiquid has a tendency to follow the out-side wall. The elbow
or
turn should be perpendicular to the manifold, as is shown in Figure 5.If
thii
is notpossi-ble, a tee and cap or a mixing length after the turn may be used.
The liquid must be distributed into heat exchangers,
air
coolers and other typesof
equiprnent. Oftenit
is necessary to rotate an elbow to the shell of an exchangerin order to distribute the liquid on the baffling ,.r.a.rg"-ment. Tees and caps or mixing'lengths are also used on
the inlets to heat exchangers. On the outlet, liquid
dis-tribution is not usually important.
The severity
of
the operation and theduty
(size of heat release)of
the service as selected to provide liquiddistribution
are
determinedby an
economic balance.Therefore, each case must be looked at individually.
Flow
distribution
in
monifold piping
systemsis
afunction
of
pressure losses through each lateral system.IN
484,000 0.670 277 0.310IN
OUT 652,000 0.678 150 0.412 OUT Flow lb/hr. Molecular weight Viscosity Cp. 533,000 8.49 0.01132 365.000 5.99 0.0099Three cases r,vill be considered
to
determinethe
bestpiping layout. Economics prevents putting valves in each
of the 16 sections.
Cqse.
l.
One headerwith
16 ]aterals on the inlet and outlet is shorvn in Figure B.(I) THBEE PASS
ST RATI FI ED
Flow Pattern. From Baker's1,2 two-phase flow
corre-lations the type of flow is determined aIter each lateral take-ofl and is shown
in
Figure 8. The flow patterns gofrom mist to annular to slug to stratified flow. This flow pattern change is not acceptable.
Pressure Drop. The pressure drop calculations are a
trial and error procedure to determine the exact
distribu-tion. Using Bernoulli's Theorem and velocity head coeffi-cients from the literature5,6,7 the initial pressure and flow distributions (assuming equal distribution) are as follows for the first and sixteenth pass:
o
First Branch System: AP1:
10.46 psiincluding
cooler loss
o
Sixteenth Branch System. APro:
8.24 psi including cooler loss.'. Percent flow not distributed
:
4.89 percentOn a service with a large duty this 4.89 percent of mal-distribution
of
flowis not
acceptable. CaseI
isnot
agood system.
Cose
ll.
One tapered header with 16 laterals on the inletand outlet is shown in Figure 9.
Flow Pattern. From Baker's1'2 two-phase flow correla-tions the type of flow is determined to be in mist flow the
total length of the header on both the inlet and the outlet. Pressure Drop. The pressure losses were calculated the same way as Case
I
only the expansion and contractionlosses were considered.
'
First
Branch System:aP,:1i.86
psi
includingcooler loss
o
Sixteenth Branch System: APre:
9.55 psi including cooler loss.'. Percent flow not distributed
:
4.64 percentOn a service with a large duty this 4.64 percent of mal-distribution of flow is not acceptable. There is 8.73 per-cent rnore pressure drop than
in
CaseI
and the tapered header is expensive. CaseII
is not a good system.16-6.. LATERALS SPACED AT
8
SECTIONSFig. 8-Shows single 24-irch headers with 6-inch laterals (example problem-Case
I).
Fig. 9-Shows tapered (24
x
8-inch) headers with 6-inchlaterals (example problem-Case
II).
Fig.
l0-Shows semi-symmetrical manifold piping layout (example problem-CaseIII).
Cose
lil.
A
semi-symmetrical manifold piping system isshown in Figure 10.
Flow Patterns:
AII
of
the pipingis
designed so that only mist flow is encountered. All turns into headers haveto be rotated corectly so that the liquid is evenly
dis-tributed.
Pressure Drop. The pressure losses were calculated as
in Case I.
o First
Branch System: AP,:
9.34
psi
including cooler losso
Fourth Branch System:APr:9.97 psi
includingcooler loss
Percent flow not distributed
:
1.96 percentIn Case
III
the florv is in the dispersed legion throughout the system. The flou, distribution is the best that can beeconomically justified. This is good piping la.vout.
LITERATURE CITED
Indqirrg Tere: Ctrmputations-l0, Design-4,8, Distribution-7, Fluid Flow-4,7,
Layout-4,6. Liquid Phase-5, Piping-9, Vapor Phase-5.
NOTES
Plont
Loyout
qnd
Piping
Design
for
i,tinimum
Cost
Systems
Afier
process
ond
equipment ronditions
ore
sel, plont loyout
con
be
the
lorgest
single
cost
sover
in
HPI
plonts.
Line sizes
ond
pressure drops
depend
on
pipe
lengrh
ond configurotion.
Use
these guides
io
moximum
plping
system
economy
Robert Kern, The M. W. Kellogg Co., New York
PrprNc EcoNoMy is closely
related
to three areas of plant design:o
Equipment layouto
Piping designa. Line sizing and flow slntems
b. Piping layout, and
o
Piping detailsThese areas are interdependent; without an economical
Plonl
Loyoutqnd
Piping Economy. Plant layout canbe the biggest single cost saver in refinery and
petrochem-ical plant
design, after process and equipment designposibilities have been exhausted. Savings can be rcalized
not only
in
piping but alsoin
the cost of pumping com-pression andutility
cost. Oftena
layout can eliminateequipment
(for
example, pumpswith well
arrangedstandbys).
The most important document issued
to
the layoutengineer
is
the process flow diagram (PFD). This hasto
be evaluatedfor
an economical plant arrangement.From
a
layout standpoint, three typesof
lines can bedistinguished.
Main Process Flow Lines. First, lines which represent
the main process flow. Such streams pass
through
furnaces,
reactors
and dryers, then they continue astower bottom and feed
inlet
to
the next tower, oftenwith exchangers and pumps between them. These lines
will
bethe
shortestif
towers are arrangedin
processflow sequence as close to each other as equipment sizes
and access space permits.
With
smaller interconnecting lines, towers can be located further apart without much increasein
piping costil
other economies can thus berealized.
For
example: the groupingof
condensersbe-tween two towers can result
in
a
shorteningof
cooling water lines;a
conrmon steam line can be designed for grouped reboilers. Grouped condensers and reflux drumswill
permita
common supporting structure. Figure 1shows
an
exampleof
alternative tower arrangements.Many configurations are possible and justified
if
shorten-ing
of
these process lines is the ultimate result.Process flow is not always
a
simple straight through flow but can split into two or three streams, as isoftL
donewith a
numberof
distillation columns. Subsidiarycircuits to process flow must also be considered such as
the refrigeration circuits
in
ammoniaor
ethylene units.are generally large diameter lines and should have pref-erence over the
first
group which arewually
smallerprocess lines.