(
("''
''l SAFETY/HEAT TRANSFER
'-
i11ta.
Tcwer pressure relief calculation A practical, less-complicated method for different emergency cases S. RAHIMI MOFRAD, Petrofac Engineering & Construction,Sharjah, UAE
I n he hydrocrrbon pr ocesring indusr ries and chem ical plan rs. f Ir;criorar ion rorers are some of rhe morr complicared equ ip'{ r
menr lrom a reliel load srudy point ofview. fhis is due ro rhe variety of tower design parameters and operating conditions. The process designer is responsible for recognizing the different causes of over-pressurizacion and evaluating the relielload considering all possible operating paramerers such as difFerent feed composirions, ca, ies and producr specifications, as well as start/end ofrun or siimmer/winter operation. Failure to recognize these effects is likely to expose che column to danger ofoverpressure and the possibiliry ofcatastrophic failure. A conservative and most common method for determining rower reiiefrequirements is to design rhe pressure protection system based on total gross overhead vapor flo\i/rare. Rec€nt tower reliefstudies show that rhis i. not always consewative, since there is evidence rhat a"ppto".L reliefvalve sized on this basis can be even undersized.l The best available rechnique for a reiiefstudy is dynamic simulation but it requires comprehensive data on equipment specifica, tions arld dimension, hydraulic informarion and control system derails. Mosr of these data are not available or finalized wheq a reliefsystem is designed. That is why reliefload calculations using dynamic simulation is best suited for revamps or debotdenecking pro.jects. Moreover, dynamic model va.lidation wirh respec to hear and material balance results and dynamic model reaction time compared with actual plant operarion is another concern when dealing with dynamic simulation. There are many emergency ca.ses applicable ro rowers: :ctric power failures (feed pump, reflw pump, air-cooled conddnser fan, pumparound pump, overhead gas product compressor failures or a combination ofthese) . Cooling failures (water-cooled condenset air-cooled condenser, pumparound pump or cooling water pump failures) . Control/manual valve malfunctions (feed control valve, reflux control valve, vapor product confiol valve or cooling water valve closures; liqLrid producr control valve, reboiler sream control valve or heater fi:el gas control valve wide openings). The reliefrate for all of*rese upset condirions can be estimated by following a three,step proceduie: reboiler pinch study, relief condition simulation and overhead energy balance. As column pressure increases from normal operating pressure to relieving pressure, column temperature rises. The reboiler temperature difference may reduce and consequenrly lead to lower heat input to the tower at relieviog condition. The ffrst srep should be ro determine che reboiler duty at the relieving condition (pinch study).
. Data collection
, T-,N, T"*,u, U" A, Qy from
I
reflecr,
2. Recognize that the heating media inlet remperature will not change when the tower pressure is increasing from operating to
relieving, rherelore
f* q
|
,,u.
3. Obtain to,,,pfrom simulating rhe bottom liquid composition by using process simuladon sofrware. The iemperature corresponds to the boiling remperarure of the liquid botrom .. composition at relieving pressure. The bottom composition at the relieving condirion is a function of m"rry pararrr.ters str.h as the tower's internal conffguration, cause ofoverpressure, feed and , reflu-x composirion, tower sump hold,up time, trays liquid hold- . up, etc. Ifrhe towert intemal configuration caused fie light liquid coming off rhe tray to go direc y to the reboiler, the temperaiure pinch will be less probable than when rhe lighr liquid is mixed
wirh sump liquid. Differenr reboiler ourler remper:cure" will be achieved rhen the cause of overpressure is feed failure compared with reflux ' or cooling failure, because different compositions will go to rhe-', reboiler. Fig.2 shows che effecr ofthe rower's inrernal configura, .-tion on bottom liquid composition. The efect ofother paramlters should be evaluated by the designer. If the designer has no idea about the composirion, using the operating composirion for relieving condition is often conservative.
. Initial checking
Reboiler pinch study.
1. Collecr r,o1q, t"*,tt
flow diagram (PFD) or hear exchanger darasheer. Fig. he schemari. represenrarion of reboiier remperarures.
r
a process
Fsrimaring the reboiler duq rr relier ing condir ion. Qp. i. lasr and easy ifone ofrhe following conditions is estab]ished: 1. Ifthe fired heater is used as a column reboiler, *re temperature pinch rarely (if at all) occurs because firebox r.-p"r.trrr., can easily reach 1,700"F-1,800"F. The rower reliefsruJy should be carried out with normal reboiler dury (ea = ed.
2.If t",,,R> f,,.p rhen rhe heat transfer in the reboiler will gradually be suppressed while the tower pressure increases from SEPTEMBER
2oo8
]
]49
SAFETY/HEAT TRANSFER Q, I \ -e. f /.,"^ -/., vJ
!n.a=t"ut.R
1. Beboiler ir once-through type (tray draw-of0? 2. Column sump has baffle for reboiler draw-om
fl)
3. Calculate the estimared heating media outlet remperature ar relieving condition:
,,.,
Bottom holdup time is less than 1tF15 min?
^
=r.._,
xF,..,
-r".,.,)
(2)
4. Calctrlate Qa = LI,A LMTDR whete: LMTD
r
(3)
(4)
=
^IP_A 5. Calculate the new rario of QalQ1, and compare it with the previous value. If they are not the same go ro Step 2, otherwise operating to relieving pressure. The reliefrequirement will be zero, due to zcro reboiler dury at this condition (Q,? = 0). 3.If U,A (T",,tN r,,,,a) > Q,y, rhen there is enough differential temperature benveen both sides ofthe heat exchanger at relieving .ondition so that the lemperarure pinch will never occur. The tower reliefstudy should be carried out with normal reboiler duty
(Q.r=
Qry)
Otherwise, the following mechod can be used ro specify the reboiler dury at relieving condition:
. Final calculation 1. Guess Qa/Qy (first guess could be 0.6). 2. Calculate the estimaced process side inlec lemperarure ar relieving condition:
go ro \rep 6. 6. If QalQyis less rhan 1.0, the reboiler duty will be decreased to Qa ar relieving condition otherwise normal reboiler duty is . utilized for rhe reliefstudy ofthe tower
Relief condition simulation. To preclict rhe effects of the tower's internal stream and tray hydraulics on heat and mass transfer and consequenrly on tower reliefload, which is ignored in common merhods, rhe following method is recommended: 1. Evaluare rhe latent heat ofthe tower bortom liouid ar relieving condirion. Ifthe rower design pr...ure is n."r oi high.. than the liquid crirical prcssure, rhe physical properties, including latcnt heac cannot be calculated easily. Referring to refere.r..s i PROTECTING YOUR ADVANTAGE
i4 lNLlNEaType4X
CANCEPT" Type
4X
Alultt ar3 patnt
Latch
[wo
Dau
D'connect
Type 4X
Standard, Stylized, Modified or Custom En(losure Protedion, Fast from Hoffman, Hoffman has more sta n ess stee enc osuTe so utions to protect vital e ectr cal systems and electronics in dirty, corroslve and wet environments. 5e ect from an assortment ofJype.+X models, includinq NL NE@ or CONCEPT' sty lzed enc osr..tres. Choose mu ftpoint latches or our UL 4X certified, 3 point POWERGLIDE' atching We also feature two door cablIets and disconnect enclosures for all popuiar switch brands. Each in dozens of standard sizes;in wall mountorfree-standing models. Modiry slandard enc osures to get exactly what you need quick y. We' I also work with you to co develop or customlze solltion5 for your app cat on Contact your ocal Hoffman distr butor, or ocate your sales off ce at hoffmanon line.com Select 187 at www.HydlocarbonProcesring.com/Rs
ZONEX" ATEX-Certified
for Hazardous Locations, ZONEX* Encl05ures protect eledrical equipment from harsh environments in EEx ezone 1 and Zone 2 locations. Available from stock, or modifled per customer request with terminal block installed. @2008 Notrman EnclosLrrc5
hc.
SAFETY/HEAT TRANSFER and 3 is r€commended for this purpose. 2. CalcuLare thc reboiler vapor retu.n flowrate at rclieving
,o rdiriorr b' di.idirgrlr.rebo'lcr Jug rhe
rr
rclievirrg.ordiri,,n
l,m
pinch studv, Qp bv latent heat.
3. N{odel rhe rorver as an absorber using simularion solnvarc. Set the tower pressure at relicving pressure and use above vapor flow as inlet g:s flow to the tower ro ger rhe absorber overhead flowrate, ,41p 4. Thc following iterns should be done belore running the sirnulation: . Availabiliry offced and rellux srreams have ro be checkcd by dre proccss ciesigner. S?hile the to\yel pressure migrates fiom operating to design due co emergency cases, fccd and/or reflu-x purnps may not be able to keep the same liquid ilow against higher des, tination pressure. This is seen especiaily with centrifirgal pumps Lrecause the shurolf prcssure of a cen!rifugal pump is generally 2!% higher than operaring diffcrenrial pressure. Ifrhe differential pressure ofa feed purnp at normal flow is 4.0 barg, it can produce i.0 barg diffirenrial pressures ar zero flow. for example, i1: che towcl operating pr€ssure increas€s as a result ofrellux controlvalve lailure fiom 2.0 barg ro design pressure of3.5 barg, lecd florvwill also stop. To*,er pressure will increrse i.5 barg while the pump\ harge pressure cannot increase more than 1.0 barg. Ifa feed failure case is studiecl, then rhe reflux strearn should Lre considered as absorbent. . If a reflux I:rilure case is srudied, rhen rhe leed is raken as absorbcnt. Sometimes the absorbcr will not converge because the liquid stream inside the cor.ver is totally vaporized due ro conract rvith
hor, high-florv vapor coming lrom the reboiler. In this case, the sLrmmation of the reboiler vapor and absorbenr florv should bc
Petrop
us
I I
prSuc ----l----I I I I
Reflux
I
I I I
I
Heat balance
boundary - ---
I
- - - -- - - -- -
-J
raken as rhe overhead llowrare of the rower. Simulation convergence failure may also happen in case ofthe reboiler vapor being totally condensecl by a cold feed/reflrx slream. In this case, the rower ovcrhead rate and reliefload are null.
Overhead energy balance. The rower overhcad vapor flo*,s to the cotdensing systen in which ir is fuily or partially condensed. Thc required reliefload is calculated by encrgy bdance f,r the overrr.ad .y.rer. Fig. J .l'oru. r he.. \enrar i., onligur.,.on ofa rorver over-head system and heat balance boundarv limic. The
'"116*irg l-err bal.:n.e crr' be urirren lor normal oper:rcing conditions:
, .pe.;lc bounJly ",
committed to a sustainable future
NYDROCARBON PROIESS
NG
SEPTEMBER 2OO3
I r::
SAFETY/HEAT TRANSFER
LH:lv'la^Ho.N-4r-Mt , Hr., -M^,*Hr,* - Mr,rHr., =o
M*.n
H
r., (5)
At relieving condition, the heat accumulation is not zeto. Rewriting the equation for relieving condition will give: L,H = Mo,^Ho.r -,y' n - M tpH L,n Mr,"H*,*- M *,*H r,a - Mv.nHv,n
Flooding: Condenser flooding happens when reflu,t or product streams fail, due to electric power or control valve failure causing loss ofoverhead condensation. Ifoverhead accumulator surge time is less tharr 10 minutes, the overhead condensers
will
(6)
streams ro i'hd the ma-ximum required relief rare. If such a scenario s envisaged, it should be noted that the rernaining air-cooler cooling capacity after electric power fiilure, due to a natural convection
effect, is usually 20% ro 307o ofthe normal dury for induced draft air coolers and 10% to 150/o for forced-draft air coolers. lfa watercooled condenset is used, cooling capaciry seems unaffected by power failure; but, since hajfof the cooling water pumps are usually driven by electrical motors, power failure can cause a 5070 reduction in dre warer-cooled condensert dury If all of the cooling water supply pumps are turbioe driven, the condensers will run at iull capaciry when the power is out. Ifa combinarion of air-cooled and water-
{f,$l
x
i
be flooded
reliwing condition. The surge capacity ofan overhead accumulator is calculated based on its fiee volume above normal liquid levels. Vapor blanketing: If the gas product path is blocked as a result of power failure, (e.g., failure ofa downstream gas compressor), non-condensable gases will accumulate in the overhead condenser and prevent condensable material from reaching it. ln this condition, if the safety valve is located on top ofrhe tower, the condenser will lose most ofits condensation capaciry Placing the safety valve on the overhead accumulator will allow noncondensable marerial to be swept out ofthe system when a safery valve opens and the nomal condensation capacity of the overhead system can be taken into account for ihe reliefstudy. The required reliefrate is obtained by dividing the accumulated heat fiom Eq. 6 by the enthalpy ofoverhead vapor, 119 p. As a matter of fact, the amount ofoverhead gas that should be released from the system to remove energy imbalalce is calculated.
,:..,,r power Electric failure.
.. t :',/;i:.i - .-
.'
lJ
..:
Determining rhe relief load in an
eleccric power failure condirion requires a careful plant or system analysis to evaluate what equipment is affected and how the equipment failure affects plant operation. Electric power failure should be anallzed in the following ways:
VEB E IN IGTE F0 LLK0 RP ER - FAB GMBH & CO. KG
R
rKEN
vrww. glnrulofi onrri,ncl
RSI Simcon, an IFP group company. is a leading global supplier of dynamicsimulation solutionsforengineering studies, operator training and plant operation in the Upstream Oil & Gas, Refining, Petrochomicals. Chemicals, LNG, and Power industries.
Our leading platform INDISS , an open environment, is a rigorous and high fidelity tull scope simulation package with extensive first principles based process &thermodynamic librariesOur Operator T.aining Simulators (OTS)are used for training operators on normaland abnormal conditions, for improving plant safety and for optimal plant operation. Our Dynamic Engineering Simulators (DES) are used to validate the preliminary design, ;nvestigate process controllability and operability, and perform va.ious "what-if" studies. Our offices are located in Houston, Grenoble (France) and Beljing.
Ov€r t2OO.irnulaiorr worldwids tor mor€ than 10O cuslom€r3 Select 189 at www.Hydro(arbonProcessing.com/RS
h
with liquid and their condensation capaciry will be negligible at
The required reiiefload is calculated with respect to the condensation capacity ofthe overhead system at relieving condition, 44. Since derermining rhe.ondensarion capaciry in :n up.et condition is not simple due to variations ofoperating paraneters, most ofthe designers prefer to use normal condenser dury for the reliefstudy. This assumption is valid as long as the following situations do not suppress the condenser dury: Electric power failure: Sometimes the designer considers the simultaneous failure of rhe condensing system and feed or reflu-.<
._
cooled condensers are h service, the total condensation capaciry after power cutoff is calculated based on the previously mentioned rules.
Seled 1 90 at www.HydrorarbonPro(essing.com/RS
i;i
i
'''
,:]
i,':
f SAFETY/HEAT TRANSFER
' As a local power failure, where one piece of equipment is alfected . As an intermediare power failure, where one distribution centet one motor control cenrer, or one bus is affected.'fhe equipment affecred by an intermediare power lailure will dispose ofits coorent into the flare header sinulraneously. . As a roral power failure, lvhere all electrically operared equip, ment are simultaneously affected. The affected ecluipment by roral power failure will dispose ofirs conrenr inro rhe flare header simultareously. This is usually the governing case ofaflare system design, especially in oil refineries. For determining tower reliefrequirements, the designer should review diflerenr electric power failure scenarios or any reasonable combination ofemergency cases (e.g., simultaneous failure ofleed and an overhcad gas product compressor) to find the maximum possible column relief rate. Flow cooling failure. All columns
manral valve by an oper:rlor, cooling-water supply system failure or supply pipe rupture.
. Air-cooled condenser failure can happen
because ofrwo main The firsr is fan failure, due ro a mechanical or oower failure that causes loss ofmosr ofthe condensarion capacity. Ii rhis case, 4p will be 20olo to 300/o ofqp for induced draft air coolers arrd 10% to l5o/o of qyfor forced-draft air coolers becau.se ofnanral conveclion effects. The second is louver closure that may result from autom;ric .onrrol leilure, me. hrni.al link;ge iarlLrre or de,rru!ri\e vibr.jrion on a manually posirioned louver Louver closure on air,cooled conreasons.
densers is considered
rotal loss ofcooling capaciry. happens as a result offailure of any componenr in a circuir. It will lead to hear accumularion in rhe system and increase the tower overhead flowrate lhat can be easily pred i.ered by , m ular ing r he rnwer rr r h. rcli.ving contlirion. as a
. Pumparound circuit failure
Absorbent failure. are eqr.ripped
with different
rypes ofcondensing facilities including water-cooled and eir-cooled condensers, pumparounds or a combination ofthose co meet prod-
t specifications, supply reflux and optimize tower and reboiler -oimensions. Failure ofthis equipmenr due ro a cooiing medium or power failure can cause rower overpressure. The above described method can bc utilized for determining reliefrequirements in case ofcooling fiilure. If rhe duration ofcooling fiilure is grcater than the overhead accumulator liquid holdup time, reflu* is losr. -fherelore, the simuladon should be carried out without reflux. . Vatet-cooled condenser failure can tai
Reboiler tube rupture. Reboiler tube ruprure will causc huge amounc ofhor hearing-media (steam, hor water or hor oil) to be released into rhe boiling process liquid inside che reboiler There are some guidelines ud references inA?I-521 Section 3.18 for calcular-
M3 FlexDlementru Texss Inc provides engineertdr, dmpin replocemen8 for early Bendix-type spacers
The American Petroleum Inscirute ApI-
521 Section 3.7 states,4 that generally no relief requirement reslLlts from absorbenc failure. Howeveq in an acid gas removal unit where large inlet vapor quanriries may be rernoved in rhe absorbcr, absorbent loss could cause a pressure rise to reljef pres'ure.,in.e thedown,(re:rr.),terr rral nor bc"Jequ,re ro hanJl. ror LrJ the in.rea.ed flou. I "Lh individual .;,e mu.t be c(Ldied proc\,\ JI d in,rrrrmenrarion .htracreri.ri<..
M3 Technology SIMTOTM
Advanced Scheduling, Planning and Optimization Solutions www.m3tch.com
>
SIMTO Scheduling 0il Refining Petrochemical LNG (liquefaction & regasification)
Terminals
>
SIMTO M-Blend Multi blend recipe optimization Gasoline, Crude, Fuels, Asphalt
Naphtha olefin plant feedstock
>
SIMTO Dock Manager Jetty/berth scheduling
>
SIMTO Distribution Supply & distribution optimization
>
SIMTO Planning Workspace Sales and operation planning
10850 Richmond Ave., Suite 290, Houston, TX 77042 Tel: +1.713.784.8285 o Fax: +1.832.553.1893 Select 191 at www.HydrocarbonProcessing.com/RS
Select'1 92 at www-Hydro(arbonPro.essing.com/RS
f
|
i i '
', '
.
I
i l
I !
rnedia hear content by the larent heat ofprocess liquid. The neit
in reliefcondirion !o
ellect of rhe tower's incernal stream on the reboiler vapor and speci$' rhe overhead vapor flow, ,44p 4. Energy balance around rhe overhead sysrem will give che torver reliefload in the reboiler tube ruprure condirion. see rhe
l j
i i I
i
I I
j
t,,
, 'SAfETY/HEATTRANSFER i '
For tower relief requiremenr during an externsl fire see references 5 and 6.
Control valve failure,
Loss
of instrument air drives all air-
operated valves ro their specified fail posirions. This action may result in overpressure ifrhe specified valvc failrrre posirions were not s€lected ro prevenr overpressure. Likewise, clcctric power failure can drive conrrol sysrems and electrically operated valves to their speciffed failure positions. Failure of conrrol valve mechanical parrs may lead them ro stay -ilueposite oftheir sali positions. The reliefrare to emergency cases caused by controi velve hilure such as feed/reflux control valve closLLre, €las block.-d outler or reboiler sleam -lnd hearer luel gas conrrol valve wide open can be calculatecl considering rhe guidelines here and reference 7. The last rwo cases wiil result in reboiler operation at design duty. Procedures outlined above in the reliefcondirion simulation and overhead energy balance will clerermine the reliefrequirements. lr
ud depressuring systems, 4rh edirion, N{fch 1997. \4oi,d. '. Pd,-ri .,no \ NoruJ/, D,.iHnnsror pr,.rer.t<...<,dunrrp l-'^. -.1, : , h.,ho-.tan r a,. De..n F; 6 ",,,. rclicf valves t001. pp t-_L Ouderkirik, R., "tugorousty si,e fo. ,"pe-i;ricJ fluids,, CE? , for p,rssure rclieving
.
Augusr 2002,
7
pp.341.
Mofrad, S. Rahini, "Relief rate calculation for controt valve faiiure.,,
Hldncarbon Protesing,linuary 2008, pp.
toj
109.
s a process enq neer \a/ rh p€rrofac Constr!ction. He has worked on a variety of fare and reliel system projects, rinq nq frcm simpte flareanctretiefvd ve stud es to detaied hydrau ic engineer nq dergn Mr Rahimi has ako deveoped and taLrghr courses on pressure retef load caicutatron and f are network desgn Hereceved hs BS degree frorn Shtraz LJnrvelsity,lran Eng neering &
and his lvlS degr€e in chernica engineering from Sharif Unrversity ofTechno oqy
NOMENCL{TURT
,4
tl
AH
l.Vl? M t r
Hear exch.rnger roral Jre.r
\.f:rs enthalpy Ac.umularion ofhcrr Log mean renpcrature differencc, calculatcd by Eq. 4 Mass flowmle Reboiler proces side temperarurc Reboile. hearnrg medir side remperarure
z7 Condenser du+ Reboilerduw 14 Hrer .rchr|se. ovcrall herr fickor ar clean condirion
Q
tra
nsfer coef
azr
. Typicolly cuts tumoround decontqminotion time by 507o or more . Deoils, degosses, oxidizes sullides ond reduces wcste
Inler Oudet
in d single step Usuolly eliminotes the need for frac tqnks; reduces temporory piping , Treotment is merged inio shutdowa operoiions Scfe for wqstewcter treoiment sysiems
a, ovcrh.xd slream Y Vllpo. producr srream
Z
'
Liquid produn sr.anr drarv off ream No.malcoDdirion
lYl \%rer
ff R
Relieving coDdirion
LITER,{TTJRE
With on MSDS roting oI 0-0-0 qnd on qll-oqueous, VOC-free, biodegrodoble formulq, Z]'rne-Rowt' UN657 continues to be the environmentolly sole solution ior decontominoting hydrocqrbon processing units worldwide.
It's Eca-nomicqj too.
SUBSCRIPTS
u
CITFI)
I B.alrirrd, M. and D. G. Duffeft,'Avoiding conmon mis'3kes in sizing disrillarion satery valves, Ergin€ting,ltly 9, I984, pp. 78 84.
Zyme-Flow UN 657 is port of c fu.ll rcn ge aI innovolive deconlominqdo! chemisfries and plcnning sen'ices from UIL Contqcl us todqy to lec, l whc:t we can do lot you,
C/rrziral
:Von. K. \1. A. k \n h , d C. c. /nni,,gd. 'lhnnoovrrn.c m.,Fod-.o p, '.u, Fief j.,.m des;gn p,rrnler€r'," ,ry,,/)/ pp. 50.
1l
)
Saeid Rahimi Mofrad
Fxternal fire,
i
I
,t ',
ir_
ing ruprure flowrare. Once the trrbe rupture flowrate is found, rhe designer can c;Jculate rhe reboiler vapor rate bydividing the headngsrage is to simulare the tower
,-.
Zda Eg,ilbna
LJL)y:;:;::r:;nr
241, 2006,
'Mof.ad, S. Rahima xnd S. Norouzi, 'Designing rbr presure relerses during ifts Pan I, H)drod1'ho 1)".6j;/r.(, November 1007, pp. 65-67. 'Arrericln Perroleuni Insrirure, API RP 5tt, Cuidc
WoRLDuIDE
LEADER IN REflNERY DEC'NTAMINATIoN +1 580.252.8500 wwwzymef low.com
Zyme+ow 5. mark of Unred Laboratorie5 nternai onat. LLC o20os
Select 193 at wwwHydrocarbonprocessing.com/RS
,,
l l