Campbell 2 .hal 333-366 Bab 18

18

GLYCOL DEHYDRATIONJ)ehldration rs the process of removing water from a gas andior liquid so that no condensed waterIt'rs present in the system. Inhibition is the process of addin-e some chemical to the condenseci waierso hyCretes cannot form.

Dehydration is generallyvents water from condensing insion/erosion problems. In some instances. inhibition may be the preferred process, particularly in

refrigeration (-40'C [-40'F] and above).sweet systems employing moderate

Natural gas is commercially dehydrated in one of three ways

preferred, if economically and mechanically ieasible, becausc it pre-the system. This water is the source of both hvdrates and corro-

Glycol Dehydration

Mol Sieve, Si l ica Gel, or Act ivated Alumina

Refrigeratirn u,ith Glycol or Methanol lnjection

l .

2 ..3. Condensation

Glycol dehydration (absorption) is the most common dehydration process used to ineet pipelincsales specifications and fielci requirements (gas lift, fuel, etc.). Adsorption processes are used to obtainvery low water contents (0. I ppm or less) required in low temperature processing such as deep NCLcxtraction and LNG plants. Condensation is commonly uscd as a riehydration process when moderatelevels of refrigeration are employed or in pipeline transportation. An inhroitor such as ethylene glycol(EG) or methanol is used to prevent hydrate forrnation. but it should be noted that the actual water ex-traction mechanism is condensation.

Four glycols are used for dehydration and/or inhibition:

l. Monoethylene glycol (MEG) - ofren refened2. Diethylene glycol (DEG)

3. Triethylene glycol (TEG) _-- eor.tzrf4. Tetraethylene glycol (TREG)

Triethylene glycol (TEG) is the most common glycol used in absorption systems.Monoethylene glycol (MEG) is the most common glycol used in ,elycol injection systems. All glycolare hygroscopic, which means they have an affinity for rvater.

to as ethylene glycol (EG) ]-, cf,q.t

to?6:if6n

cla.[.1d*irn

The basicerning the choice

properties of these -elycols are shown in Appendix l8A. The major properties gov-of glycol for a given application are:

Viscosity

Vapor pressure

Solubility in hydrocarbons

CHAPTER 18 333

GLYCOL DEHYDRATiCN

In absorption dehydration systems the solvent (glycol) should be hygroscopic. noncorrosive,

non-volatile, easily regenerated to high concentrations, insoluble in liquid hydrocarbons- and unreactive

with hydrocarbons. COl and sulfur compounds.

Several of the glycols come close to meeting all of thcse criteria. Dicthylene (DEG),

triethylene (TEG) and tetraethylene (TREG) glycols all possess suitable traits. However. almost l00o

of the glycol clehydrators use TEG.

DEG is somewhat cheaper to buy and sometimes is used for this reason. It is also sometimes

used as an inhibitor in addition to an absorbent. However. by the time it is handled. stored, and added

to the units there is often no ;eal savings. Compared to TEG, DEG has a larger carry-over loss. offers

lower dewpoint depression. and regeneration to high concentrations is more difficult. For these rea-

sons. it is difficult to justity a DEG unit, although a few are built each year.

TREG is more viscous and more expensive than the other processes. The only real advantage

is its lower vapor pressure which reduces absorher carry-over loss. It may be used in those relatively

rare cases where glycol dehydration will be employed on a gas whose tcmperattlre exceeds about 50oC

I r 22'F] .

In recent years some units have employed propylene glycol (PG). PG is the least toxic glycol

anci has a lower affinity for aromatics, but PG has a much higher vapor pressure than TEG, and a much

low'er flash point.

This chapter will conccntrate on TEG, even though property data are shown in Appendix l8A

fbr several glycols. Some of the ..;ystem characteristics apply also for all glycols.

T|IE BASIC GLYCOL DEHYDRATION UNIT

Figure l8.l shows the basic glycol unit, regardless of the glycol used. Not shown is any inlet

gas cooling equipment that may be a part of the dehydration unit. When it is possible to cool the en-

tering wet gas with air, water or another process stream ahead of the absorber. do so. Such ccoling is

the least expensive fomr of dehydration.

The entering wer gas, free of liquid water. enters the bottom oi the absorber (contactor) and

flows countercunent io the glycol. Glyccl-gas contact occurs on trays or packing. Bubble cap trays

have becn used histoncally but structurcd packing is more common today. The dried gas leaves the

top of the absorber.

The lean glycol entcrs on the top tray or at the top of the packing and llows downward, ab-

sorbing water as it goes. It leaves rich in water.

It is convenient to use ihe word "rich" to describe the bottom of the absorber and the word

"lean" for the top. At the bottom, both the entering gas and glycol leaving are rich in water; at the top

cnd thcy bot l t arc lcan in watcr.

The rich glycol leaves the bottom of the absorber and flows to a ref'lux condenser at the top of

the stil l column. The rich glycol then enters a flash tank where most of the volatile components (en-

trained and solubie) are vaporized. Flash tank pressures are typically 300-700 kPa [44-102 psia].

Leaving the flash tank the rich glycol flows throLrgh the glycol fiiters and the lean-rich erchanger

where it exchanges heat with the hot lean glycol. The rieh glycol then errters the stil l column where

the water is remi,rved bv distil lation.

334 VOLUME 2: THE EQUIPMENT MODULES

THE BASIC GLYCOL DEHYDRATION UNIT

F. o:x < - otd_ : sE ^ u > i i

F ^ < =< l n ol a Y z

o

a.p

tN

L\

s)

bi

t-

' . t _ .

t i ;a;

< E )>\q

cf.-

:'N : .

\ , o" ) -L l

>' V t

{r L.A\ - .k-- ' t ^ \ r

- v l

>YJ.

c.)o

arl

o q ) v.A; i c L 'd - l ^

t" !'rL t _

- V/ ,r ,@

bgl\i .'tt, A 4.i { , - . ' , '-J , ' - Y-fA-

q

€:a

L. . * <: . \- , . v

l-tr. < I ^

^\ \4

( )zr <p x

I , \ -

i -<. \)6 -

\- 5\4

t t i l I^ O ! .

_fX :N

F E1 4 ui @

v r

@

Fl

a

aaU

Io

Io

I

U TF I

< Fu ar o

oo

uFI.J

5

rciFuI

\-

{at aU Ei P

z L

o o

< @

- - v l

' - ' t l

q-

EoF

J Oo s< z( |o

I

6UF

z

U)

I

I

I

o zr <p x

I >@ -< i

U

? T 6s; i l

) )) )/?\I Y E\-/

T\

Yz

0 rF Uz 9< *U JJ @

0

\

5 t ; 5L > [ [

a

ozdc

EFa

ccFa

. > > . =f i5o**P=37 .'

=

rUJ

@UE

u cF U- )< xU Y

zJ l) )@ o

./--_\ [ 'n

\ v z

I > iI t x

x l1f'(l ' l o

- U: o

a 0 3

: Hii

335CHAPTER 18

GLYCOL DEI'TYDRATION

The still column and reboiler are often called the regenerator or reconcentratoi. This is u,here

the glycol concentration is increased to the lean glycol requirement.

The regeneration unit shorvn is designed to operate at prevailing atrnospheric pressure. The

initial thermal decomposition temperatures of the glycols are shown in the table belorv:

Equilibrium*"Water Delvpoint

@ lqq wq"Fl3'C [37"F]3"C [37"F]-8"C [18 'F ]

'Decomposition Lean Glycol*

Gly.ol _ lgmperature Concentratign' rytfo

EGDEGTEG

165"C [329 'F ]i64'C [328'F]206'C [404'F]

96.09 7 . 198.'7

i _:fBEG::. _23.8'c.t4q:Il >.ee.=--_ =.-.__ ,.-$TIQ"FL. ::* E,quil ibriurn concentration at decomposition temperature and I atm

1i . 1{.-1.c ongl t rq!!q:t! c o I q1Ut:-.g1 1 e fi

These are the temperatures at which measurable decomposition begins to occtlr in the presence

of air. DEG is no more stable than EG because it pyrolyzes in contact with carbon steel.

ln the normal unit containing no air (oxygen), it has been found that one can operate the

reboiler very close to the above temperatures lvithout noticeabie decomposition. The compositton of

the lean glycol is set by its bubblepoint composition at the regenerator pressure. Maximum concentra-

tions achievable in an atmospheric regenerator operating at decomposition temperature are also shown

in the above table.

lf the iean glycol concentration required at the absorfrer (to meet the dewpoint specification) is

higher than the maximum concentrations above. thel sorne- methoci of further rncreasing the glycol

concentration at the regenerator must be incorporated in the unit. Virtually all of these methods in-

volve lowering the partial pressure of the glycol solution either by pulling a vacuum on the regenerator

or by introducing stripping gas into the regenerator.

A typical str ipping gas systern is shown in Figure lS.2(a). Any inert gas is sui table. A part of

the gas being dehydrated. or exhaust from a gas-powered glycol pump (ii used), is suitable. The quan-

tity required is small. The stripping gas may be introduced directly into the reboiler or into a packed

"stripping coiumn" between the reboiier and surge tank. In theory, adding gas to a packed unit be-

tween the reboiler and surge tank is superior and will result in lower stripping gas rates. If introduced

directly to the reboiler. it is common to use a distributor pipe alcng the bottom of the reboiler.

A seconcl stripping gas alternative is to close the stripping gas loop and use a material like iso-

oclane, as shorvn in Figure 18.2(b). It vaporizes at reboiler tcmperature bui can be condensed and

separated from the water in a three-phase separator. The stripping solvent is then pumped back to the

regenerator to complete the stripping loop. Sold under the trade name DRIZO@, this unit has the aci-

vantage of providing verlv high stripping gas rates with little or no venting of hydrocarbons. Glycol

concentrations in excess of 99.99 wt%o have been achicved with the DRIZO@ process. It has an added

ailvantage of crnciensing and recovering aromatic hydrocarbons from the stil l column cverhead. In

practice, these units often operate with a stripping solverri. which is not iso-octane but a mixture of aro-

matic, naphthenic and paraffin hydrocarbons in the Cs-Cro range.

Figure 18.2(c) shows a third regenerator al ternat ive cal led a COLDFINGERE. The

COLDFI)JGERI process achieves glycol enrichmcnt by passing a cooling medium (often the rich gly-

col) through a cool "finger" inserled in the surge tank .,'apor space. This condenses a water-TEG


THE BASIC GLYCOL DEI-iYDRATION UNIT

(rr ttf{ '+ Qg,? 1 rtlTv'f, f ii,8v"ntc'r""X;:l"t'*"

Qctto -----e -s t'r'r'-"1.(

a) Str ipping gasVent Gases to Flare

or Recycle Flue Gas

DRlzo'SEPARATOR

Cooling Medium

Water RichTEG MixtureTo Stil lColumn

b) DRtzo' c ) COLDFINGERo

Figure 18.2 TEG Regeneration Alternatives

mixture which is rich in water. This mixture is drawn out of the surge tank by means of a trough be-low the "coldfinger" and is recycled back to the regenerator. The H2o partral pressure in the surgetank vapor space is thus lowered and the lean glycol concentration increased. It is claimed that leanTEG concentrations of 99.5-99.9 wto% have been achieved in COLDFINGER.& units without the use ofstripping gas, although ' small amount of gas is introduced into the surge tank for pressure balancing.

The unit shown in Figure l8.l is typical. Figure 18.3 shows an example of an offshore unit.The inlet scrubber is in the bottom of the absorbei- Three-phase separation is required. The gas risesthrough a "chimney tray" to the absorber. The hydrocar-bon and water are separated as shown.Three-phase separation saves on deck space antl is less expensive, but many of the existing units areunsatisfhctory because they provide inadequate separation. This is paticulaily true when the absorberutilizes structured packing.

Glycol Pump

CHAPTER 18 337

GLYCOL DE!.IYDRATION

Lean glycol

Hydrophobic

+Vy'aler lc sea

B Separator-Contactors SumP Tank

16 GlYcol PumPs(8 operating 8 spares)

3 Glycol Regenerators

Steam tovent stack

I , 1'"n.9't'o't t

waler lree Igasolne I

\ I Gtvco l Pump rT*l Suciion header

Rich giycolheader

Lrnalog launch barrel

Sphere launcher

Gas to vent slack

Reclaimed gasolrneLean glycol

Pipel ineto shore

(400 NlMsctd)

8 Wel ls(50 Mt\.4scfd each)

Fiqure 18.3 Exampie Offshore Glycol Dehydration Installation

\The rich TEG from the chimney tray goes to a degassing pot (flash tank) which is operated at a

high enough pressure to send the gas to fuel or recompr'ession. In some systems the pressure is suffr

cient merely to enter the main flare system. The purpose of this is several fold: (l) use or dispose

safely any volatile components picked up by the TEG in the absorber, and (2) minimize the presence of

corrosive sulfur compounds and carbon dioxide in the high temperature reboiler.

The true solubilrty cf paraffin hydrocarbons is very lorv in the glycols. But, separator carry-

o1,er and entrainment does introduce hydrocarbons into the rich glycol. Many of these are "heavier"

than air and can be a safety problem unless disposed of properly. In addition. aromatic components are

very soluble in TEG. These can also be a saf-ety and environmental concem when discharged to atmo-

spherc at the top of the regenerator.

Both sulfur compounds and carbon clioxide are soluble in rvater and react to sorne degree with

the glycols. The degassing in the flash tank prior to the stripping column reduces their concentration

in the glycol somewhat. This degassing is more efficient if the rich glycol is first preheated, usually in

the rich-lean glycol exchanger.

The glycol cooler is a gas-TEG exchanger in Figure 18.3, as opposed to cool ing medium in

Figure 18.1. This is a common method of cool ing the lcan glycol. I t is s imple and inexpensive and

"nirr., the glycol enters the contactor warmer than the gas. The gas-glycol exchanger can restrlt in

excessively high lean TEG tenlperatures at low gats flowrates'

TJie r ich- iean exchanqer in Figure 18.3 is a coi l in the surge tank. This is inexpensive to bui ld

but often results in poor heat transfei'arrd higher reboiler duties. Pipe-in-pipe exchangers are widely

used in this service.


MINIMUM LEAN TEG CONCENTRATION

Fi-'rre 18.3 shows a gas tired reboiler. The usc of hot oil, stearn, waste he.ii or elcctrical resis-tance cci is l re al l sui tabl : i l ' thci are rc".di ly a. . 'a i labie at the s! te. Thc use of electr ical resistance heat-ing on oflshore installatiotts is ctrst-ef-fective and more convenient if area ciassification does not allowa flrcd heater.

No filter is shown in li igure l8 j. I prefer locating the particulate fi!ter at sorne point ahead ofthc reboi ler to minimize the'-gunk" accunrulat ing therein. Foieffcct ive operat ion i t is imperat ive t f iatfui l - f low. glycol f i l ters bc instal led in the system.

BASIC PROCESS DESIGN FACTORSAll factors controi i inc thc behavior of absorpt ion systems also apply for TEG del iydrat ion. In

fact, front a process viewpoint. TEG is one of the sirnpler absorption processcs being cmplcved in thepetroleum industry.

To properly design a unit one nceds to know maximum and minimum gas flow rate. lnaxinrurrrand minimum temn.rature anti pl'cssure. qas colnposition and required water dewpoint or \\,ater contentof the out let gas. From thesc one can calculatc:

l. The minintwtt cott(etllt 'utiott o1- TEG in the lean solution entering the top of the absorber

, required to mect out lct gas water specif icat ion.

I Z The lean TEG circulation rate requircd to pick up the necessary amount of water fiorn thc' gas to lneet thc outlet gas watcr content specification.

3. The amount of abscrbcr coirtact required to produce the nccessary approach to equilibriurnrequired in ( l) above at thc chosen circulation rate.

To obtain these answers i t is l recessary to have a vapor- l iquid equi l ibr ium corrclat ion fbr aTEG-water system. Frorn this basic input, one can evaluate or size equiprnent an<l devclop mechanicalspcci f icat ions.

The procedure that follows is straightforward and can be perfonnecl manually. In all but a fewexccpt ional appl icat ions, i t wi l l g ive rcsults as rel iable as more r igorous methods. Fol lowing an out l ineof the basic calculat ion procedurc. each major equipment cornponent wi l l be reviewed.

MINIMUM LEAN TEG CONCENTRATIONI f rvater-saturated gas is placed in a stat ic cel l with a given concentrat ion of TEG-u'ater solu-

tion at a fixed P and T, equilibrium would be attained in time. Assuming the liquid had a sufficientlylow rvater concentration, waicr rvould transfer to this liquid frorn the gas. At equilibrium. the mol fi-ac-t ion water in the gas divided br i ts mol fract ion in thc l iquid equals the K value fbr this systcm.

Figurcs 18.4(a & b) are based on equi l ibr ium data publ ished by Parr ish, " ,

o1.t l ' \ t t Severalequi l ibr iurn comelat ions(tE t ' t ' \ ' " havc been presented since 1950. Previous edit ions of the Car-npbel ltexts and the GPSA engineering Data Book presented an equilibn;;n comelation based on the work ofWorleyll ' l. l i. In general, the corrclations of Worleyr/8:/, Ros,1run/i'\ -ti and Parrish/1s // agree reasonablywell and are adequate for most 1l'G system designs. All are limited by the ability to accurately mea-sure the equilibriurn concentration of water in the vapor phase above TEG solutions. The parrish cor-relation has been included in lni- gdition because equilibrium water concentralions in the vapor phasewere determined at inf ini te di lutrorr (essent ial ly 100% TEG). The othercorrelat ions use extrapolat ionsof data at lower concentratiorr trr cstimate equilibrium in the infinite dilution region. The effect ofpressure on TEG-water equi l ibr iunr is smal l up to about l3 800 kpa [2000 psia]/r8+,t .

CHAPTER 18

-iF G

339

GLYCOL DEHVDHATION

40

zv

;c

=q)

U . Z U

:J

, =s -40.LU

-bu

/al -80

3020 40

Contactor TemPerature, "C

100

80

60

40

20

99.0

99.5

u-;c.

=o

f' E

:=

tI l

-20

-40

-60

-80

99.9

99.95

99.97

99.99-1 00

1 0 0 1 1 0

Contactor TemPerature, "F70

Figure i8.4 Equil ibrium HrO Dewpoint vs. Temperature at Various TEG Concentrations

120

I

t E u

Concentration.wt% :::::4"-"'-95.0

'JZ:=-99.0

99.5

e e 9

99.95'-

99.97

99.99

-

-

Hl l

l i

VOLUME 2: THE EQUIPMENT MODULES


A TEG absorber is essent ial ly isothermal. The hcat o1'solut ion is about 2i l k l /kg [91Btuilbm] of water absorbed in addition to the latent heat. But. the mass of water absor-becl plus themass of

'fEG circulated is small relative to the mass of gas. So, the inlet gas tel.nperaturc controis.

The tetrperature rise due to heat of absorption and heat transfcr betu,een the glycol and the gas seldon-rexceeds 1-2"C [2-4"F] except when dehydrating at pressures below about 1000 kPa [145 psia]. ln lorvpressure sei.vicr siii.ne telnperature adjustment may be desirable.

Example 18.1: What equi l ibr ium water dewpoint could be obtained at 40'C [104"F] wirh a leanglycol solut ion containing 9q.5 wt% TEGI

In F-igure 18.4, locate 40'C [04'F] on the abscissa. go vertically to the 99.,5 1y19'ul ine and then horizonral ly ro the ordinare. Read - i9.C [-2.F].

This water der,vpoint could be attained in a test cell but not in a rcal absorber. The gas andTEG arc not in contact for a long enough time to reach equilibrium. ln addition, the gas thcoreticallyleaves the top tray of the absorber in equilibrium with the TEG leat,ing the tray, not entering. Numer-ous tests show that a well designed, properly operated unit rvill have an actual water dervpoint 5-lO"Cl[9-18"F] higher than the equi l ibr ium dewpoint. This "approach" to equi l ibr ium depends on rhe glycolcirculation rate and number of contacts in ihe absorber and is used to specify rninirnum lean glycolconccn#at iorr . Thc procedure is as lol lows.

I1. Establish the desired outlet water dewpoint needed from sales contract specifications or

lrom mininrum system temperaluie.

2. Subtract the approach from (l) to find the cor.--sponding equilibrium u,ater dewpoint.

Enter the value in (2) on the ordinate of Figure 18.4 and draw e horizontal l ine

4. Drerw a vertical line from the inlei gas temperature on thc abscissa.

5. The intcrsect ion of thc l ines in Steps (3) and (4) establ ishes rnininrum leal TEG conceptra-tron required to obtain the water dewpoint in Step (1).

If rl 'ater conteltt is specified or calculated in rnass per unit gas vclume, a warer oontent. pres-sure, dewpoint temperature correlation is required (see Figure 6.1 in Volurne I or Figure l8C.l in Ap-pendix l8C). Notc that the equilibrium water dewpoints on the ordinate of FigLrre 18.4 are based onthe assunlption the condensed water phase is a metastable liquid. At lorv dewpoints the true condensedphase r.vill be a liydrate. TIiu equilibrium dewpoint temperature above a hydrate is higher than thatabove a metastable liquid. Therefore, Figure 18.4 ma1, predict dsu,points which are lorver than can ac-tually bc achteved. The difference is a function of temperarure, prcssrrre and gas composition'but canbe as much as 8-12'C [15-20"F]. Figure 6.5 shows the relat ionship between the metastable and hy-drate dewpoints for one gas. When dehydrating to very low dewpoints, such as those required up-stream of a refiigeration process, the TEG corrcentration should be sufflcient to drv the sas to thehydrate dewpoint and not just the metastable dewpoint.

The dashed line in Figure 18.4 at about 98.5 wto/o represents the concentration of lean TEG thatcan be prodr-rced routinely in a regenerator operating at stanclard atltospheric pressure and 204"C[400'F]. This is a safb value for design and specification purposes. Concentrations of 98.7-98.8 wt%are common; some to 99. I wtgb have been reported but represent a special case where incoming hydro-carbons provided natural stripping and/or the atmospheric pressure u,as lower than l0l kpa [14.7 psia].

Since the initial capital cost of ordinary gas stripping accessories is small, they always shouldbe included. conditions can chan-ee to where they may be required.

CHAPTER 18 341

GLYCOL DEHYDRATION

Example 18.2: The gas sales contract specifies an outlet $'ater content of 100 kgi lOb std rnr [6lbrn/MMscfl at a pressure of 69 bar [000 psia]. l 'he inlet gas temperature is40'C. What minimum lean TEG concentration is required?

SI Solution: For 100 kg/l0" std mi and 69 bar, the equivalent dewpoint from a watercontent correlat ion (Figure l8C.l(a)) is -2oC. I f we use an 8"C approachthe equi l ibr ium dewpoint is -10'C. 'From Figure 18.4(a) at -10'C and40oC contact temperature, lean TEG concc'rtration :99.0 wtoh

FPS Solution: For 6 lblMMscf and 1000 psia the equivalent dewpoint from a water con-tent correlat ion (Figure lBC.l(b)) is 28"F. An approach of l4"F gives anequrlibrium dewpoint of 14"F. From Figure 18.4(b), lean TEG concentra-tion equals 99.0 wto/o.

It is necessary to fix a lean TEG concentration for subsequent calculations. For the first cor-r-sideration, use the results from Figure 18.4. If the concentration obtained is less than 98.5 wto%, use98.5 wt% for the calculation unless you plan to reduce the reboiler temperature below 204"C [400"F].

The minimum lean TEG concentration may not be the one eventually used. A higher concen-trat ion thap this may bc specif ieC to minirnize circulat ion rate and opt imize cost.

tI

Absorber Design

The absorber (contactor) is where the water is removed from the gas by the proccss of physicalabsorption. The amount of water removed depends on three factors:

. glycol concentration

. glycol circ;lation rate

' number oi contacts in absorber

The effect of glycol concentration on v"'ater removal is il lustrated in Figure 18.4. As the TEGconcentration increases, the equilibrium water dewpoint decreases. In an absorber this would mean alower water concentration in the gas at the absorber outlet, hence higher water removal.

In a real absorber the gas leaving the absorber does not reach equilibrium with the incominglean TEG. This is the reason for the "approach" discussed in the previous section and applied in Ex-ample 18.2. Approach depends on two parameters - circulation rate and number of contacts. Thetheoretical outlet water dewpoint (water content) is set by the lean glycol purity. The actual waterdewpoint depends on the circulation rate and numher of trays or height of packing. This is il lustratedin F igure 18 .5 .

As you can see in Figure 1tt.5, adding contacts or increasing the circulation rate increases w.r-ter removal. However, the curves in Figure 18.5 reach an asymtotic value of about 0.94. This watcrremoval represents the equilibrium dewpoint condition at the top of the contactor. If we have an infi-nite number of contacts or an infinite circulation rate, rhe waicr removal percentage would stil l bc 94%and the approach would be zero. At this point, the water removed depends exclusively on the leanTEG concentration. Our goal is to design a glycol system which is economically viable. We have sev-eral optinns. We can select a contactor with several contacts and a low circulation rate or one with fewcontacts and a high circulation rate. In fact, there exist an infinite number of chorccs to meet our out-let dewpoint specif i cat ion.

342 VOLUME 2: THE EQUIPfuIENT MODULES


0.80

0.75

0.70

0.65

TEG Circulation Rate, liters TEG/kg HrO

Figure 18'5 water Removal vs. TEG Circulation Rate for Several Contactor Theoretical Stages(99.0 wt% TEG)

Il/here:

t l

o

Eq)CEq)(s=

c'F

(g

lr

Most designs use a circulat ion rate of l5-40 l i ters TEC/kg H2o absorbed [2-5 US gal l lb Hroabsorbed]' This is near the economic optitrtrm. Higher circulation rates result in a lar-eer regeneratio'system, higher energy consumption, and higher coabsorption of aromatic hydrocarbons. Lower ratesrequire a taller contactor and may result in poor tray/packing hydraulics.

The relationship bet'*'een circulation rate and the number of contacts can be quantified by useof the Kremser-Brown equation (see Chapter 17), a shortcut absorber calculation.

, _ Y N * r _ Y r A N ' l _ A= _

" Y N * r - Y o A N * l - l

Eu = efl iciency of absorptionYr-r : mols of rvater in entering (wet) gas per mol of gas enterin_e

Yl : mols of water in leaving (dry) _eas per mol of gas enteringYo = mols of water in equil ibrium rvith the incoming lean glycol per mor of gas

entering

A = absorption lactor, A : L,VKL : glycoi circulation ratc, tnols/unit t irneV : _eas tlorv rate, rnolsiunit timeK = vapor liquid equilibrium ratio (y/x) for water in a water-gas-TEG systemN : nunrber of theoretical contacts in absorber

(18 . r )

CHAPTER 18343

GLYCOL DEHYDRATION

In TEG systems, the mol fraction, y, may be used instead of the regular absorption parameter,Y due to the concentrations involved.

This means thst the left hand sidc of Equation 18.1 becomes

Y r " ' + l - Y l

Y N + t - Y o

where "y" refers to the mol fractron of water rather than

Fnrther, since the mol fraction of water can bevolume by a fixed conversion factor,

a rate per mol of incomine gas.

converted tc mass of $'ater per standardgas

w=Bys ,o

W6

I(here.

Equation 18.1

B

v

ma

= conversion factor: mol fraction H2O

y be rewritten in terms of W.

._9r .kg' /10 ' s td ml

76 r 000

_ w N * r - w r A N - l - A" 3

w N * r - w o A N + r _ l( I 8 .2 )

WN*t

w lw o :

rvater content of entering (wet) gas

water content of leaving (dry) gas

water content of gas in equil ibrium withincoming lean TEG

TEG concentration and contactor T and

kg/ I 06

kg/ I 06

kg/ I 06

P.

std ml

srd ml

std mr

FPS

lbm/lv{Mscf

lbm/MMscf

lbm/MMscf

Wn is a function of the lean

Figure 18.6 is a plot of Equat ion 18.2 and is convenient fbr manual calculat ions. This useswhat could be called an overall absorpiion .factor. L6 is the rate of lean TEG entering the top tray andVy11 is the gas rate entering the bottom tray. Even though the (L,^/) changes slightly throughout theabsorber the effect of these changes are effectively canceled by changes in the "K" value, so use of(LglVN+lK) as the average absorption factor has little effect on tne accuracy of the method.

Thc lcft-hand ordinate of Figure i8.6 is called absorption elJicienclt o/'absorption, Eu. It is theactual amount of water removed, divided by the maximum amount theoretically removabie. The val-ues of N encompass the range of theoretical stages usually employed in TEG contactors.


MINIMUM LEAN TEG CONCENTRATTON

=t=, t '- l

+ l :Z l 2

= l=.;

t!J

4

)t "(o (huut

Efficiency of Absorption, Eu vs. Absorption Factor, A

calculation of Lean TEG Rate for a Given Absorption Efficie'cy and N _

l. Calculate yo (or Wo;

2_ Determine absorption efficiency

3' Use Equation l8'2 or Figure 18.6 to find absorption factor A fbr a given value ofN

4' Knowin-e Vp'1 and K, solve A for L6, the lean TEG circulation ratecalculation of N for a Given Lean TEG Rate and Absorption Efficiency _

l. Calculare yp (or W6)

2. Determine absorption efficiency

3. Calculate absorption factor A

4. Determine N from Equation lg.2 or Figure l8.6

It is usual to repeat the calculation to obtain three lean glycols rate/absorber contact values thatsatisfy the required absorption efficiency. The final choice is based on economic considerations. Thisusually involves sclection of standard desisns.

This calculation should be made at the lowest pressure a'd highest temperature anticipated forthe entering wet gas, to obtain the maximum water loading. Unfortunately, the tendency is to choose adesign temperature lower than that actually incurred.

CHAPTER 18345

GLYCOL DEHYDRATION

The overall tray efficiency in a well-designed TEG unit.;vill vary from 25-40%. lt is recom-

nren,Jed that25o/o be used for most applications. This provides an affordable safefy factor to help com-

pensate for the inherenf errors in the design specifications.

tquilibrium RelotionshipsVarious studies have been made of the equilibrium behavi,or of water in the TEG-water sys.

tem.(t8I' 18 3'18'6) All provide rather consistent data. The use of an activity coefhcient (Y) is a conve-

nient and reliable method for calculating K. Using this relationship

K=(y*Xy) = IY!

K : equilibrium constailt for water in a TEG-water system

y* : mol fr. water in the gas at safuration over 100%o liquid water (from regular water

( l 8 .3)

Where:

content correlation)

T : activity coetficient for water in the TEG-water system as found from Figure 18.7

w : water content on a mass per volume basis, at safuration, as found from a regular

?/ water content correlation

B : 761000 when W : kg/106 std m3 47 400 when W : lbm/MN{scf

0.85

0.80 I- T T - l - T - r T - t f f i l l l

t J . t 5

C

:0q)

=()

0.70

u.o3

0.50

i ra$ o,{ -t90

a,6)<M91 92( r ,dg

93 94 95 96

TEG Concentration, wt %

97 98 99 100

Qt4- -i . .ure 18.7 Activity Coeffic;rnt for HrO Concentration at Various Temperatures(t8'7)


MIN|MUM LEAN TEG CONCENTRATION

Notice that y (and thus K) varies with TEG concentration, which in furn chan-ges across the ab-sorber' So, a ntean or average K cani.;.-,t be determineC until the cirrulaticn rate is fixed. This in'clvesa trial and error calculation. in most cascs both L and K are determined at top tray conditions, i.e. usethe lean TEG concentration and lean TEG circulation rate. As stated previously, the increase in Kfi'om the learr to rich TEG is roughly proportional to the increase in L/v, so the absorption factor(A : L/vK) remain:; relativerv uncharrgeci thrcughout the absor,ber.

In the absorption etficiency term.

yo = K xc and W, = (W)(y ) (x6 )xe : mol li water in the lean TEG enterins the absorber

This may be calculated from Xol., the weight percent TEG in the lean solution entering the ab-sorber. This must be not less than the minimum'aiue required from Figure 1g.4.

^ 0 -

l0r - arLl 8

i o0 -xs l

l 8

( l 8 .5 )

Where = mol fr. of H2O= wtu/o TEG in lean glycol

94 95 96

TEG Concentration (Xn1), wt%

Figure 18.8 Mol Fraction HrO, xo, vs. TEG Concentration, Xo,

( r 8.4)Ilhere:

X o t

lstX9

Xet

xri

N

Ic

.F

(g

tL

o

Equation 18.5 is shown graphically in Figure 1g.g.

CHAPTER 18347

GLYCOL DEHYDRATION

The above procedure using the Kremser-Brown approach gives results comparable to computersinrulat ion.

The use of other equilibrium K values will have some effect on contactor design. The requiredlean glycol concentrations may differ but the difference is normally less than the random error in pro-cess specifications and is statistically insignificant.

Example 18.3: Calculate the circulation rate of 98.7 wtolo lean TEG needed to dry 106 std m3/d

[35.4 MMscfd] of gas at 7.0 MPa [000 psia] and 40'C [04'F] in a six tray ab-sorber (1.5 theor. stages) to achieve an exit gas water content of 117 kgl0'std m'

[7 lbm/MMscf]. The inlet water content is 1100 kg/106 std m3 (saturated gas)

[68.s lbmA4Mscl].

l . From Equation 18.5, xo : 0.099

2. From Figure 18.7, y : 0.66

3. W is the water content of saturated sas at 7.0 MPa and 40oC or li00 ke/106 std m3tnls case.

4. From Equation 18.4, Wo : (1100X0.66X0.099) : 71.9 kg/106 std m-l

5. The left-hand side of Figure 18.o is:

(1100 - 111) / (1100 - 71.9) : 983/1043 : 0 .9s6

6. From Figure 18.6, for N : 1 .5, A: 7 .3

1. Lo : (AXK)(VN*r)

f l 100)(0.66)From Equation 18.3, K =

761-000 = 0.000 954

8 .

9.

10 .

1162kmollh

So, Ls : (1.3)(9.54 x 10-4X1762):12.3 kmolrh

MW lean glycol : (0.099X18) + (0.901XI50) : 137 kgikmol

kg TEGIh : (r2.3)(r37) : 1685

Density of TEG : 1.12 kg/liter

Circulation rate is 1685ll.l2 :

In one hour ( l100 - 117)124 or 41.0 kg H2O is absorbed

Circulation ratio is l504l4l :36.7 liter TEG/kB HzO absorbed

In FPS units the calculation follows the same format.

| & 2. The same

3. W: 68.5 lbm/MMscf

4. w0 : (0.66)(68.5X0.099) : 4.48lbm/N4Mscf

5 . (68 .5 - 7X68.5 - 4 .48) : 0 .960

6 . A : 1 . 5



Example 18.3 (Cont 'd):

t 68 .51(0 .66 ) . . ̂t . K - _ = 0 . 0 0 c 1 9 - 5 447 400

V : (35 . -+X l l0 r = 3894 lb rno l ih r

i -0 - AK\ / \ .1 = (7 .5 t f .54 x t0 +x3894) - 21 .g lbn io i r i r r .

8 . M W : 1 3 7 l b m l b r n o l

9 . l b m / T E G / h r : ( 2 7 . 9 \ ( 1 3 7 ) : 3 E 1 7

10. Density of TEG is 9.33 lbm/US gal

Circulat ion rate is l8 l7/9.33 - 409 US gal/hr

In one hour a total of 90.7 lbm of warer is absorbed

Circulation rario is 409190.7 : 4.5 US eal TEGilbm u,ater absorbed

When using the Kremser-Bro'uvn method, the tenns V and L must be expressed in rnolar units.This requires calculation of the MW of the TEG solution. The molecular weieht of a TEG-water solu-tion may be calculated as follows:

M W = l 8 x u + 1 5 0 ( l - x 6 )

The actual circulaticn rate used intions always dif!'er to some degree fiompumps come in.discrete capacities. It is

FPS Solut ion:

the unit may be different from this becausethose specified in the design. In adtlition,sound practice to choose other components

( 1 8 . 6 )

operating condi-TEG circulationbased on the ca-

pacity of the circulation pump.

For the previous example , would we buy a 1..-5 theoretical tray absorher? Probably not! Thccirculation rate caiculated is toward the high end of the economic range. An absorber rvith 1.75-2.0theoretical trays might well be specified to provide valuable flexibility and inexpensive "insurance."

The shortcut calculation method presenied in this secrion is stil l somewhat tedious to do byhand. Appendix l88 presents graphical solutions to this shortcut method taking into account the waterremoval, circulation ratio, lean TEG concentration and number of theoretical stases in the contactor.

Example Rework the previous examplel 8 8 .

SI Solut ion: [-ater removal =Win

From Fisure 18B.2 (N : 1.5), at

The circulation ratio is 34-35 Iitevalue calculated in Example 18.3.

water removal = w'n - v

From Figure

Win

(N : 1 .5 ) , a t

The circulation ratio is 4.3-4.4calculated in Exarnple 18.3.

mple using the absorber performance curves in Appendix

win -wuur = l_.]!9:_l_LZ

=0.g94w i n 1 1 0 0

1.5), at 98.7 wt% TEG

4-35 Iiters TEG/kg H2O which is very close to the 36-37p le 18 .3 .

w i n - w o u , _ 6 9 . 5 - 7 : 0 . g g g

win 69.5

1.5), at 98.7 wt9lo TEG

3-4.4 US gal/lb H2O which is very close to the 4.5 value

CHAPTER 18 349

GLYCOL DEHYDRATION

Absorber DesignThe design of the absorber (contactor) is based ol.t two parameters:

1. gas rate which determines the contactor diameter, and

2. number of contacts. which determine the contactor height.

In some cases the contactor may also contain an integral scrubber designed to remove entrained drop-lets and solids fiom the gas prior to entering the absorption section. ln addition. an internal gas-glycolexchanger is inclLrded at the top of the contactor in some designs.

The contactor dianteter depends almost exclusively on the gas rate and is virtualiy inclependentof the glycol rate. This is due to the lorv liquid loadings ernployed in glycol contactors. Two tvpes ofcontactor internals are used in glycol systen-ls - t rays (usual ly bubble cap) and packing (usualh'struc-tured).

Bubblecap trays have been histor ical ly used in TEG s) 'stems. This was certainly the casc unt i lthe rnid-1980s. Since then, many contactors have been instal led with structured packing. This is pr i -rnaril,v due to thc superior gas handling capacity of structured packing relative to trals. Suirl TubeTrays. a proprietary Shel l technology, have been recent ly introduce6.rtat t Shel l c lairns signi f icant ca-pacity increases over stnlctured packing.

When trayed contactors are used. bubblecaps are preferred over other types of trays (r.alve andsieve ) due to their higher turndown ratio and generally better efficiency at low liquid rates. Likervise,when a packed contactor is employed. structurcd packing is preferred over random parking because ofits higher capacity, better turndou; and superior performance at low liquid rates. Randorn packing issometimes used in small diameter contactors (less than 0.6 m 124 inl) for convenience.

Figure 18.9 shows a bubblecap tray, section of structured packing and random packings.

BubblecapTray

PallRings

Figure 18.9 Bubblecap Tray, Random and Structured pac.king


MINIMUM LEAN TEG CONCENTRATTON

ConlaCor Diomeler

As stated earlier, the diameter of a glycol contactor is determined, almost exclusively by thegas rate. If a contactor is operating near flood, changes in the giycol rate can have a noticeable effecton the glycol canyover, but for design purposes the liquici rate i: typically not a factor in absorber siz-ing.

The calcuiation of diameter can proceed two ways. The frrst method employs the SoudersBrown equation (Equation I l.l I ), frequently used to size separators.

r i 0 . 5

V:K,IeL _es

I\Ps )

- S I ]allowable gas velocity .it

-

lIt : sizing parameter:

?*r'rti,t"ooto".o,r, , J;lJirT'L, Ipg = gas density kg/mi

Ipr = liquid density kg/mr Ipr : for TEG systems 1120 kg/mi I

( i 8 .7)

Where.

The calculation of the diameter follows

( 1 8 . 8 )

ftlsec

0.18 fusec0.30-0.34 ft isec

lbm/ftl

lbmifts

69.9 lbm/fC

d= -Fq^\ / n 'n

Where;

The act'-ralmetric rate.

d = contactor diameterga : actual gas flowrate

gas flowrate can be calculated from

:r+mr/s I

either the mass flowrate or the

_:i_!kg/s

kgim3

ft

ft3lsec

stanciard vclu-

( t 8.e)

( 18 .10 )

m9 a - -

nr g

Were.

Ll'here

m

p"

mass flowrate

gas density

o,=#*d(+)[+)'Qstd

Prtd

p

Tu

T.td

gas flowrate in standard volumes

standard pressure

actual flowing pressure

actual flowing temperafure

standard temperature

gas compressibility factor at florving conditions

scf/day

psia

psia

"ROR

FPS

CHAPTER 18 351

GLYCOL DEHYDRATIOI"I

A second sizing equation uses an F. value which is related to the kinetic energy of the gas

1pv21. this is the more popular method for sizing packed towers.

( 1 8 .1 1 )F.

ESI

Where.

The sizing parameter, F' depends on the fype of packing and the packing density, but for most struc-

tured packinBs, F, : 3.0 in SI units [2.5 in FPS units].

It should be noted that for contactors containing structured packing the gas handling capacity

may be limited by the mist extractor, not the packing. This is particularly true when the glycol viscos-

ity exceeds l5-20 cp. This is the approximate viscosiry range of TEG at 32-38'C [90-100'F].

ln an actual design it is sound engineering practice to size the contactor for a gas rate 20-30%

higher than the expected rate. This contingency provides contactor capacity for changes in flow rate

and pressure and for pessimistic reservoir engineers.

allowable gas superficial velocity

F, - sizing parameter

mls

Pao 5

Example 18.5:

SI Solution:

A glycol contactor is to be designed to handle I x 106 std ml [35.4 MMscfd] ofgas at 40"C and 70 bar [015 psia]. The gas compressibility factor is 0.85 and the

MW : 19.0. Size the contactor for both bubblecaps and strucfured packing.

1) Calculate the gas density

(7000)(1e) --60 kgf ml

2)

3 )

4)

Calculate allowable velocity. v

Bubblecaps. v:K,[ql : !s- ] =o.or, (1t29:60' | =0.23mls" ( . Ps ] \ 60 )

Calculate the actual volumetric rate, qa

q. = jrl r!*)l,]-],,,:' 990 990rl!r-)r..)ro $rrd 86 400 [ P, / | T"a J '

' 86 400 \ 7000/\ 288 )

= 0 . 1 5 4 m , / s

Calculate contactor diameter, d

=0 .92m1(0.23X3.14)

(0.8s)(8.314)(3 r3)

5) For structured packing

Fs 3 .0

h

352

J% J6o= 0.39 m/s


*{1tr#rifin


Example 18 5 (Cont'd):

=0 .71 m

FPS Solution: l) Calculate the gas densitv

(P)(Mw)= - =

( l0 l s ) ( le )=3.J5 lbm/ f t1n

zRr (0.85)(10.73)(564)

2) Calculate ailowable velocity, v

Bubblecaps, v =K. (p ' ' _ p"

lo t =0. , s (eq '2 -1 ' ts ) =0.756f t lsec' l Pg I \ 3 '1s )

3) Calculate the actual volumetric rate, qa

^ ( ' , \ ( T , ) 35 .4xnb ( t 4 .7 ) f 564 ) . ^q . , =E l

r s t u ' l - l l z l : : | _ l l - ^ ^ l (U .85 )r : E6 4oo I P " r I T "o i 86 400 \ 1015 ) \ s2o )

= 5.41 ft ' /sec

Calculate contactor diameier. d4,\

d - =3 .0 f t

s) For strucfured packing

F . 2 .5\ = . = = l . j n i s e c

l n . 1 7 7 51 r g v - ' '

, { _u -

\\

Conluclor Height

The contactor height is determined by the number of equilibrium contacts required and effi-

ciency of the mass transfer. For trayed contactors the conversion from equilibrium stages to actual

trays is accomplished by using a tray efficiency. The tray efficiency is measure of the approach to

equilibrium and can be calculated from either vapor phase or liquid phase compositions.

The overall tray efficiency is defined as follows:

tr" 0 \ c r l l l

No. of Equilibrium Stages

No. of Actual Trays( 18 .12 )

For glycol contactors using bubblecap trays Eoverau typically ranges from 25-30o/o. (This is

equivalent to a Murphree plate efficiency of approximately 45-50%o). For most engineering calcula-tions an overall tray effrciency of 25o.h will yield satisfactory results.

CHAPTER 18

rH

353

\

GLYCOL DEI{YDRATION

A minimum spacing of 610 mm [24 in] is recommended. It is essential that a stable foam not

fill the gas space between trays to prevent excessive glycol loss. This spacing also allows a suitable

liquid level in the downcomers.

Tray hydraulics design is critical because of the low circulation rate. Liquid can bypass caps

or valves in some areas of the tray, ineffective gas-liquid contact can occur with low gas rates, and tray

liquid levels can be unstable. In situations like this, absorber perforfuance can vary markedly with gas

and liquid rate. A higher than calculated liquid rate may be necessary to provide the tray efficiency

required. A minimum weir length of 50% of the tower diameter is required to ensure adequate liquid

distribution across the tray.

For packed towers, equilibrium stages are converted to packing heights using an HTU (Height

of a Transfer Unit) or HETP (Height Equivalent to a Theoretical Plate (Stage)).

HETP and HTU are related concepts and depend primarily orr the gas and liquid properties, gas

and liquid rates as well as the surface characteristics and densiti' of the packing.

The mass transfer in a packed tower is continuous, and does not take place in discrete steps as

in trayed towers. It is for this reason that many packing manufacturers prefer to work in HTUs. The

HTU is multiplied by the Number of Transfer Units (NTUs) to arrive at the packing height. For

overall mass transfer controlled by resistarrce on the gas side, the nunber of transfer units may be cal-

culatcd from t l lc fol lowine equat ion.

I(here

Y r" F O vNTU= | '

. i v -Y*J T

NTU, : number of transfer units

yu - : concentration of water in the bottom of the contactor

yt : concentration of water at the top of the contactor

y* = equil ibrium water concentration

( r8 .13)

A similar concept is employed in heattransfer calculations. If we assume the operatingand equilibrium lines to be straight, as shown inFigure 18.10, then

NrU= Yu -Y, r"f 4l l l (18.14)

Ayr , -Ay , I Ay , /

VI/here: Ayt : Yu-Yu*

Ay, : Y , -Y t *

A typical HTU for structured packed tow-

ers depends on gas density and packing ple butfor packing with a specific area of 250 m'lm' 176ft2lft3l an HTU of 0.8 m[2.6 ft] gives good results

for preliminary calculations.

Despite the iechnical superiority of

HTU/NTU approach many companies still use

HETP/NTS method.

thetheFigure 18.10 Operat ing and Equi l ibr ium Lines for a

Typical Glycol Contactor

-i*+'



NTS stands for the Number of Theoretical Stages anci is identicai to the "N" values in Equa-tiens i8.l and 13.2. E','en thor'3h mas: transf:r in packeC to'r', 'ers is ccntinucus the I{ETPAITS methodis more commoil.

The relationship between NTS and NTU can be estimated frorn Equation 18.15.

I4/here

NrS _ HTU _ ( l _ l iA )NTU HETP In A

A : absorption factor (LnlVs*1K)

( r8 . ls )

specific area of 250 m2M3 U6The HETP of structured packing

\\

For TEG contactors containing structured packing with aft2lft3l, typical HETP values range from about I .6-2.0 m [5.3-6.5 ft].varies with contactor operating parameters as follows:

increasing Pressure

Increasing Gas Flowrate Increases HETP

, Increasing Liquid Flowrate Decreases HE,TP

Decreases in HETP mean better mass transfer

lncreases HETP

i' Increasing Specific Area of Packing I Decreases HETP

Example 18.5: Calculate the height of packing required to provide 2 equilibrium stages in a TEGcontactor.

Assume HETP : 1.75 m 15.74 ftl

Pack ing he igh t : (2 ) ( l 75) :3 .5 m

Pack ing he igh t : (2 ) (5 .74) : 11 .5 f t

It is customary to acid about 10% additional packing (usually 1-2 layers) to account for operatingcontingencies and allow for distribution of the gas and glycol at the top and bottom of the packings.

Liquid Di$ributor Design

In packed contactors, the design of the liquid distributor is critical. The liquid distributor en-sures that the incoming lean glycol is uniformly distributed across the packing. Several different dis-tributor designs are used but in general all consist of a main header box and a series of rectangularlaterai flow channels. as shown below.

l. Distributor must be level

2. Weir openings or drip tubes must be resistant to plugging

3. Drip point density should be 80-100/m2 7l-Otft214. Minimum glycol circulation rate should be about I (rn3A;/m2l I0.4 US gpm/ft2l

5. Area available for gas flow should be a minimum of 40-50% of cross sectional area

6. Drrp points should be less than l0 mmlll2 inl from the top of the packing to avoid splash-ing and droplet re-entrainment

CHAPTER 18

-

355

GLYCOL DEHYDRATION

Lateral Distributor

Main Header Box

\The distance between the

of 0.6 m t2 ftl. An examPle of a

Demister l'Iat

ManholeA t r r n n l l n l e t

Liquid Distributor

Structured Packing

Riser CaP

Gas Riser

ChimneY TraY(used if integral scrubber

installed in base of contactor)

top of the distributor and the mtst

structured packed tower is shown lneliminator should be a mtntmum

Figure 18 .11 .

To.rs D (minimum 0.15 m [6 in]-f-o.ts m [6 in]' typical

II o.oo m 1z tt1

0.2-0.4 m [8-16 in]depends on distributor design

Packed height2.5-4.0 m [8-13 ft] ryPical

0.40 m [16 inl minirnum

hc (025 m [10 in] tyPical)

0.5 D or 1 m [3 ft] minimumGlycol Outlet

Gas Inlet

Manhole

Vortex Breaker

Note: Column inside diameter = D Scrubber Liquid Outlet

L

l-rgure 18.1 1 Example Strucli jred Packed Absorber


TEG REGENERATION

Sotne vetrriors offer an absorption svstem oiher ihan a conventional vertical unit. One designuses rwo or three cocurrent contaciors in "e.ies scparated by scrubbers. The cocurent contractors arestatic mixers. This dcsign may have a lower installed weight when compared to a conventionalcontactor, and may offer a cost advantage r,,hen a high dewpoint depression is not required.

Rich TEGto Regenerator

Static mixers have also been used upstream of conventional contactors to add some nortion ofa theoretical stage to the unit.

TEG REGENERATIONThe required lean TEG concentration is produced in the rcgenerator. The regenerator consists

of a reboiler, stil l column and in some cases a gas stripping column. The lean TEG concentration iscontrolled by adjustment of reboiler temperature, pressure and the possible use of a stripping gas. Solong as no stripping gas is used, the concentration of the lean TEG leaving the reboiler is independentof the r ich TEG entcr ins.

The concenfration of rich TEG leaving the absorber is found by a $'ater material bala'rcearound that absorber. By oefinition

wt% Rich TEG =mass lean TEG

mass lean TEG + mass water absorbed + mass water in lean TEG(100)

The mass quantities in this equation may be found per unit of time or per unit of gas qr glycoiflow. In any case, the values used depenci on circulation rate. As we have seen in previous sc-ctions.TEG rate depends on dewpoint requirements, lean TEG concentration, number of absorber contacts andeconomics.

The rich TEG concentration may be calculated from Equation 18.16.

Rich TEG =(p)(lean TEG)

( 1 8 1 6 )p + (ricR)

FPS

Where:

I

IRich TEGRegenerator

pCR

rich TEG

lean TEG

lean TEG density

circulation ratio

wt% TEG in rich TEG solution

rvt9/o TEG in lean TEG solution

l.l2 kgil iter

liters TEG/kg H2O

9.3 lbtuS gal

US galTEGnb H2O

Dry Gas

CHAPTER 18

-

357

GLYCOL DEHYDRANCN

Regenerotion Poromelers

Because regeneration takes place near atmospheric pressure, under esserrtially ideal gas condi-tions, the calculation is routine. Figure 18.12 has been calculated to predict regenerator perfor-mance.(18'8)

The minimum wto% lean TEG on the top abscissa is found from Figure 18.4. The wt% of richTEG on the bottom abscissa is found from Equation 18.16. The diagonal lines in the lower left portionof Figure 18.12 represent various amounts of stripping gas for units where the stripping is sparged intothe reboiler.

Three temperature lines are shown. Where high concentrations are desired, the specification of204'C [400'F] is normal uniess the gas being dehydrated contains oxygen. This is close to the thermaldecomposition temperature (in air). In the usual case where the natural gas oxygen free, the use of204"C [400"F] has proven satisfactory.

The diagonal lines at upper right in Figure 18.12 represent the effect ofregeneration pressurein mmHg.

Unless a vacuum is being used, it is customary to use the 760 mmHg line for design calcula-tio;is.

Notice that at 760 mmHg pressure and a reboiler temperature of 204'C, Figure 18.12 shows alean TEG concentration of 98.7 wto/o. lf in using Figure 18.4 you obtain a concentration less than this,use 98.7 wto/o as the desired concentration when utilizing Figure 18.12.

The general procedure for using Figure 18.12 is as follows:

l. Atmospheric Pressure, No Stripping Gas.

wt% rich glycol is not a variable. Proceed vertically from 0stripping gas and temperature line intersection. You willread 98.7 wt% TEG at204oC:98.4 wto/o at 193"C.

2. Atmospheric pressure, Stripping Gas.

a. Proceed vertically from B to temperature line andthen horizontally.

b. Proceed vertically from A.

c. Intersection of two lines from A and B fixes amountof stripping gas.

3. Vacuum, No Stripping Gas.

a. Proceed vertically from intersection ofand temperature line to atmosphericmrnHg).

b. Proceed horizontally from point in (a)line necessary to fix value of point B.

0 gas linel ine (760

to pressure


a(!

c'a'-

Ec(U

o

F$q)

c)

:q)

ccq)

faac)o-o=

.E

oCl

LL

(U(n(d

c)c(U

L

c)o_

(uc)o(Dtt

c(d=

F6tu

o-c(U

z

(\l

diF

oJct,ll.

n : -c i 3

q)

U)

.C

ct)^ . c:O d

0)

()

-c

iic

N Ou ) U J

F

TEG REGENERATiON

CHAPTER 18 359

GLYCOL DEHYDRATIOII

In that rare case where both stripping gas and vacuum are useC, procedures (2) and (3) arecombined.

As a general rule, vacuum is avoided unless necessary to simplifu unit operation. Vacuumpumps can be a nuisance and air (oxygen) ingress accelerates glycol degradation. An ejector can beused to produce necessary vacuum in the righL circumstances.

Figure 18.12 is based on I equilibrium stage in the regenerator. Most regenerators will containmore than I equilibrium stage, particularly if a stripping column is installed between the reboiler andsurge tank. Figure 18.13 shows the lean TEG concentration vs stripping gas rate for units emnloying astripping column. In Figure 18.13, "N" refers to the number of theoretical stages in the stripping col-umn. The approximate HETP in the stripping column is 0.8 m 12.7 ftl for 5/8 in pali rings and 0.4 m

tl.3 ftl for structured packing.(18e)

Stripping gas rates seldom exceed 75 m3 (std)/m3 TEG [10 scf/gal] unless lean TrG concentia-tions in excuss of 99.99 u'toh are required. If these concentrations are required, an alternate designsuch as DRIZO@ or an adsorotion svstem should also be considered.

ReboilerHeat input to the regenerator is provided in the reboiler. The heat source is usually direct fired

with the fire-tubes immersed in a glycol bath. Other heat sources include hot oil (or other heat transferfluid) steam, or electric resistance heating.

In any event, the temperature of the glycol in the reboiler should not exceed 204"C [400"F]due to the degradation of TEG at higher temperatures. However, in order to maintain the glycol bathat a temperature at or slightly below 204'C [400'F] it is necessary to maintain the heat transfer surfaceat a temperature above this value. This can lead to some degradation of the giycol in contact with theheat transfer surface. For this reason the following maximum flux rates should not be exceeded.These flux rates, in turn, will set the heat transfer area.

Example 18.4: An example is shown in Figure 18.12 for use of stripping gas and vacuum. A96.85 wt% rich glycol enters a regenerator using 0.03 m3 of stripping gas per literof glycol solution [4 scf{JS gal]. Proceeding to 204"C and then vertically, onereads 99.14 wto/o if atmospheric pressure is used. If a vacuum is employed andthe absolute pressure is 500 mmHg, the lean glycol concentration is 99.41 wt%"

Direct Fired 19 kW/m2 [6000 Btu/hr-ft2]

24 kw lm2 [7600 Btu,/hr-ft2]

, 24 kW lmz [7600 Bru/hr-ft2j

i tz.s twlm2 1+ooo Bru/hr-ft2l

Most glycol reboilers maintain the bath temperature near 204"C [400'F]. Lower temperaturesmay reduce degradation but result in lower lean TEG conuentrations which, in turn, rlscessitate highercirculation rates or higher stripping gas rates.

Pressure effects on regenerator operation are not often fully appreciated. Lean TEG concentra-tions are reduced by backp;essure on the rcgenerator. T;pical backpressure for veniing to a LP flaresystem or conCenser system may be as much as 3.5-7 kPa [0.5 to I psi]. For a regene,*.or operating at


TEG REGENERATION

100.0 9

Stripping Gas Rate, (stci) m3/m3 TEG30 45 60

O T A L" " " 0

204'C [400"F] with no stripping gas, the effect of pressure in lean TEG concentration is about 0.014wt o/olWa [0. ] wt %lpst].

The reboiler dury depends on the TEG circulation ratio (liters TEG/H2O [gal TEG/lbm HzO]),the efficiency of the rich-lean TEG exchanger, the reflux ratio, stripping gas rates and effectiveness ofthe insulation. Heat balances indicate a required reboiler duty of 250-300 kJiliter [900-1075 BtuiUSgall. The reboiler duty should actually be sized to deliver 130-140% of the expected duty to providefor start-up, insulation losses, etc. A design value of 400 kJiliter [1430 Bfu/US gal] will typically pro-vide sufficient heat input flexibility to meet any expected operating condition. The reboiler dutyshould always be sized based on circulation pump capacity, not the expected circulation rates.

o\

= vv.ui

ooo 98.8

ulFc

3 eB.6J

Figure 18.13 Effect of Stripping Gas on Lean TEG Concentrations for Regenerators with Stripping Columns

CHAPTER 18 361

GLYCOL DEHYDRATION

Still [olumnThe still column is the "fractionator" portion of the regenerator. The column mav be packed or

trayed. Packed columns are more common and the packing is typically a random packing such asstainless steel slotted rings. Packing sizes range from 16 mm l5/8 inl to 5l mm [2 in] depending onthe still column size with the larger packing used in the larger diameter columns. Structured packinghas also been used, and in very large units (diameters > I m [3 ft]). trays have been installed.

The still column is sized based on standard packed tower sizing correlations. Since the vaporloading is often tied to the glycol circulation rate many correlations have been developed which esti-mate the still column diameter as a function of TEG circulation rate. One such correlation is shownbelow and is based on 25 mm I in] slotted ring packing.

d = (AXqrec )o 5

d : diameter of packed torver

grrc - g lycc l c i rcu lat ion rate

A : empirical constant based on (1 in. pall r ings)

( r8 .17)

SI FPS

I(here: mm

m ' t n

210

ln

US gal/min

4 .0

The reflux ratic employed in TEG systems is very small. L/D values typically range from 0.1to 0.2 mols/mol. This is equivalent to condensing l0-209l, of the total overhead vapor stream. The re-flux rate should be the minimum reqrrired to maintain the still overhead temperature at the boilingpoint of water for the partial pressure of water at the top of the regenerator. When stripping gas isused, the water partial ptessure will be less than I atm. In fact, this is the princrple which results inlower water concentrations in the lean TEG. Consequently, when stripping gas is used, the partialpressure of H2O will be less than 1 atm, hence the boiling point of water will be lower as well.

Figure 18.14 shows the recommended still column overhead temperature as a function of theTEG circulation ratio and stripping gas rate for a still column operating at I atm.

Reflux is normaliy supplied by using the rich glycol stream circulating through a condensingcoil inserted in the top of the still column. Cooling water or a fin-fan cooler could also be used. How-ever, using the rich glycol stream is the preferred method for cost, simplicity and energy efficiency.Heat fansfer coeffrcients in the reflu' coil typically range from ll0-230 Wm2.K [20-40 Btu/hr-ft2-'F].

Glycol-Glycol (Leon-Rich) l|eol Exchonger

This is a basic heat exchanger. Its efficiency has adirect etl-ect on reboiler heat load. The rich glycol fromthe absorber (Pt.l) enters at a temperature 5-l3oC [9-23'F]wanner than the inlet gas due to the reflux corrdenser duty.The lean glycol from the regenerator (Pt. 3) enters usuallyat about 204"C [400'F]. The exchanger is designed so thetemperature of the lean glycol at Pt. 4 should not begreater than 60-65'C [140-149'F].

In most cases a l5-20"C 127-36"F1approach in theheat exchanger is desirable. If it is too high, the reboilerand glycol cooler duties will increase.

VOLUME 2: THE EQUIPMENT MODULES362

TEG REGINERATION

Circulation Ratio, galTEG/lbm H.O3 4 5

( ) 8 5

q,

:f( d A n

c)F

TL

1BO E:l(s

trr / u o

F

20 30 40

Circulation Ratio, l i ters TEG/kg HrO

Figure 18.14 Recommended Stil l Column Top Temperatures

In the lean-rich exchanser

Q l - + = Q r - z = m r ( h z - h r ) = m 3 ( h 3 - h a ) ( 18 .18 )

The easiest way to perform this calculation is to look up the average heat capacity of the glycolin Appendix 18A and multiply it by AT across the heat exchanger to find Ah for the lean mixture. Re-member Ah - CeAT.

Double-pipe type, with bare or finned fubes, or plate-type heat exchangers are frequently used.Plate exchangers are preferred, especially for offshore or large units, because they are more compact,lighter and cheaper. They are, however, susceptible to fouling and plugging and it is imperativi theglycol is clean and filtered.

Some small dehydration units use a coii in the surge drum to exchange heat between the richand lean glycol stream. Not only are these coils limited in surface area but the overall 'U' values tendto be low. As a result they may not be adequate for applications where the available reboiler heat islirnited. They also may not cool thc' lean glycol adequately to meet the pump temperature limit. Theyshould only be considered for small packaged units having a reboiler duty less than 150 kW t50bMBtu/hrl.

CHAPTER 18 363

GLYCOL DEHYDRATION

Leon Glycol Cooler

A final glycol cooler is required so that the lean glycol entering the top of the contactor is

cooled to within 5-10"C [9-18"F] of the gas temperature entering the top tray. The lean glycol is often

cooled with a lean glycol-gas heat exchanger. Gas-glycol exchangers are inexpensive and compact but

the lean TEG temperature can increase to unacceptable levels at.low gas rates. Although air or

water-cooled glycol coolers risk over-cooling the lean glycol, they are often used since control of the

glycol temperature is then independent of the gas flow.

If a gas-glycol exchanger is used we do not recommend the type which employs an integral

coil in the top of the absorber column due to less effrcient heat transfer and problems of inspection and

maintenance.

FILTERS

Good filtration is critical. The full-flow type is preferred. I recommend two filters in parallel,

with no by-pass lines, so that ful1 filtration is assured.

A cloth fabric element that is capable of reducing solids to about 100 ppm by weight is pre-

ferred. Paper and fiberglass elements generally have proven unsatisfactory. Filter size in a properly

operated glycol system should be 5-10 pm. Larger sizes (25-50 pm) may be required during staffup

and in dirry service.

li may be impossible to judge the cifectiveness of filtration by color alone. Even well filtered

glycol will often be black. But, removal of most of the solids will reduce corrosion, plugging and de-posits in the reboiler, and may reduce foaming losses. Good filtration is critical for satisfactory perfor-

mance. It is desrrable to meas.rre the pressure differential across the filter and change the elements

when it reaches the filter supplier's recommended maximum AP which is often t70 kPa [25 psi].

The use of a carbon purifier downstream from the filter often is recommended. This can pro-

duce essentially water-white glycol. Maintenance of this color has proven desirable because it tends to

increase dehydration efficiency and minimize foaming, a major source of glycol loss.

Aromatic hydrocarbons are often present in ilie rich glycol entering the regcnerator and will be

adsorbed on the carbon filter. These components will quickly reach equilibrium loading on the carbonfilter although it is likely they are eventually displaced (to some extent) by heavier hydrocarbons. Be-

cause of this high aromatic content, changeout of carbon filters requires speciai precautions to avoid

unnecessary exposure of workers to BTEX components.

ln some units, the carbon filter is installed in the lean TEG stream, upstream of the contaeror,to avoid significant aromatic loading.

Coal-based activated carbon should be used because wood-based charcoal tends to break up in

use. This carbon can be placed in a metal canister or installed as fil l into a vessel. In either case, good

screens are needed to prevent carbon loss into the system. Carbon particles, much like iron sulfide,

tend to promote a stable fcam.

Glycol filters are only effective when used. In especially dirty glycol systems, filters are oftenbypassed to avoid irequent filter change-out. The problems wjth this should be obvious. If filter plug-ging is excessive, try larger filter size and look for source of problem (e.g., poor inlet separation, deg-radation, corrosion products, etc.).


GLYCOL CIRCULATION PUMPS

SURGE DRUMThe surge drun should be sized to provide the following:' a retention time not less than 20 minutes between low and normal levels. based on desisn

circulation rate:. hold-up capacity betwcen normal level and high level;. a reasouable length of time between glycol additions;' sufficient volume to accept the glycol drained from the reboiler to allow repair or inspec-

ticn of the firetube or heating coil.

It is often vented to the rebciler but a small amount of N2 or dry fuel purge gas is sometimesneedeci to prevent water vapor from the reboiler flowing through the vent line and being absorbed bythe glycol in the surge drum. If it is not located directly below the reboiler, it should be provided witha separate purge gas supply.

Provisions should alsc be made to facilitate:' make-up of the glycol inventory from glycol storage by means of a simple hand or

air-driven pump with a check valve and filter;' the batch addition of chemicals to the glycol surge vessel, e.g. for pH control, corrosion in-

hibition, etc.

GLYCOL FIASH VESSELThe glycol flash vessel is used to remove light hydrocarbons, CO2 and/or H2S, that have been

absorbed or entrained with the glycol as well as recover the spent gas from gas-glycol powered pumps.It also seryes to separate any liquid hydrocarbons from the glycol to prevent them from entering thereboiler and causing fouling and foaming.

Both the sulfur compounds and carbon dioxide are very soluble in water and react to sorne de-gree with the glycols. Degassing in the fiash ves:el upstrean; of the regenerator reduces their concen-tration somewhat and helps mitigate corrosion in the regenerator. Degassing is more efficient if theglycol is preheated first. Preheating, to about 60-70"C [140-l58oF] is often done to decrease viscosityand facilitate degassing.

When gas-glycol powered pumps are used, the operatingpressure of the flash vessel must be[5 percent of the contactor'operating

flash vessel must operate at a pressure

The flash vessel should not contain liquid hydrocarbons. Unforfunately this is often not thecase due to poor inlet gas separation upstream of the contactor. Therefore, it is prudent to install a hy-drocarbon skim nozzle, bucket or trough and weir to collect and separate the liquid condensate fromthe r ich glycol.

GLYCOL CIRCULATION PUMPSGlycol circulation pumps may be electric driven, gas, or gas-glycol powered. The pumps

should be sized to provide a minimum of 25 percent excess capacity, and in critical seryice two glycolpumps shall be provided. each designed for 100 percent duty.

compatibie with the maximum pump exhaust pressure which ispressure. Thus, with 70 bar [1015 psia] contactor pressure theIower than l0 bar I i45 psia].

CHAPTER 18 365

GLYCOL DEHYDRATION

pumps utilizing conventional electric motor drives are normally reciproca{ing multiplex type'

A conservative, slow ii.ton speed (0.6 m/s [120 ftlmin]) pump should be used since the lubricating

properties of glycol url poor. Due to the expected turn-down of glycol systems, variable speed drives

are often used on Iarger units. They can pro'nid" the flexibilify to increase the glycol circulation rate if

needed to meet dewpoint requirements and reduce the glycol rate to operate more economically' Low

speed centrifugal booster pumps are sometimes used when NPSHA to the reciprocating pumps is mar-

ginal.

The gas-glycol powered pump (Kimray pump) utilizes the rich glycol under pressure plus sup-

plemental gu. f.; the contactor to furnish the driving energy. Since the pumping rate is proportional

to the volume of the retum glycol and gas, the pumping rate is controlled by adjusting this flow'

A cutar.vay of a Kimray pump is shown in Figure l8'15'

INLET SEPARATION

Without a doubt, most operating problems with glycol dehydration systems are a direct result

of inadequate inlet gas treatment upstream of the contactor. The upstream separator should remove liq-

uid hydrocarbons, liquid water, solids, corrosion inibitors, etc. prior to the glycol unit. John Campbell

Sr. says ,.ycu cannot afford the dehydration if you cannot afford to place an effecttve separator on the

gas inlet.,' a glycol unit is a closed system. Non-volatile contaminants remain in the glycol and must

be removed by filtration or blowdown.

Somc of the more deleterious contaminants are salt and high boiling point hydrocarbons. These

remain in the glycol. Salt precipitates in the reboiler, still column and rich-lean exch:lgers. This can

cause plugging rncreasing pressure drop and decreasing flowrates. In the reboiler, salt can coat the

firetube causing hot spots and evenfuai firetube failure'

Heavy hydrocarbons can "coke" in the reboiier causing hot spots tn the tiretuoe and plugging

in the still and stripping columns. In addition, these burnt hydrocarbons cause the glycol to become

black and promote loaming in the contactor, and cause frequent filter change outs. lf sulfur com-

pounds ars present in the gas, these can combine with the heavy hydrocarbons to form a corrosive

"sludge" which accumulates in the system.

Certainly, a properly sized impingent separator is the first step in preventing unwanted carry-

over. Many units also utilize coalescing filter separators and/or superheaters downstream of the pri-

mary separator. Frequently, a heat exchanger designed to increase the gas temperature by 5"C [9"F] is

lnstaned immediately upstream of the contactor. This serves to vaporize any entrained liquid particles

in the gas.

Many designs utilize an integral scrubber installed in the base of the contactor. While this may

provide some removal of entrained particles, these "scrubbers" are typically under-sized, particularly

when the contactor is packed with structured packing. These integral scrubbers should never be used

as a primary separator - only for secondary scrubbing'


Documents

Campbell 2 .hal 333-366 Bab 18