GPSA ENGINEERING DATABOOK ERRATA

(2004 SI Edition)
PAGE DESCRIPTION
18 Fig. 12 Change units for Energy conversion factor
715 Fig. 723 Delete commas
98 Text changes
910 Fig. 913 Corrections
1017 Fig. 104 Change equation
1018 Change example following Eq. 104
1320 Ex. 135 Change units on head
1322 Eq. 1328 Change equation
1326 Fig. 1333 Change units on head
1327 Fig. 1334 Change units on head
1328 Eq. 1335 Change equation
2019 Text changes
2031 Figs. 2063, 2064 Reference change
2034 Eq. 2010 Change equation
2042 Fig. 2084 Change axis label
215 Fig. 213 Change heats of reaction
wellhead
The assembl y of fi tti ngs, val ves, and control s l ocated at the
surface and connected to the fl ow l i nes, tubi ng, and casi ng of
the wel l so as to control the fl ow from the reservoi r.
wet gas
(1) A gas contai ni ng water, or a gas whi ch has not been dehy-
drated. (2) A term synonymous wi th ri ch gas. Refer to defi ni -
ti on of "ri ch gas".
Wobbe number
A number proporti onal to the heat i nput to a burner at con-
stant pressure. I n Bri ti sh practi ce, i t i s the gross heati ng val ue
of a gas di vi ded by the square root of i ts gravi ty. Wi del y used
i n Europe, together wi th a measured or cal cul ated fl ame
speed, to determi ne i nterchangeabi l i ty of fuel gases.
Conversion Factors
I n these tabl es, factors for conversi on, i ncl udi ng conversi ons
to the I nternati onal System of Uni ts (SI ), are based on ASTM
Standard for Metri c Practi ce, E380-91. The l atest edi ti on of
thi s publ i cati on shoul d be studi ed for more detai l on the SI
system, i ncl udi ng defi ni ti ons and symbol s.
I n cal cul ati ng deri ved factors i n the tabl es that fol l ow, exact
conversi ons were used, when avai l abl e, rather than the 7-di gi t
round-offs l i sted i n ASTM E380 conversi on tabl es. Deri ved fac-
tors gi ven bel ow are rounded to the same number of si gni fi cant
di gi ts as the source factors.
I n any conversi on of fundamental measurement uni ts, some
confusi on may resul t due to redefi ni ti on of uni ts used i n earl i er
tabl es. For exampl e, i n 1959 a smal l refi nement was made i n
the defi ni ti on of the yard, whi ch changed i ts l ength from
3600/3937 meter (or 1 i nch = 25.4000508 mm) to 0.9144 m
exactl y (or 1 i nch = 25.4 mm exactl y). The tabl es bel ow are
based on the new defi ni ti on, but one shoul d be aware that
where U.S. l and measurements are concerned, the ol d rel ati on-
shi p appl i es. Refer to ASTM E380-91, note 13, for more detai l .
Energy Units Conversion
Confusi on may ari se i n the defi ni ti on of uni ts for heat or
energy. I n the tabl es bel ow, the Btu (I T) and cal ori e (I T) are
used. These are the heat uni ts recommended by the I nterna-
ti onal Conference on the Properti es of Steam, as defi ned:
1 Btu (I T) = 1055.055 852 62 joul e (exactl y)
1 Cal ori e (I T) = 4.186 800 joul e (exactl y)
For information only, other defi ni ti ons that may be used
el sewhere:
1 Btu (Mean) = 1055.87 joul e
1 Btu (39F) = 1059.67 joul e
1 Btu (60F) = 1054.68 joul e
1 Btu (Thermochemi cal ) = 1054.350 joul e
1 cal ori e (Mean) = 4.190 02 joul e
1 cal ori e (15C) = 4.185 80 joul e
1 cal ori e (20C) = 4.181 90 joul e
1 cal ori e (Thermochemi cal ) = 4.184 000 joul e
The fundamental rel ati onshi p between the Btu and the cal o-
ri e:
grampound rel ati onshi p
Fahrenhei tCel si us scal e rel ati onshi p
or: Btu
453.592 29
1.8
= cal ori e (I T, mean, or other)
1 therm = 100,000 Btu = 105.5056 x 10
6
J
= 105,505.6 kJ = 1.055 056 x 10
5
kJ.
(Btu denotes Bri ti sh Thermal Uni ts)
(ref: Physi cal Properti es of Natural Gases,
Gas Uni e, 1988 p. 23)
Velocity
(Length/uni t of ti me)
ft/sec ft/mi n Mi l es/hr (U.S. Statute) m/sec m/mi n km/hr
1 60 0.6818182 0.3048 18.288 1.09728
0.01666667 1 0.01136364 5.08 x 10
3
0.3048 0.018288
1.466667 88 1 0.44704 26.8224 1.609344
3.280840 196.8504 2.236936 1 60 3.6
0.05468066 3.280840 0.03728227 0.016667 1 0.06
0.9113444 54.68066 0.6213712 0.2777778 16.66667 1
Energy
Ft-l bf Kgf-meter Btu (I T) Ki l o-cal ori e (I T) Hp-hr Ki l owatt-hr joul e (J)
1 0.1382550 1.285068 x 10
3
3.238316 x 10
4
5.050505 x 10
7
3.766161 x 10
7
1.355818
7.233014 1 9.294911 x 10
3
2.342278 x 10
3
3.653037 x 10
6
2.724070 x 10
6
9.806650
778.1692 107.5858 1 0.2519958 3.930148 x 10
4
2.930711 x 10
4
1055.056
3088.025 426.9348 3.968321 1 1.559609 x 10
3
1.163 x 10
3
4186.8
1980000 273744.8 2544.434 641.1865 1 0.7456999 2684520.
2655224 367097.8 3412.142 859.8452 1.341022 1 3600000.
0.7375621 0.1019716 9.478171 x 10
4
2.388459 x 10
4
3.725061 x 10
7
2.777778 x 10
7
1
FIG. 1-2
Conversion Factor Tables
1-8
where the two i mmi sci bl e l i qui d phases separate wi thi n the
vessel by the di fferences i n densi ty of the l i qui ds. Suffi ci ent
retenti on ti me must be provi ded i n the separator to al l ow for
the gravi ty separati on to take pl ace. The second category i s
defi ned as coal esci ng separati on. Thi s i s where smal l parti -
cl es of one l i qui d phase must be separated or removed from a
l arge quanti ty of another l i qui d phase. Di fferent types of i n-
ternal constructi on of separators much be provi ded for each
type of l i qui d-l i qui d separators. The fol l owi ng pri nci pl es of de-
si gn for l i qui d-l i qui d separati on appl y equal l y for hori zontal
or verti cal separators. Hori zontal vessel s have some advan-
tage over verti cal s for l i qui d-l i qui d separati on, due to the
l arger i nterface area avai l abl e i n the hori zontal styl e, and the
shorter di stance parti cl es must travel to coal esce.
There are two factors that may prevent two l i qui d phases
from separati ng due to di fferences i n speci fi c gravi ty:
I f dropl et parti cl es are so smal l that they may be sus-
pended by Browni an movement. Thi s i s defi ned as a ran-
dom moti on that i s greater than di rected movement due
to gravi ty for parti cl es l ess than 0.1 m i n di ameter.
The dropl ets may carry el ectri c charges due to di ssol ved
i ons. These charges can cause the dropl ets to repel each
other rather than coal esce i nto l arger parti cl es and settl e
by gravi ty.
Effects due to Browni an movement are usual l y smal l and
proper chemi cal treatment wi l l usual l y neutral i ze any el ectri c
charges. Then settl i ng becomes a functi on of gravi ty and vi s-
cosi ty i n accordance wi th Stokes Law. The settl i ng vel oci ty of
spheres through a fl ui d i s di rectl y proporti onal to the di ffer-
ence i n densi ti es of the sphere and the fl ui d, and i nversel y
proporti onal to the vi scosi ty of the fl ui d and the square of the
di ameter of the sphere (dropl et), as noted i n Eq 7-3. The l i q-
ui d-l i qui d separati on capaci ty of separators may be deter-
mi ned from Equati ons 7-13 and 7-14, whi ch were deri ved from
Equati on 7-3
9
. Val ues of C* are found i n Fi g. 7-23.
Verti cal vessel s:
W
cl
= C

S
hl
S
l l

(0.785) D
v
2
Eq 7-13
Hori zontal vessel :
W
cl
= C

S
hl
S
l l

L
l
H
l
Eq 7-14
Si nce the dropl et si ze of one l i qui d phase di spersed i n an-
other i s usual l y unknown, i t i s si mpl er to si ze l i qui d-l i qui d
separati on based on retenti on ti me of the l i qui d wi thi n the
separator vessel . For gravi ty separati on of two l i qui d phases,
a l arge retenti on or qui et settl i ng secti on i s requi red i n the
vessel . Good separati on requi res suffi ci ent ti me to obtai n an
equi l i bri um condi ti on between the two l i qui d phases at the
temperature and pressure of separati on. The l i qui d capaci ty
of a separator or the settl i ng vol ume requi red can be deter-
mi ned from Eq 7-12 usi ng the retenti on ti me gi ven i n Fi g. 7-22.
The fol l owi ng exampl e shows how to si ze a l i qui d-l i qui d
separator.
Example 7-3 Determi ne the si ze of a verti cal separator to
handl e 100 m
3
/day of 0.76 rel ati ve densi ty condensate and 10
m
3
/day of produced water. Assume the water parti cl e si ze i s
200 m. Other operati ng condi ti ons are as fol l ows:
Operati ng temperature = 25C
Operati ng pressure = 6900 kPa(ga)
Water rel ati ve densi ty = 1.01
Condensate vi scosi ty = 0.55 mPa s @ 25C
Condensate rel ati ve densi ty = 0.76
From Eq 7-14
W
c l
= C

S
h l
S
l l

(0.785) (D
v
)
2
From Fi g. 7-17 for free l i qui ds wi th water parti cl e di ameter =
200 mi crons, C* = 1880.
100m
3
/day = 1880
1.01 0.76
(0.55)
(0.785) (D
v
)
2
(10
6
)
(D
v
)
2
= (1.43) (10
5
)
D
v
= 390 mm
Usi ng the al ternate method of desi gn based on retenti on
ti me as shown i n Eq 7-16 woul d gi ve:
U =
W (t)
1440
From Fi g. 7-18, use 3 mi nutes retenti on ti me.
U =
(110) (3)
1440
= 0.23 m
3
A 390 mm di ameter vessel wi l l hol d about 0.12 m
3
per 1000
mm of hei ght. The smal l vol ume hel d i n the bottom head can
be di scounted i n thi s si ze vessel . The shel l hei ght requi red for
the retenti on vol ume requi red woul d be:
Shel l hei ght =
0.23
0.12
= 1.9 m = 1900 mm
Another parameter that shoul d be checked when separati ng
ami ne or gl ycol from l i qui d hydrocarbons i s the i nterface area
between the two l i qui d l ayers. Thi s area shoul d be si zed so the
gl ycol or ami ne fl ow across the i nterface does not exceed ap-
proxi matel y 100 m
3
per day per m
2
.
The above exampl e i ndi cates that a rel ati vel y smal l separa-
tor woul d be requi red for l i qui d-l i qui d separati on. I t shoul d be
remembered that the separator must al so be desi gned for the
vapor capaci ty to be handl ed. I n most cases of hi gh vapor-l i q-
ui d l oadi ngs that are encountered i n gas processi ng equi pment
desi gn, the vapor capaci ty requi red wi l l di ctate a much l arger
vessel than woul d be requi red for the l i qui d l oad onl y. The
properl y desi gned vessel has to be abl e to handl e both the va-
por and l i qui d l oads. Therefore, one or the other wi l l control
the si ze of the vessel used.
Emulsion
Characteristic
Droplet
Diameter, m
Constant,
10
C*
Free Li qui ds 200 1880
Loose Emul si on 150 1060
Moderate
Emul si on
100 470
Ti ght Emul si on 60 170
FIG. 7-23
Values of C* Used in Equations 7-13 and 7-14
7-15
Use base val ues from Fi g. 9-11 for (r
i
)
1
condi ti ons.
f

2
1
_

,
0.47

k1
k2
_

,
0.67

Cp1
Cp2
_

,
0.33

G1
G2
_

,
0.8

Di 2
Di 1
_

,
0.2

0.21
0.726
_

,
0.47

0.136
0.135
_

,
0.67

2.09
2.51
_

,
0.33

903
537
_

,
0.8

15.7
12.6
_

,
0.2
0.840
Basi s: (I nsi de Area)
(r
i
)
2
f(r
i
)
1
and (r
i
)
1
0.000 67 from Fi g. 9-11
(0.840) (0.000 67)
0.000 563 (m
2
C)/W
From Fi g. 9-10 (see Note ), the rati o of the second to the
fi rst pressure drop i s:
(P
i
)
2
= (f) (P
i
)
1
Use base val ues from Fi g. 9-11 for (P
i
)
1
condi ti ons.
f

1
_

,
0.2

G
2
G
1
_

,
1.8

2
_

D
i 1
D
i2
_

,
1.2

Np
2
Np
1
_

0.21
0.726
_

,
0.2

537
903
_

,
1.8

751
614
_

12.6
15.7
_

,
1.2

1
1
_

,
0.285
(P
i
)
2
= (f) (P
i
)
1
= (0.285) (1.37) = 0.390 kPa/m
For a 9.15 m tube l ength the total 9.15(0.390) = 3.569 kPa
5. Cal cul ate the shel l si de pressure drop and resi stance to
heat transfer wi th the rel ati onshi ps shown i n Fi g. 9-10,
the val ues shown i n Fi g. 9-11, and the data shown i n
Fi g. 9-13.
G
(215 784)
(3600)(101 283/1 000 000)
591.8 kg/(m
2
s)
From Fi g. 9-10 (see Note ), the rati o of the new to the
ol d resi stance i s:
(r
o
)
2
= (f) (r
o
)
1
Use base val ues from Fi g. 9-11 for (r
o
)
1
condi ti ons.
f

1
_

,
0.27

k
1
k
2
_

,
0.67

Cp
1
Cp
2
_

,
0.33

G
1
G
2
_

,
0.6

D
o2
D
o1
_

,
0.4

0.34
0.549
_

,
0.27

0.132
0.133
_

,
0.67

2.33
2.27
_

,
0.33

646.4
591.8
_

,
0.6

19.05
15.7
_

,
0.4
0.998
(r
o
)
2
(f)(r
o
)
1
(0.998)(0.000 49) 0.000 49 (m
2
C)/W
From Fi g. 9-10 (see Note ), the rati o of the new to the
ol d pressure drop i s:
(P
o
)
2
= (f) (P
o
)
1
Use base val ues from Fi g. 9-11 for (P
o
)
1
condi ti ons. Ob-
tai n tube rows crossed between baffl e wi ndow centroi ds
from Fi g. 9-13.
RC
2
= 23 (RC
1
= 10 per note on Fi g. 9-11)
Obtai n the number of crossfl ow spaces, whi ch i s one more
than the number of baffl es, from Fi g. 9-12.
SP
2
= 19 [ SP
1
= 1 per note on Fi g. 9-11 si nce (P
o
)
1
i s for one baffl e space.]
f

1
_

,
0.15

G
2
G
1
_

,
1.85

2
_

D
o1
D
o2
_

,
0.15

SP
2
SP
1
_

RC
2
RC
1
_

0.34
0.549
_

,
0.15

591.8
646.4
_

,
1.85

750
660
_

15.7
19.05
_

,
0.15

19
1
_

23
10
_

,
38.3
(P
o
)
2
= (f) (P
o
)
1
= (38.3) (1.7) = 65.1 kPa
6. Cal cul ate the tube metal resi stance.
D
o
19.05 mm
D
i
15.75 mm
k
w
50 W

(m C) from Fi g. 9-8.
r
w

D
o
2 1000 k
w
ln

D
o
D
i
1
1
]
0.000 036(m
2
C)/W
?. CaIcuIato tho ovoraII hoat transfor coofficiont.
r r
i

A
o

A
i

_

,
+ r
o
+ r
w
+ r
fo
+ r
fi

A
o

A
i

_

,
0.000 563

0.0182
0.0151
_

,
+ 0.000 49 + 0.000 036 + 0.000 35
+ 0.000 2

0.0182
0.0151
_

,
0.0018
U
1
r

1
0.001 78
555.6 W/(m
2
C)
8. Compare the requi red heat transfer coeffi ci ent cal cu-
l ated i n step 3 to the val ue cal cul ated i n step 7. (Avai l abl e
U = 561.8; requi red U = 570.4) The avai l abl e val ue i s
1.6% l ess than the requi red val ue, and the cal cul ated
pressure drops are l ess than the pressure drops al l owed
i n Fi g. 9-12. Therefore, by these cal cul ati ons, the uni t wi l l
perform adequatel y.
CONDENSERS
The purpose of a condenser i s to change a fl ui d stream from
the vapor state to the l i qui d state by removi ng the heat of
vapori zati on. The fl ui d stream may be a pure component or a
mi xture of components. Condensati on may occur on the shel l
si de or the tube si de of an exchanger ori ented verti cal l y or
hori zontal l y.
Condensi ng the overhead vapors of a di sti l l ati on col umn i s
an exampl e of condensi ng a mi xed vapor stream. A verti cal
exchanger fl anged di rectl y to the top of the col umn mi ght be
used. The condensed l i qui d drai ns back i nto the col umn coun-
tercurrent to the vapor enteri ng the condenser. The major con-
cerns i n desi gni ng thi s type exchanger are keepi ng the vapor
vel oci ty suffi ci entl y l ow to prevent fl oodi ng the exchanger and
eval uati ng an appropri ate temperature profi l e at the condens-
i ng surface to determi ne an effecti ve temperature di fference.
The techni cal l i terature addresses cri teri a for fl oodi ng deter-
mi nati on
1
and speci al fl ow characteri sti cs of fal l i ng l i qui d
fi l ms. A useful esti mate for determi ni ng an effecti ve tempera-
ture di fference can be made by assumi ng an i sothermal con-
densate fi l m at the saturati on temperature of the l ast
condensate formed. I f the condensi ng temperature range ex-
9-8
ceeds 5C, consul ti ng a speci al i st i s recommended for a more
ri gorous cal cul ati on procedure.
The condensi ng of a pure component occurs at a constant
temperature equal to the saturati on temperature of the i n-
comi ng vapor stream. Frequentl y a vapor enters a condenser
superheated and must have the sensi bl e heat removed from
the vapor before condensati on can occur. I f the condensi ng sur-
face temperature i s greater than the i ncomi ng vapor satura-
ti on temperature, the superheat i n the vapor i s transferred to
the col d surface by a sensi bl e heat transfer mechani sm (dry-
wal l condi ti on). I f the condensi ng surface temperature i s l ess
than the saturati on temperature of the i ncomi ng vapor, a con-
densate fi l m wi l l be formed on the col d surface. The sensi bl e
heat i s removed from the vapor at the condensate-vapor i nter-
face by vapori zi ng (fl ashi ng) condensate so that the heat of
vapori zati on i s equal to the sensi bl e heat removed from the
vapor. Under thi s wet wal l condi ti on, the effecti ve tempera-
ture of the vapor i s the saturati on temperature, and the effec-
ti ve heat transfer mechani sm i s condensati on. The
determi nati on of the poi nt i n the desuperheati ng zone of a
condenser where drywal l condi ti ons cease and wet wal l
condi ti ons begi n i s a tri al and error procedure. A method fre-
quentl y empl oyed to gi ve a safe approxi mati on of the requi red
surface i s to use the condensi ng coeffi ci ent and the CMTD
based on the vapor saturati on temperature to cal cul ate the
surface requi red for both the desuperheati ng zone and the con-
densi ng zone.
The fol l owi ng Exampl e 9-2 wi l l i l l ustrate the use of the heat
rel ease curve to cal cul ate the surface requi red and the LMTD
for each zone i n a condenser for a pure component appl i cati on.
Example 9-2 A propane refri gerant condenser i s requi red
to condense the vapor stream usi ng the heat rel ease curve as
shown i n Fi g. 9-14. Thi s stream enters the condenser super-
heated and l eaves the condenser as a subcool ed l i qui d. Assume
that a si ngl e-tube pass, si ngl e-shel l pass, counterfl ow ex-
changer i s used so that LMTD correcti on factors do not appl y.
Note that the propane i s on the shel l si de. The overal l heat
transfer coeffi ci ents for each zone are as fol l ows:
Desuperheati ng: (82C to 42C)
U
v
= 396.6 W/(m
2
C)
h
v
= 630.4 W/(m
2
C)
Condensi ng: (42C to 42C)
U
c
= 794.4 W/(m
2
C)
Subcool i ng: (42C to 35C )
U
L
= 649.7 W/(m
2
C)
Solution Steps
1. Cal cul ate the surface temperature (outsi de wal l ) on the
vapor si de at the refri gerant stream i nl et usi ng the fol -
l owi ng equati on:
2

Leakage use TEMA tol erances
tube hol e = 576 19.05 0.4 = 13 789 mm
2
shel l crack = 762 4.445 0.60 = 6 385 mm
2
20 174 mm
2
Wi ndow fl ow area:
wi ndow area = 0.237 (762)
2
= 137 613 mm
2
tubes i n wi ndow = 208 0.785 (19.05)
2
= 59 255 mm
2
78 358 mm
2
l eak 20 174 mm
2
98 532 mm
2
Cross fl ow area:
Free area = 171.45 mm 473.08 = 81 109 mm
2
Leakage = 20 174 mm
2
Net cross fl ow area = 101 283 mm
2
BP = baffl e pi tch = 473.08 mm
FD = free di ameter = 171.45 mm
Note: wi ndow area cross fl ow area
FIG. 9-14
Propane Condensing Curve
FIG. 9-13
Heat Exchanger Detail Design Results
9-10
Sound power l evel cannot be measured di rectl y but must be
cal cul ated from sound pressure l evel s (SPL) dB. I n metri c
terms, thi s i s known as Lw.
Sound pressure l evel , known as SPL, or Lp i n metri c termi -
nol ogy, i s the audi bl e noi se gi ven i n deci bel s that rel ates to
i ntensi ty at a poi nt some di stance from the noi se source. I t can
be rel ated to octave bands or as an overal l wei ghted l evel
dB(A).
Wei ghted sound l evel s rel ate the deci bel (l oudness) to a fre-
quency. Ears can easi l y pi ck up hi gh-frequency noi ses (both
i ntensi ty and di recti on) but are rel ati vel y i nsensi ti ve to l ow-
frequency noi se. For a stereo system hi gh-frequency speakers
must be very careful l y l ocated i n a room for best resul ts, but
l ow-frequency bass speakers can be pl aced anywhere, even out
of si ght.
There are three basi c wei ghti ng systems: A, B and C. The
"A" system, dB(A), most cl osel y rel ates to our ear, the "B" sys-
tem, dB(B), has some speci fi c uses and the "C" system, dB(C),
i s consi dered unwei ghted.
The dB(A) i s the most common wei ghti ng system. I t ex-
presses sound l evel s wi th a si ngl e number i nstead of havi ng
to speci fy the noi se of each octave band.
Note that the sound range of ACHEs (at cl ose range) i s typi -
cal l y between 80 and 105 dB(A).
ACHE Noise
Whether the concern i s for overal l pl ant noi se or the noi se
exposure of pl ant workers i n the vi ci ni ty of the fans, a di fferent
type of noi se speci fi cati on must be used.
Overal l , noi se l i mi tati ons from an ACHE are typi cal l y a
sound power l evel (PWL) speci fi cati on for each uni t. Thi s l i m-
i ts the contri buti on of each uni t (typi cal l y two fans) to the
pl ant noi se as a whol e. Thi s i s usual l y needed where noi se at
the pl ant boundary i s consi dered. Contri buti ons of each part
of the pl ant must be careful l y control l ed i f overal l pl ant noi se
i s l i mi ted. PWLs can be expressed as wei ghted l evel dB(A) or
someti mes even by l i mi tati ons on each octave band.
I f worker protecti on i s the mai n concern, a l i mi tati on of
sound pressure l evel at 1 m bel ow the bundl e wi l l 574-,- be
i mposed as "SPL dB(A) at 1 m". The OSHA l i mi tati on i s 90
dB(A) for ei ght-hour exposure, but 85 dB(A) down to 80 dB(A)
i s not uncommon.
Predicting Fan Noise
Each fan manufacturer has propri etary equati ons for pre-
di cti ng fan noi se. API Gui del i nes use the general formul a:
PWL = 56 + 30 l og

MPM
304.8

+ 10 l og HP Eq 10-4
Thi s cal cul ates PWL as dB(A)
Propri etary noi se equati ons are based on actual tests at
vari ous speeds and operati ng condi ti ons consi deri ng the fol -
l owi ng effects:
Fan di ameter
Fan ti p speed
Bl ade Type
Bl ade pi tch angl e
I nl et condi ti ons
Horsepower
Note: Logs are common l ogs (base 10).
For exampl e:
4.267 meter fan
237 RPM (3177 m/mi n ti p speed)
Altitude above Sea Level Barometer Atmospheric Pressure
meters feet mm Hg abs. in Hg abs. kPa (abs) psia
0 0 760.0 29.92 101.325
14.696
153 500 746.3 29.38
99.49
14.43
305 1000 733.0 28.86 97.63
14.16
458 1500 719.6 28.33
95.91
13.91
610 2000 706.6 27.82 94.18
13.66
763 2500 693.9 27.32
92.46
13.41
915 3000 681.2 26.82 90.80
13.17
1068 3500 668.8 26.33
89.15
12.93
1220 4000 656.3 25.84 87.49
12.69
1373 4500 644.4 25.37
85.91
12.46
1526 5000 632.5 24.90 84.32
12.23
1831 6000 609.3 23.99
81.22
11.78
2136 7000 586.7 23.10 78.19
11.34
2441 8000 564.6 22.23
75.22
10.91
2746 9000 543.3 21.39 72.39
10.50
3050 10 000 522.7 20.58
69.64
10.10
Repri nted wi th permi ssi on of John M. Campbel l and Company.
FIG. 10-20
Altitude and Atmospheric Pressures
8
10-17
25.1 HP
Fi nd sound power l evel
PWL= 56 + 30 l og 10.4 + 10 l og 25.1
= 100.5 dB(A)
When consi deri ng mul ti pl e noi se sources (fans) use the
rel ati on:
PWL
N
= PWL + 10 l og N Eq 10-5
The sound power l evel for 2 adjacent fans i s the PWL of
one fan pl us 10 l og 2 or PWL
2
= PWL + 3.
A doubl i ng of the noi se source adds 3 dB.
Noi se attenuates wi th di stance by the equati on:
SPL (at di stance R) = PWL - 20 l og (3.28R) Eq 10-6
Where R i s i n meters from the center of the source. Meas-
ure R as a "l i ne of si ght" di stance.
Consi der the noi se an observer at grade hears at 15.24 m
from an operati ng ACHE wi th the fan on and the l i ne-of-si ght
di stance from grade to the center of the fan i s actual l y 18.9 m.
Remember what the ear hears i s SPL, the noi se energy i s
PWL.
Assume PWL = 100.5 dB(A).
SPL = 100.5 - 20 l og 62 = 64.7 dB(A)
Thi s al so assumes background noi se i s at l east 10 dB qui eter.
Note: I f both fans were runni ng, the SPL woul d have been
67.7 dB(A).
When consi deri ng the noi se at 1 m beneath the uni t, the
dri ve system and motor noi se become domi nant at l ower ti p
speeds.
Factors that i nfl uence thi s noi se are:
Motor noi se
Bel t or gear noi se
Beari ng noi se
Refl ected noi se from supports
Background noi se
Gear noi se i s especi al l y si gni fi cant i n a forced draft uni t.
Noise Testing
Frequentl y, the ACHE must be tested for confi rmati on that
i ts noi se does not exceed speci fi cati ons i mposed by the pur-
chaser. There are two basi c types of tests normal l y performed
before shi pment:
a) Measure SPL dB(A) (Sound Pressure Level ) at "1 m be-
l ow the bundl e or fan guard" dependi ng on whether
the uni t i s i nduced or forced draft.
b) Measure PWL (Sound Power Level ) usi ng "hemi spheri -
cal power l evel test." PWL i s speci fi ed as ei ther a dB(A)
wei ghted val ue or by octave bands.
The "SPL at 1 m" test i s by far the most common and l east
expensi ve. Usual l y, onl y one or two measurements are re-
qui red. The answer i s i mmedi ate and read di rectl y from the
noi se meter.
The hemi spheri cal test i s far more compl i cated and expen-
si ve. Several hours, many techni ci ans and a l arge crane are
requi red to perform thi s test. Ful l detai l s of the test are gi ven
i n API Recommended Practi ce 631 M, June, 1981.
7
REFERENCES
1. A.P.I . Standard 661, Ai r Cool ed Heat Exchangers for General
Refi nery Servi ces.
2. Bri ggs, D. E., Young, E. H., Convecti on Heat Transfer and Pres-
sure Drop of Ai r Fl owi ng Across Tri angul ar Pi tch of Tubes,
Chemi cal Engi neeri ng Progress Symposi um Seri es, Vol ume 59,
No. 41, 1963.
3. Cook, E. M., Ai r Cool ed Heat Exchangers, Chemi cal Engi neer-
i ng, May 25, 1964, p. 137; Jul y 6, 1964, p. 131; and August 3,
1964, p. 97.
4. Gardner, K. A., Effi ci ency of Extended Surfaces, Trans ASME,
Vol ume 67, 1945, pp. 621-631.
5. Robi nson, K. K., Bri ggs, D. E., Pressure Drop of Ai r Fl owi ng
Across Tri angul ar Pi tch Banks of Fi nned Tubes, Chemi cal En-
gi neeri ng Progress Symposi um Seri es, Vol ume 62, No. 64, 1966.
6. Rubi n, Frank L., Wi nteri zi ng Ai r Cool ed Heat Exchangers, Hy-
drocarbon Processi ng, October 1980, pp. 147-149.
7. API Standard 631 M, Fi rst Edi ti on, "Measurement of Noi se from
Ai r-Cool ed Heat Exchangers," June, 1981.
8. John M. Campbel l Co. "Gas Condi ti oni ng and Processi ng, Vol . 2,"
Ei ghth Edi ti on.
10-18
Answer: t
2
= 130C (approxi matel y) from Fi g. 13-32.
Fi g. 13-34 gi ves the approxi mate brake power requi red for
the compressi on. I t i ncl udes compressor effi ci enci es i n the
range of 60 to 70%.
Example 13-5 Gi ven: Wei ght fl ow, w, = 30 000 kg/h
head = 150 kNm/kg
Fi nd: Brake Power
Answer: Gp = 1700 kW from Fi g. 13-34.
Fi g. 13-37 predi cts the approxi mate number of compressor
wheel s requi red to produce the head. I f the number of wheel s
i s not a whol e number, use the next hi ghest number.
Calculating Performance
When more accurate i nformati on i s requi red for compressor
head, gas horsepower, and di scharge temperature, the equa-
ti ons i n thi s secti on shoul d be used. Thi s method appl i es to a
gas mi xture for whi ch a P-H di agram chart i s not avai l abl e. To
cal cul ate the properti es of the gas, see Fi gs. 13-6 and 13-7. Al l
val ues for pressure and temperature i n these cal cul ati on pro-
cedures are the absol ute val ues. Unl ess otherwi se speci fi ed,
vol umes of fl ow i n thi s secti on are actual vol umes.
To cal cul ate the i nl et vol ume:
Q
(w) (8.314) (T
1
) (Z
1
)
(M) (P
1
)
Eq 13-25
I f we assume the compressi on to be i sentropi c (reversi bl e
adi abati c, constant entropy), then:
H
i s

ZRT
M (k 1)/k

P
2
P
1
_

,
(k 1)/k
1
1
1
]
Eq 13-26
Si nce these cal cul ati ons wi l l not be wheel -by-wheel , the
head wi l l be cal cul ated across the enti re machi ne. For thi s, use
the average compressi bi l i ty factor:
FIG. 13-25
Typical Centrifugal Compressor Cutaway
13-20
Zavg

Z
1 + Z
2

2
The heat capaci ty rati o, k, i s normal l y determi ned at the av-
erage sucti on and di scharge temperature (see Fi gs. 13-7 and
13-8).
Isentropic Calculation
To cal cul ate the head:
H
i s

Z
avg
RT
1
M (k 1)/k

P
2
P
1
_

,
(k 1)/k
1
1
1
]
Eq 13-27a
whi ch can al so be wri tten i n the form:
H
i s

8.314
M

Z
av g
T
1
(k 1)/k

P
2
P
1
_

,
(k 1)/k
1
1
1
]
Eq 13-27b
The gas horsepower can now be cal cul ated from:
Ghp
(w) (H
i s
)
(
i s
) (3 600)
Eq 13-28
FIG. 13-28
Compressor Performance, Low Compression Ratio
FIG. 13-29
Compressor Performance, Higher Compression Ratio
FIG. 13-27
Compressor Head
13-22
FIG. 13-33
Head
Z = 1
13-26
FIG. 13-34
Power Determination
FIG. 13-35
Efficiency Conversion
13-27
The approxi mate theoreti cal di scharge temperature can be
cal cul ated from:
T
i deal
T
1

P
2
P
1
_

,
(k 1)/k
1
1
1
]
Eq 13-29
T
2
T
1
+ T
i deal
Eq 13-30
The actual di scharge temperature can be approxi mated:
T
actual
T
1

P
2
P
1
_

,
(k 1)/k
1
1
1
]

i s
Eq 13-31
T
2
T
1
+ T
actual
Eq 13-32
Polytropic Calculation
Someti mes compressor manufacturers use a pol ytropi c path
i nstead of i sentropi c. Pol ytropi c effi ci ency i s defi ned by:
n
(n 1)

k
(k 1)
1
1
]

p
Eq 13-33
(See Fi g. 13-35 for conversi on of i sentropi c effi ci ency to
pol ytropi c effi ci ency.)
The equati ons for head and gas horsepower based upon
pol ytropi c compressi on are:
H
p

Z
avg
RT
1
M (n 1)/n

P
2
P
1
_

,
(n 1)/n
1
1
1
]
Eq 13-34a
whi ch al so can be wri tten i n the form:
H
p

8.314
M

Z
av g
T
1
(n 1)/n

P
2
P
1
_

,
(n 1)/n
1
1
1
]
Eq 13-34b
Ghp
(w) (H
p
)
(
p
) (3 600)
Eq 13-35
Pol ytropi c and i sentropi c head are rel ated by
H
p

H
i s

p

i s
Eq 13-36
The approxi mate actual di scharge temperature can be cal -
cul ated from:
T
2
T
1

P
2
P
1
_

n 1
n
_

,
Eq 13-37
Mechanical Losses
After the gas horsepower has been determi ned by ei ther
method, horsepower l osses due to fri cti on i n beari ngs, seal s,
and speed i ncreasi ng gears must be added.
Fi g. 13-36 shows l osses rel ated to the shaft speed and casi ng
si ze for conventi onal mul ti stage uni ts.
Beari ngs and seal l osses can al so be roughl y computed from
Scheel s equati on:
Mechani cal l osses (Gp)
0.4
Eq 13-38
To cal cul ate the brake horsepower:
Bhp Gp + mechani cal l osses Eq 13-39
Compressor Speed
The basi c equati on for esti mati ng the speed of a centri fugal
compressor i s:
N (N
nomi nal
)

H total
(No. of wheel s) (H max / wheel )
Eq 13-40
where the number of wheel s i s determi ned from Fi g. 13-37.
Nomi nal speeds to devel op 30 000 N m/kg of head/wheel
can be determi ned from Fi g. 13-23. However, to cal cul ate the
maxi mum head per wheel , the fol l owi ng equati on based on
mol ecul ar wei ght (or more accuratel y, densi ty) can be used.
H max/stage 15,000 1,500 (M)
0.35
Eq 13-41
Thi s equati on wi l l gi ve a head of 30 000 N m/kg for a gas
when M = 30 and 33 000 N m/kg when M = 16.
Casi ng Si ze
Max Fl ow
(i nl et m
3
/h)
Nomi nal Speed
(rpm)
1
12 700
10 500
2
33 900
8 200
3
56 000
6 400
4
93 400
4 900
5
195 000
3 600
6
255 000
2 800
FIG. 13-36
Mechanical Losses
13-28
economi c advantage versus methanol recovered by di sti l l a-
ti on. At cryogeni c condi ti ons (bel ow 40C) methanol usual l y
i s preferred because gl ycol s vi scosi ty makes effecti ve separa-
ti on di ffi cul t.
Ethyl ene gl ycol (EG), di ethyl ene gl ycol (DEG), and
tri ethyl ene gl ycol (TEG) gl ycol s have been used for hydrate
i nhi bi ti on. The most popul ar has been ethyl ene gl ycol because
of i ts l ower cost, l ower vi scosi ty, and l ower sol ubi l i ty i n l i qui d
hydrocarbons.
Physi cal properti es of methanol and methanol -water mi x-
tures are gi ven i n Fi g. 20-34 through Fi g. 20-37. Physi cal prop-
erti es of the most common gl ycol s and gl ycol -water mi xtures
are gi ven i n Fi g. 20-38 through Fi g. 20-49. Tabul ar i nformati on
for the pure gl ycol s and methanol i s provi ded i n Fi g. 20-56.
To be effecti ve, the i nhi bi tor must be present at the very
poi nt where the wet gas i s cool ed to i ts hydrate temperature.
For exampl e, i n refri gerati on pl ants gl ycol i nhi bi tors are typi -
cal l y sprayed on the tube-sheet faces of the gas exchangers so
that i t can fl ow wi th the gas through the tubes. As water con-
denses, the i nhi bi tor i s present to mi x wi th the water and pre-
vent hydrates. I njecti on must be i n a manner to al l ow good
di stri buti on to every tube or pl ate pass i n chi l l ers and heat
exchangers operati ng bel ow the gas hydrate temperature.
The vi scosi ti es of ethyl ene gl ycol and i ts aqueous sol uti ons
i ncrease si gni fi cantl y as temperature decreases, and thi s must
be al l owed for i n the rati ng of refri gerati on-pl ant exchangers
and chi l l ers.
The i nhi bi tor and condensed water mi xture i s separated
from the gas stream al ong wi th a separate l i qui d hydrocarbon
stream. At thi s poi nt, the water dewpoi nt of the gas stream i s
essenti al l y equal to the separati on temperature. Gl ycol -water
sol uti ons and l i qui d hydrocarbons can emul si fy when agi tated
or when expanded from a hi gh pressure to a l ower pressure,
e.g., JT expansi on val ve. Careful separator desi gn wi l l al l ow
nearl y compl ete recovery of the di l uted gl ycol for regenerati on
and rei njecti on. Fi g. 20-57 shows a fl ow di agram for a typi cal
EG i njecti on system i n a refri gerati on pl ant.
The regenerator i n a gl ycol i njecti on system shoul d be operated
to produce a regenerated gl ycol sol uti on that wi l l have a freezi ng
poi nt bel ow the mi ni mum temperature encountered i n the sys-
tem. Thi s i s typi cal l y 75-80 wt%. Fi g. 20-58 shows the freezi ng
point of various concentrations of gl ycol water sol uti ons.
The mi ni mum i nhi bi tor concentrati on i n the free water
phase may be approxi mated by Hammerschmi dts equati on.
25
d =
K
H
X
I
MW
I
(1 X
I
)
Eq 20-5
X
I
=
dMW
I
K
H
+ dMW
I
Eq 20-6
Where K
H
(gl ycol s) = 1297 to 2222 and
K
H
(methanol ) = 1297
The K
H
range of 1297 to 2222 for gl ycol s refl ects the uncer-
tai nty i n the val ue of thi s parameter. At equi l i bri um, such as
for a l aboratory test, 1297 i s appl i cabl e as i l l ustrated on Fi g.
20-60. I n some fi el d operati ons, however, hydrate formati on
has been prevented wi th gl ycol concentrati ons correspondi ng
wi th K
H
val ues as hi gh as 2222. Thi s i s because hydrate sup-
pressi on wi th gl ycol s depends on the systems physi cal and
fl ow characteri sti cs (i.o., tho systom dynamics, configuration,
Iocation and mothod of gIycoI injoction, amount of froo wator,
etc.) as wel l as the properti es of the gas and the gl ycol . There-
fore, i n the absence of rel i abl e fi el d-test data, a system shoul d
be desi gned for a K
H
of 1297. Once the system i s operati ng, the
gl ycol concentrati on can be reduced to tol erabl e l evel s.
Eq 20-5 and Eq 20-6 shoul d not be used beyond 20-25 wt%
for methanol and 60-70 wt% for the gl ycol s. For methanol con-
centrati ons up to about 50%, the Ni el sen-Buckl i n equati on
26
pro-
vides better accuracy:
d = 72.0 l n(x
H
2
O
) Eq 20-7
Note that x
H
2
O
i n Eq 20-7 i s a mol e fracti on, not a mass frac-
ti on. Expressi ng mol e fracti on i n terms of mass fracti on, dew-
poi nt depressi on i s pl otted agai nst the wei ght percent
methanol i n Fi g. 20-59.
Maddox et al.
27
presents a method of esti mati ng the requi red
i nhi bi tor concentrati on for both methanol and EG. The method
i s i terati ve but converges easi l y after a few i terati ons.
Fi gs. 20-60 thru 20-64 provi de a compari son of vari ous i n-
hi bi tor correl ati ons wi th experi mental data.
28,29,30
Experi men-
tal data at very hi gh i nhi bi tor concentrati ons i s l i mi ted.
Once the r equi r ed i nhi bi tor concentr ati on has been cal -
cul ated, the mass of i nhi bi tor r equi r ed i n the water phase
may be cal cul ated fr om Eq 20-8
m
I
=
X
R
m
H
2
O
X
L
X
R
Eq 20-8
The amount of inhibitor to be injected not only must be sufficient
to prevent freezing of the inhibitor water phase, but also must be
sufficient to provide for the equilibrium vapor phase content of the
inhibitor and the solubility of the inhibitor in any liquid hydrocarbon.
The vapor pressure of methanol is high enough that significant
quantities will vaporize. Methanol vaporization losses may be esti-
mated from Fig. 20-65
31
Fig. 20-65 is extrapolated above 4800 kPa
(abs). Recent studies indicate Fig. 20-65 may underestimate vapor
phase methanol losses at higher pressures. Glycol vaporization
losses are generally very small and are typically ignored in calcula-
tions.
FIG. 20-30
Solution Sketch for Example 20-8
20-19
tor countercurrent to the gas fl ow. Water-ri ch gl ycol i s removed
from the bottom of the contactor, passes through the refl ux
condenser coi l , fl ashes off most of the sol ubl e gas i n the fl ash
tank, and fl ows through the ri ch-l ean heat exchanger to the
regenerator. I n the regenerator, absorbed water i s di sti l l ed
from the gl ycol at near atmospheri c pressure by appl i cati on of
heat. The regenerated l ean gl ycol exi ts the surge drum, i s
partl y cool ed i n the l ean-ri ch exchanger and i s pumped
through the gl ycol cool er before bei ng reci rcul ated to the con-
tactor.
Eval uati on of a TEG system i nvol ves fi rst establ i shi ng the
mi ni mum TEG concentrati on requi red to meet the outl et gas
water dewpoi nt speci fi cati on. Fi g. 20-68 shows the water dew-
poi nt of a natural gas stream i n equi l i bri um wi th a TEG sol u-
ti on at vari ous temperatures and TEG concentrati ons. Fi g. -
20-68 can be used to esti mate the requi red TEG concentrati on
for a parti cul ar appl i cati on or the theoreti cal dewpoi nt depres-
si on for a gi ven TEG concentrati on and contactor temperature.
Actual outl et dewpoi nts depend on the TEG ci rcul ati on rate
and number of equi l i bri um stages, but typi cal approaches to
equi l i bri um are 611C. Equi l i bri um dewpoi nts are rel ati vel y
i nsensi ti ve to pr essur e and Fi g. 20-68 may be used up to
10 300 kPa (abs) wi th l i ttl e error.
Fi g. 20-68 combi nes the resul ts publ i shed by Parri sh, et al .
35
coveri ng TEG puri ty up to 99.99 mass percent and those re-
ported by Buckl i n and Won i n 1987
52
coveri ng TEG puri ty up
FIG. 20-61
Hydrate Inhibition with Methanol:
Hammerschmidt vs. Experimental Data
28
FIG. 20-62
Hydrate Inhibition with Methanol:
Nielsen & Bucklin vs. Experimental Data
28
FIG. 20-63
Hydrate Inhibition with Methanol:
Nielsen & Bucklin
26
vs. Experimental Data
29
FIG. 20-64
Hydrate Inhibition with Methanol:
Maddox et al.
27
vs. Experimental Data
29
20-31
The sol ubi l i ti es of pure H
2
S and CO
2
i n pure TEG are gi ven
i n Fi g. 20-76. These charts appl y to pure TEG or to the l ean
TEG, whi ch i s essenti al l y pure. For the ri ch-TEG sol uti on l eav-
i ng the bottom of a contactor, i t i s approxi matel y correct to
appl y them to the TEG porti on.
Heavi er paraffi n hydrocarbons are essenti al l y i nsol ubl e i n
TEG. Aromati c hydrocarbons, however, are very sol ubl e i n
TEG, and si gni fi cant amounts of aromati c hydrocarbons may
be absorbed i n the TEG at contactor condi ti ons. Thi s may pre-
sent an envi ronmental or safety hazard when they are di s-
charged from the top of the regenerator.
Vapor-l i qui d equi l i bri um constants (K-val ues) for benzene,
tol uene, ethyl benzene, and o-xyl ene i n TEG sol uti ons are pre-
sented i n RR-131.
43
Thi s data i ndi cates that at typi cal contac-
tor condi ti ons approxi matel y 10-30% of the aromati cs i n the
gas stream may be absorbed i n the TEG sol uti on. Based on the
data i n RR-131 at 6900 kPa (abs), 38C and a TEG ci rcul ati on
rate of 25 l i ters TEG/kg H
2
O absorbed, the approxi mate per-
centage absorpti on of aromati cs from the gas i nto the ri ch TEG
are cal cul ated as:
Benzene 10%
Tol uene 14%
Ethyl benzene 19%
O-xyl ene 28%
Aromati c absorpti on i ncreases wi th i ncreasi ng pressure and
decreasi ng temperature. Aromati c absorpti on i s di rectl y re-
l ated to TEG ci rcul ati on rate. Hi gher ci rcul ati on rates resul t
i n i ncreased absorpti on. Aromati c absorpti on i s essenti al l y i n-
dependent of the number of contacts i n the absorber so one
method of mi ni mi zi ng aromati c absorpti on i s to use tal l er con-
tactors and mi ni mi ze TEG ci rcul ati on rates.
Most of the aromati c components wi l l be stri pped from the
TEG sol uti on i n the regenerator.
Fl ash tank si zi ng shoul d be suffi ci ent to degas the gl ycol
sol uti on and ski m entrai ned l i qui d hydrocarbons, i f necessary.
A mi ni mum retenti on ti me of 3-5 mi nutes i s requi red for de-
gassi ng. I f l i qui d hydrocarbons are to be removed as wel l , re-
tenti on ti mes of 20-30 mi nutes may be requi red for adequate
separati on. Fl ash tank pressures are typi cal l y l ess than 520
kPa (abs).
Regenerator si zi ng requi res establ i shi ng the reboi l er duty
and, when hi gh TEG concentrati ons are requi red, provi di ng
suffi ci ent stri ppi ng gas.
A qui ck esti mate of reboi l er duty can be made usi ng Eq 20-10.
Q = (0.116) (L
g
) Eq 20-10
Eq 20-10 i s approxi mate and usual l y gi ves val ues whi ch are
hi gher than the actual duty. A more ri gorous determi nati on of
reboi l er duty i s shown i n Exampl e 20-12.
Example 20-12 Determi ne reboi l er duty for condi ti ons i n the
previ ous exampl e. Assume the ri ch TEG temperature enteri ng
the regenerator i s 150C and the reboi l er temperature i s 200C.
Gl ycol Reboi l er Duty: Basi s 1 m
3
TEG
Sensi bl e Heat:
Q
s
= m C
p
T
FIG. 20-69
Water Removal vs. TEG Circulation Rate at Various
TEG Concentrations (N = 1.0)
FIG. 20-70
Water Removal vs. TEG Circulation Rate at Various
TEG Concentrations (N = 1.5)
FIG. 20-71
Water Removal vs. TEG Circulation Rate at Various
TEG Concentrations (N = 2.0)
20-34
and the peri od of regenerati on shoul d be decreased by mul ti -
pl yi ng i t ti mes the rati o V /V
mi n
. A method more exact than
Fi g. 20-88 for cal cul ati ng the mi ni mum superfi ci al vel oci ty i s
to use Eq 20-16, but to consi der i t i n terms of (P/L)
mi n
and
V
mi n
i nstead of (P/L)
max
and V
max
.
General Comments
The regenerati on cycl e frequentl y i ncl udes depressur-
i ng/repressuri ng to match the regenerati on gas pressure
and/or to maxi mi ze the regenerati on gas vol ume to meet the
vel oci ty cri teri on. I n these appl i cati ons, the rate of depressur-
i ng or repressuri ng shoul d not exceed 350 kPa per/mi nute.
Some appl i cati ons, termed pressure swi ng adsorpti on, regen-
erate the bed onl y wi th depressuri zati on and sweepi ng the bed
wi th gas just above atmospheri c pressure.
Moi sture anal yzers for very l ow water contents requi re care
to prevent damage to the probes. When i nserted i nto the beds,
sampl e probes and temperature probes must be i nstal l ed to
reach the center of the gas phase.
Sol i d desi ccant towers are i nsul ated external l y or possi bl y
i nternal l y. I nternal refractory requi res careful i nstal l ati on
and curi ng, usual l y before the desi ccant i s i nstal l ed. I t saves
energy but the greatest benefi t i s i t can dramati cal l y reduce
the requi red heati ng and cool i ng ti mes. Thi s i s often an i mpor-
tant benefi t for systems where regenerati on ti mes are l i mi ted.
The pri mary di sadvantage i s the potenti al for wet gas bypass-
i ng the desi ccant through cracks and defects i n the i nsul ati on
duri ng the adsorpti on cycl e.
Example 20-13: 2.85 10
6
Sm
3
/day of natural gas wi th a mo-
l ecul ar wei ght of 18 i s to be processed for ethane recovery i n
a turbo-expander pl ant. I t i s water saturated at 4140 kPa (abs)
and 38C and must be dri ed to 101C dew poi nt. Determi ne
the water content of the gas, and the amount of water that
must be removed; and do a prel i mi nary desi gn of a mol ecul ar-
si eve dehydrati on system consi sti ng of two towers wi th down-
fl ow dehydrati on i n one tower and up-fl ow regenerati on i n the
other. Use 4A mol ecul ar si eve of 3.2 mm beads (i .e., 4x8 mesh).
The regenerati on gas i s part of the pl ants resi due gas, whi ch
FIG. 20-84
Mole Sieve Capacity Correction for Unsaturated Inlet Gas
FIG. 20-85
Mole Sieve Capacity Correction for Temperature
FIG. 20-86
Inlet and Outlet Temperatures During Typical Solid
Desiccant Bed Regeneration Period
FIG. 20-87
Allowable Velocity for Mole Sieve Dehydrator
20-42
Chemistry The overal l equi l i bri um reacti ons appl i ca-
bl e for H
2
S and CO
2
and pri mary and secondary ami nes are
shown bel ow wi th a pri mary ami ne
9
. A qual i tati ve esti mati on
of the vel oci ty of the reacti on i s gi ven.
For hydrogen sul fi de removal
RNH
2
+ H
2
S RNH
3
+
+ HS

Fast Eq. 21-1

RNH
2
+ HS

RNH
3
+
+ S

Fast Eq. 21-2
The overal l reacti ons between H
2
S and ami nes are si mpl e
si nce H
2
S reacts di rectl y and rapi dl y wi th al l ami nes to form
the bi sul fi de by Eq. 21-1 and the sul fi de by Eq. 21-2.
For carbon di oxi de removal
2 RNH
2
+ CO
2
RNH
3
+
+ RNHCOO

Fast Eq. 21-3

RNH
2
+ CO
2
+ H
2
O RNH
3
+
+ HCO
3

Sl ow Eq. 21-4
RNH
2
+ HCO
3

RNH
3
+
+ CO
3

Sl ow

Eq. 21-5
FIG. 21-3
Approximate Guidelines for Amine Processes
1
MEA DEA
(9)
DGA

Sulfinol MDEA
(9)
Aci d gas pi ckup, m
3
/100L @ 38C,
normal range
(2)
2.33.2 2.856.4 3.55.40 3.012.75 2.26.4
Aci d gas pi ckup, mol /mol ami ne,
normal range
(3)
0.330.40 0.200.80 0.250.38 NA 0.200.80
Lean sol uti on resi dual aci d gas,
mol /mol ami ne, normal range
(4)
0.12 0.01 0.06 NA 0.0050.01
Ri ch sol uti on aci d gas l oadi ng,
mol /mol ami ne, normal range
(3)
0.450.52 0.210.81 0.350.44 NA 0.200.81
Sol uti on concentrati on, wt%, normal
range 1525 3040 5060
3 components.
Vari es. 4050
Approxi mate reboi l er heat duty, kJ/L
l ean sol uti on
(5)
280335 235280 300360 100210 220250
Steam heated reboi l er tube bundl e,
approx. average heat fl ux,
Q/A = MJ/(h m
2
)
(6)
100115 7585 100115 100115 7585
Di rect fi red reboi l er fi re tube,
average heat fl ux,
Q/A = MJ/(h m
2
)
(6)
90115 7585 90115 90115 7585
Recl ai mer, steam bundl e or fi re tube,
average heat fl ux,
Q/A = MJ/(h m
2
)
(6)
7090 NA
(7)
7090 NA NA
(7)
Reboi l er temperature, normal
operati ng range, C
(8)
107127 110127 121132 110138 110132
Heats of reacti on;
(10)
approxi mate:
kJ/kg H2S 1420 1290 1570 N/A 1230
kJ/kg CO2 1920 1700 2000 N/A 1425
NA not appl i cabl e or not avai l abl e
NOTES:
1. These data al one shoul d not be used for speci fi c desi gn purposes. Many desi gn factors must be consi dered for actual pl ant desi gn.
2. Dependent upon aci d gas parti al pressures and sol uti on concentrati ons.
3. Dependent upon aci d gas parti al pressures and corrosi veness of sol uti on. Mi ght be onl y 60% or l ess of val ue shown for corrosi ve systems.
4. Vari es wi th stri pper overhead refl ux rati o. Low resi dual aci d gas contents requi re more stri pper trays and/or hi gher refl ux rati os yi el di ng l arger
reboi l er duti es.
5. Vari es wi th stri pper overhead refl ux rati os, ri ch sol uti on feed temperature to stri pper and reboi l er temperature.
6. Maxi mum poi nt heat fl ux can reach 230 to 285 MJ/(h m
2
) at hi ghest fl ame temperature at the i nl et of a di rect fi red fi re tube. The most sati sfactory
desi gn of fi retube heati ng el ements empl oys a zone by zone cal cul ati on based on thermal effi ci ency desi red and l i mi ti ng the maxi mum tube wal l
temperature as requi red by the sol uti on to prevent thermal degradati on. The average heat fl ux, Q/A, i s a resul t of these cal cul ati ons.
7. Recl ai mers are not used i n DEA and MDEA systems.
8. Reboi l er temperatures are dependent on sol uti on conc. fl are/vent l i ne back pressure and/or resi dual CO2 content requi red. I t i s good practi ce to operate
the reboi l er at as l ow a temperature as possi bl e.
9. Accordi ng to Total .
10. The heats of reacti on vary wi th aci d gas l oadi ng and sol uti on concentrati on. The val ues shown are average.
10.
21-5