SECTION 19

Fractionation and Absorption

Fractionation
Fractionation is a unit operation utilized to separate mix­ of a separation is directly related to the relative volatility
tures into individual products. Fractionation involves sep­ of the components and the required purity of the product
arating components by relative volatility (a). The difficulty streams.

FIG. 19-1
Nomenclature

a′t = tube flow area, m2 NP = number of passes in a tray

at = total tube flow area, m2 Nt = number of tubes
A = absorption factor used in Equation 19-28 ∆P = pressure drop, kPa
Ac = cross sectional area, m2 q = moles of saturated liquid in the feed per mole of feed
At = heat transfer surface, m2 Q = heat transfer duty, kw
AAM = tray active area, m2 Qc = condenser duty, kw
ADM = tray downcomer area, m2 R = reflux ratio, moles of reflux divided by moles of
ATM = tower cross sectional area, m2 net overhead product
b = exponent in Equation 19-5 and 19-6 Re = Reynolds number, dimensionless
B = bottoms product rate, moles/unit time s = relative density
C = coefficient in Equation 19-11, m/h S = number of stages
CAF = vapor capacity factor, corrected, m/s ST = stripping factor used in Equation 19-31
CAFo = vapor capacity factor, uncorrected, m/s SF = separation factor defined by Equation 19-1
CFS = vapor loading, m3/s TS = tray spacing, mm
D′ = diameter, m UD = overall heat transfer coefficient, W/(m2 • °C)
D = distillate (overhead) product rate, moles/unit time v = specific volume, m3/kg
DT = tower diameter, m vmax = maximum velocity, m/h
Ea = absorption efficiency, Equation 19-30 vi = specific volume of the inlet, m3/kg
Es = stripping efficiency, Equation 19-32 vo = specific volume of the outlet, m3/kg
f = friction factor (Moody friction factor/100), m2/dm2 V = vapor rate, moles/unit time
F = feed rate, moles/unit time V1 = vapor rate leaving top tray, moles/unit time
Fp = packing factor Vmax = volumetric vapor flow rate, m3/h
FF = flooding factor used in Equation 19-17 VDdsg
*
= downcomer velocity (uncorrected), m3/h/m2
FPL = flow path length, m VDdsg = downcomer velocity (corrected), m3/h/m2
Gt = mass velocity, kg/(h • m2) Vload = vapor loading defined by Equation 19-13, m3/s
Gp = tower vapor loading, kg/(m2 • s) Vo = stripping medium rate, moles/unit time
GPM = tower liquid loading, m3/min w = weight flow, kg/h
H = enthalpy, kJ/kg x = liquid mole fraction
HETP = height of packing equivalent to a theoretical plate X = liquid rate, moles/unit time
HTU = height of a transfer unit Xm + 1 = moles of a component in the rich oil entering a
K = equilibrium K-value, y/x stripper per mole of rich oil entering the stripper
Lo = liquid reflux rate, moles/unit time x1 = moles of a component in the lean oil per mole of
Lp = liquid loading, kg/(m2 • sec) rich oil
L = liquid rate, moles/unit time Xo = moles of a component in the liquid in equilibrium
Lt = tube length, m with the stripping medium per mole of entering
rich oil
Lm + 1 = rich oil entering the stripper, moles/unit time
y = vapor mole fraction
m = number of stripping stages
Yi = moles of any component in the lean gas leaving
M = mass flow rate, kg/h
the absorber per mole of rich gas
n = number of absorber stages
Yn + 1 = moles of any component in the entering rich gas
Nm = minimum number of theoretical stages per mole of rich gas

19-1
 FIG. 19-1 (Cont’d)
Nomenclature

Yo = moles of any component in the gas in equilibrium BP = bubble point feed stream
with the entering lean oil, per mole of rich gas bottom = bottom of the column
Z = static head, m calc = calculated value
corr = corrected value
Greek
D = distillate (overhead)
α = relative volatility F = feed
βij = volatility factor defined in Equation 19-5 G = gas
θ = correlating parameter in Equation 19-7, 19-8 HK = heavy key
σ = surface tension, dyne/cm i = any component
ρ = density, kg/m3 L = liquid
ε = efficiency LK = light key
µ = viscosity, Pa • s m = minimum
Subscripts n = tray number
avg = average value top = top of the column
B = bottoms VF = vaporized feed stream
v = vapor phase

Virtually all gas processing plants producing natural gas duce stripping vapors. The vapor rises through the column con­
liquids require at least one fractionator to produce a liquid tacting the descending liquid. The vapor leaving the top of the
product which will meet sales specifications. The schematic of column enters the condenser where heat is removed by some
an example fractionator in Fig. 19-2 shows the various compo­ type of cooling medium. Liquid is returned to the column as re­
nents of the system. Heat is introduced to the reboiler to pro­ flux to limit the loss of heavy components overhead.
Internals such as trays or packing promote the contact be­
FIG. 19-2 tween the liquid and vapor streams in the column. Intimate
Fractionation Schematic Diagram contact of the vapor and liquid phases is required for efficient
separation. Vapor entering a separation stage will be cooled
which results in some condensation of heavier components. The
Condenser liquid phase will be heated which results in some vaporization
of the lighter components. Thus, the heavier components are
Cooling medium concentrated in the liquid phase and eventually become the bot­
Vapor tom product. The vapor phase is continually enriched in the
light components which will make up the overhead product.
The vapor leaving the top of the column may be totally or
Reflux
Reflux drum partially condensed. In a total condenser, all vapor entering the
condenser is condensed to liquid and the reflux returned to the
column has the same composition as the distillate or overhead
product. In a partial condenser, only a portion of the vapor en­
Overhead
product tering the condenser is condensed to liquid. In most partial con­
densers only sufficient liquid will be condensed to serve as re­
Rectifying
section
Internals of various flux for the tower. In some cases, however, more liquid will be
kinds promote transfer condensed than is required for reflux and there will actually be
two overhead products, one a liquid having the same composi­
between countercurrent
liquid and vapor
Feed tion as the reflux and the other a vapor product which is in
equilibrium with the liquid reflux.

Stripping ­Equilibrium Stage Concept

All calculations are performed using theoretical (equilibri­
section

um) stages. A fractionation column may be considered as a se­

Reboiler
ries of equilibrium flashes with two feeds and two product
Vapor streams (Fig. 19-3). The vapor enters the flash from the stage
Steam or other below at a higher temperature and the liquid stream enters
heat source
from the stage above at a lower temperature. Heat and mass
transfer occur in this stage such that the exiting streams are a
bubble point liquid and dew point vapor at the same tempera­
Bottom ture and pressure. The compositions of these phases are related
product by the equilibrium relationship of yi = Kixi (See Section 25).

19-2
This relationship, along with heat and material balance consid­
erations, is the basis for all fractionator design. FIG. 19-3
Basic Fractionation Model
­Types of Fractionators
The number and type of fractionators required depend on Overhead Product

the number of products to be made and the feed composition. Heat

Typical NGL products from a fractionation process include:
Removed

• Demethanized Product (C2+)

• Deethanized Product (C3+)
• Ethane/Propane mixtures (EP) Equilibrium Stage

Commercial Propane
Equilibrium Liquid from
• Liquid Vapor Stage Above
Feed
• Propane/Butane mixture (LPG)
• Butane(s)
• Butane/Gasoline mixtures
• Natural Gasoline Vapor from Equilibrium
Stage Below Liquid
• Mixtures with a vapor pressure specification
See Section 1 for definitions and Section 2 for product specifica­
tions. Heat
Input Bottom
An example fractionation train used to produce three prod­ Product

ucts is illustrated in Fig. 19-4. The feed stream contains too

much ethane to be included in the products; thus, the first

FIG. 19-4
Fractionation Train

Moles/hr

◊1 ◊2 ◊3 ◊4 ◊5 ◊6 ◊7
C1 1.5 1.5
C2 24.6 22.2 2.4 2.4
C3 170.3 7.5 162.8 161.9 0.9 0.9
iC4 31.0 31.0 0.9 30.1 30.1
nC4 76.7 76.7 76.7 72.1 4.6
C5+ 76.5 76.5 76.5 0.9 75.6
Total 380.6 31.2 349.4 165.2 184.2 104.0 80.2
m3/day 156.48 117.95 110.87

19-3
column is a deethanizer. The overhead stream is recycled to the PRODUCT SPECIFICATIONS
upstream processing plant or sent to a fuel system. The bottom
product from this column could be marketed as a deethanized A material balance around the column is the first step in
product. The second column, a depropanizer, produces a specifi­ fractionation calculations. In order to perform this balance, as­
cation propane product overhead. The bottom product, a bu­ sumption of the product stream compositions must be made.
tane-gasoline mixture, is often sold to a pipeline without fur­ There are three ways of specifying a desired product from a
ther processing. The third column, a debutanizer, separates the fractionator:
butane and gasoline products. This separation is controlled to • A percentage recovery of a component in the overhead or
limit the vapor pressure of the gasoline. The overhead butane bottom stream.
product can be sold as a mixture or an additional column can be
used to separate the iso-butane and normal-butane. • A composition of one component in either product.
Another class of fractionators uses no external reflux con­ • A specific physical property, such as vapor pressure, for
denser to produce liquid for contact with the fractionator vapor. either product.
One such tower is a demethanizer as is found in cryogenic plants
(Fig. 19-5). The top feed being 12 mol % liquid at low tempera­ The recovery and composition specifications can be used di­
ture provides the reflux. This liquid along with the other feeds rectly in the material balance. However, property specifications
provides the liquid loading for this tower. The reboiler is the con­ are used indirectly. For instance if vapor pressure is a desired
trol point for the bottom product purity. The overhead composi­ specification of a product, a material balance is performed with an
tion is a function of the upstream process units. assumed component split. The calculated vapor pressure of the
resulting stream is then compared with the desired value and the
Applications also exist for a fractionator with a top liquid material balance redone until reasonable agreement is reached.
feed. These are commonly used for crude stabilizers or deetha­
nizers. As with the demethanizer, this column produces a spec­ In a multicomponent mixture, there are typically two com­
ification bottom product and an overhead stream whose compo­ ponents which are the “keys” to the separation. The light key
sition is determined by upstream process units. This is an component is defined as the lightest component in the bottom
economical approach to producing a single product but is limit­ product in a significant amount. The heavy key component is
ed in separation efficiency. Better recovery or sharper separa­ the heaviest component in the overhead product in a significant
tion can be achieved by adding a reflux condenser and rectifying amount. Normally, these two components are adjacent to each
section. other in the volatility listing of the components. For hand calcu­
lations, it is normally assumed for material balance purposes
that all components lighter than the light key are produced
overhead and all components heavier than the heavy key are
produced with the bottom product. By definition, the key com­
FIG. 19-5 ponents will be distributed between the product streams.
Demethanizer Example Example 19-1 — For the given feed stream, estimate the prod­
uct stream compositions for 98% propane recovered in the over­
head product with a maximum iso-butane content of the over­
head stream of 1%.

Feed:
C2 2.4
C3 162.8
iC4 31.0
nC4 76.7
C5 76.5

349.4 moles

Solution Steps

For Propane (light key):

Moles in overhead = (0.98) 162.8 = 159.5
Moles in bottoms = 162.8 – 159.5 = 3.3

For Ethane:
Moles in overhead = 2.4 (100% to overhead)
Since the isobutane (the heavy key) is 1% of the overhead
stream, the sum of propane and ethane must be 99% (all n-C4
and C5+ are in the bottoms). Thus:
159.5 + 2.4 161.9
Total Overhead Moles = = = 163.5
0.99 0.99
Moles of iC4 = 163.5 – 161.9 = 1.6

19-4
 The overall balance is: od. Practical exchanger design limits the process to a 10°C ap­
proach to the ambient summer temperature. This translates to
Feed Overhead Bottoms a process temperature of 46 to 52°C in most locations. With
Comp. Moles Moles Mole % Moles Mole % cooling water, process temperatures of 35 to 41°C are possible.
Below about 35°C, mechanical refrigeration must be used to
C2    2.4    2.4    1.5 — —
achieve the desired condensing temperature. This is the most
C3 162.8 159.5 97.5    3.3    1.8 expensive cooling method from both a capital and operating
iC4 31.0    1.6    1.0 29.4 15.8 cost standpoint. Generally, it is desirable to operate at as low a
pressure as possible to maximize the relative volatility between
nC4 76.7 — — 76.7 41.2
the key components of the separation. However, if reducing the
C5 76.5 — — 76.5 41.2 pressure requires a change to a more expensive cooling method,
Total 349.4 163.5 100.0 185.9 100.0 this is usually not a desirable choice.
In actual operation, the lighter than light key components In some cases, the overhead from the column must be com­
and heavier than heavy key components will not be perfectly pressed to sales or another process unit. In this case a higher op­
separated. For estimation purposes and hand calculations, per­ erating pressure may be desired to reduce compression power.
fect non-key separation is a useful simplifying assumption.
Other items must be considered which will limit pressure se­
­Key Parameters lection. If an operating pressure is too high, the critical tempera­
ture of the bottom product may be exceeded and the desired
Two important considerations which affect the size and cost separation cannot be achieved. Additionally, the pressure cannot
of a fractionation column are degree of separation and compo­ exceed the critical pressure of the desired overhead product.
nent volatility.
The selection of a partial or total condenser is fixed by the
The degree of separation or product purity has a direct im­ disposition of the overhead product. A total condenser is used
pact on the size of the column and the required utilities. Higher for a liquid product and a partial condenser for a vapor product.
purity will require more trays, more reflux, larger diameter, However, a liquid product can be produced as a vapor and sub­
and/or a reduced product quantity. One quantitative measure sequently cooled and/or compressed to produce a liquid product.
of the difficulty of a separation is the separation factor, SF, de­ There may be cases where this downstream liquefaction is eco­
fined as: nomically attractive. In most cases, the fractionation system for
a partial condenser will be cheaper and will have to be balanced
XD XB against the cost of additional downstream equipment. Before a
 XB  LK  XD  HK
SF =
Eq 19-1 reliable economic comparison can be made, the column design
must be made for each type condenser for a number of reflux
Note that Equation 19-1 defines the specification for the ratios and operating pressures.
tower design.
Typically, for most fractionation problems this factor ranges ­Reflux Ratio and Number of Stages
from around 500 to 2000. However, for sharp separations, it can The design of a fractionation column is a capital cost versus
be in the 10,000 range. The number of trays will be roughly energy cost trade-off problem. The primary parameters are the
proportional to the log of the separation factor for a given sys­ number of stages and the reflux ratio. Reflux ratio may be de­
tem. fined in several ways. For most calculations, reflux ratio is de­
The volatility of the components is usually expressed as fined as the ratio of the molar rate of reflux liquid divided by the
relative volatility, α. This quantity is computed as the ratio of molar rate of net overhead product. The reboiler duty is a direct
the equilibrium K-values of two components at a given tem­ function of the reflux ratio as the fractionating column must
perature and pressure. For fractionation calculations the α of maintain an overall heat and material balance for a given sepa­
the key components is important. Therefore: ration.

α = KLK/KHK Eq 19-2 A fractionation column can only produce a desired separa­

tion between the limits of minimum reflux and minimum stag­
This is a measure of the ease of separation. The larger α is, es. Fig. 19-6 illustrates the relationship between reflux ratio
the easier is the separation. and number of stages for a given separation. At minimum re­
flux an infinite number of stages is required. At total reflux a
minimum number of stages is required. Neither of these situa­
­DESIGN CONSIDERATIONS tions represents actual operation but are the extreme limits of
possible design configurations. Methods have been developed to
­Operating Pressure calculate both these cases in a rigorous manner.1, 2 However,
Before any design calculations can be made on a fraction­ these methods require a computer solution of tray-by-tray cal­
ation problem, a tower operating pressure must be determined. culations. In order to begin a detailed design, estimates of min­
One of the primary considerations for operating pressure is the imum reflux ratio and minimum trays should be generated
cooling medium available for the reflux condenser. The over­ from simple methods using a binary key component analysis.
head product will be at bubble point conditions for a liquid prod­
uct or at dew point conditions for a vapor product. The bubble ­Minimum Stages
point (or dew point) pressure is fixed by a desired component The minimum stages can be calculated for most multicom­
separation and the temperature of the cooling medium. ponent systems by the Fenske equation.3
The cooling media typically used are air, water, and refrig­ log SF
erant. Air cooling is normally the least expensive cooling meth­ Sm =
log (αavg) Eq 19-3

19-5
 n
XFi
FIG. 19-6 1–q= ∑
i=1 (αi – θ)/αi Eq 19-7
Relationship Between Reflux Ratio and Number of Stages
Once θ is determined, the minimum reflux ratio is:
n
XDi
(Lo/D)m + 1 = Rm + 1 = ∑
i=1 (αi – θ)/αi Eq 19-8
∞ Stages

­Number of Stages
The number of theoretical stages required for a given sepa­
ration at a reflux ratio between minimum and total reflux can
Number of Stages

be determined from empirical relationships. Erbar and Mad­

dox6 made an extensive investigation of tray-by-tray fraction­
ator calculations and developed the correlation in Fig. 19-7.
This correlation relates the ratio of minimum stages to theo­
retical stages to the minimum reflux ratio, Rm, and the operat­
ing reflux ratio, R (where R = Lo/D).
Fig. 19-7 can be used to determine an operating reflux for a
given number of stages by entering the figure at the value of
Sm/S, moving up to the line representing the value of Rm/(Rm + 1)
and reading a value of R/(R + 1). The optimum operating reflux
Minimum
Stages
ratio has been found to be near the minimum reflux ratio. Val­
ues of 1.2 to 1.3 times the minimum are common.7 Thus for a
Total Reflux

given R, a value of S can be determined from Fig. 19-7.

Reflux Ratio
Minimum This correlation is generated on the basis of bubble point
Reflux feed. If the feed is between the bubble point and dew point then
the operating reflux should be corrected. Erbar and Maddox6
proposed the following relationship to adjust the vapor rate
from the top tray for non-bubble point feed:
D
 F
Sm in this equation includes a partial reboiler and a partial con­ 1– [F(HVF – HBP)]
denser if these items are used. Vcorr = Vcalc + Eq 19-9
QC
 Lo  calc
The αavg is the column average relative volatility of the key
components in the separation. Various averaging techniques
have been proposed such as square root averaging of the α at The reflux rate can then be adjusted by material balance
the top and bottom of the column. The most common approach since:
is to use an arithmetic average:
Lo = V1 – D Eq 19-10
αtop + αbottom
αavg = ­Computation Method
2 Eq 19-4
If volatility varies widely, the approach of Winn4 is suggested, In order to determine the design parameters for a fraction­
in which a modified volatility is used: ation problem, the following method is recommended:

βij = KLK/KHK
b
Eq 19-5 1. Establish feed composition, flow rate, temperature, and
pressure.
where the exponent b is obtained from K-value plots over the
range of interest. 2. Make product splits for the column and establish con­
denser temperature and column pressure. From column
The minimum stage calculation is: pressure, calculate the reboiler temperature.
XD XB b B 1–b 3. Calculate minimum number of theoretical stages from
  XB  LK  XD  HK  D  
log
the Fenske equation (Equation 19-3).
Sm = Eq 19-6
log βij 4. Calculate minimum reflux rate from the Underwood
Note that Sm includes the partial condenser and partial reboiler equations (Equation 19-7 and 19-8).
if they exist. 5. Obtain theoretical stages/operating reflux relation from
Fig. 19-7.
­Minimum Reflux Ratio
6. Adjust actual reflux for feed vaporization if necessary
The Underwood method5 is the most widely used of the
(Equation 19-9, 19-10).
methods for calculating minimum reflux ratio. Underwood as­
sumed constant relative volatility and constant molal overflow
in the development of this method. The first step is to evaluate
θ by trial and error:

19-6
 FIG. 19-7 Solution Steps
Erbar-Maddox Correlation of Stages vs Reflux6 Estimate Product Splits from Material Balance:

Overhead Bottoms
Moles Mol % Moles Mol %
C2 21.5 4.1 — —
C3 495.4 94.9 10.2 2.0
iC4 5.2 1.0 99.8 19.3
nC4 — — 250.1 48.4
iC5 — — 56.2 10.9
nC5 — — 50.0 9.7
C6 — — 50.4 9.7
Totals 522.1 100.0 516.7 100.0

Get tower pressure at 49°C (bubble point calculation). Using

K-values from Section 25:

K α
C2 2.80
C3 0.93 2.067
iC4 0.45 bubble point pressure = 1930 kPa

Estimate the bottom temperature using K-values at 280 psia

Example 19-2 — For the given feed stream, 1102 m3/day (bub­ (bubble point calculation) assuming negligible pressure drop:
ble point feed).
K α
Desire: 98% C3 in the overhead (relative to the feed)
1% iC4 in the overhead C3 2.30 1.643
Air cooling (49°C Condensing Temperature) iC4 1.40 bubble point temperature = 121°C
nC4 1.15
iC5 0.68
Feed Composition Mol % Moles/hr
nC5 0.62
C2 2.07 21.5
C6 0.15
C3 48.67 505.6
iC4 10.11 105.0 αavg = 1.855
nC4 24.08 250.1 Determine the minimum number of trays (Equation 19-3):
iC5 5.41 56.2 94.9 19.3
 2.0   1.0 
SF = = 915.8
nC5 4.81 50.0
C6 4.85 50.4 log (915.8)
Sm = = 11 trays
100.00 1038.8 log (1.855)
Correct for change in relative volatility by using Equation 19-6:

Find the: KLK = 0.93 = βij (0.45)b (condenser)

• Minimum trays required KLK = 2.30 = βij (1.4)b (reboiler)

• Minimum reflux ratio dividing gives 2.473 = (3.111)b

• Actual trays at 1.3 times the minimum reflux ratio b = 0.798; βij = 1.759
495.4 99.8 0.798 516.7 1 – 0.798
  10.2   5.2   522.1  
log
Sm =
log (1.759)
log (512.1)
Sm = = 11.05
log (1.759)

19-7
Thus correcting for changing α did not affect the minimum tray valves. Valves of assorted weights have also been used to in­
calculation in this example. crease flexibility.
Find the minimum reflux, Rm (Equation 19-7, 8): The sieve or perforated tray is the simplest construction of
the three general types and thus is the least expensive option.
αavg αxF/(a – θ) The sieve tray is simply a plate with holes for vapor passage.
relative Although the sieve tray generally has lower pressure drop than
XF to C6+ θ = 16 θ = 15 θ = 15.8 other tray types, its main disadvantage is that sieve trays will be
C2 0.0207 68.33 0.0270 0.0265 0.0269 susceptible to “weeping” or “dumping” of the liquid through the
holes at low vapor rates and its turndown capacity is limited.
C3 0.4867 26.67 1.2165 1.1123 1.1941
Vendor-proprietary fixed valves and mini-valves are also
iC4 0.1011 13.83 –0.6443 –1.1951 –0.7098 available for increased capacity. Fixed valves can provide turn­
nC4 0.2408 11.00 –0.4506 –0.6622 –0.5518 down capacity between that of sieve trays and standard (mov­
able) valves. Larger fixed valves are used as alternatives to
iC5 0.0541 5.83 –0.0310 –0.0344 –0.0316
sieve trays in fouling services. Examples include
nC5 0.0481 5.00 –0.0219 –0.0241 –0.0223
Koch-Glitsch PROVALVE®, VG-10 (fixed valves)
C6 0.0485 1.00 –0.0032 –0.0035 –0.0033
VG-0 (fixed mini-valve)
Total 0.0925 –0.7805 –0.0978
MV-1 (movable mini-valve)
θ = 15.9
68.33 (0.041) 26.67 (0.949) 13.83 (0.01) Sulzer SVG (fixed valve)
Rm + 1 = + +
68.33 – 15.9 26.67 – 15.9 13.83 –15.9 MVG (fixed mini-valve)
Rm + 1 = 2.336 Trayed columns generally provide satisfactory operation
over a wide range of vapor and liquid loadings. Fig. 19-11 shows
Rm = 1.336
Theoretical trays at R = (1.3) Rm = 1.737
FIG. 19-8
R 1.737
Lo/V1 = = = 0.635 Top Two Trays of a Bubble-cap Column28
R+1 (1.737 + 1.0)
Rm 1.336
(Lo/V1)m = = = 0.572 Overhead Vapor Nozzle
Rm + 1 (1.336 + 1.0)
Sm/S = 0.54 (Fig. 19-7) Downcomer
S = 20.46 trays (use 21 trays)
Outlet
Weir

­TRAYED COLUMNS Inlet Weir

and
Seal Plate
­Internals
Various types of trays are used in fractionation columns.
Fig. 19-8 presents an isometric sketch of the top two trays in a Reflux
Nozzle
fractionation column with bubble caps. The bubble caps, along
with the weirs and downcomers, maintain a liquid level on the
trays. The liquid flows across the tray, into the downcomer, and
across the next tray in the opposite direction. The vapor flows Bubble
up through the caps and through the slots mixing with the liq­ Cap
uid. Assembly

Fig. 19-9 shows the vapor flow through bubble cap trays,
sieve trays, and valve trays. Due to the riser in the bubble cap,
it is the only tray which can be designed to prevent liquid from
“weeping” through the vapor passage. Sieve or valve trays con­
trol weeping by vapor velocity. The bubble cap tray has the Deflector Bubble
highest turndown ratio, with designs of 8:1 to 10:1 ratio being Insert Cap

common. Bubble cap trays are commonly found in existing gly­

col dehydration columns but new designs typically use struc­
tured packing.
Valve and sieve trays are popular due to the lower cost and Vapor
increased capacity over bubble cap trays for a given tower diam­ Riser
eter. Fig. 19-10 shows two valve designs. The upper drawing Hold-Down
Bolt Slots
shows a floating valve free to open and close with varying vapor Deflector
flow rates. The lower drawing shows a “caged” valve which pre­ Lug

vents valve loss due to erosion of the tray. Various other designs
are common such as using multiple disks and rectangular

19-8
operating characteristics for a representative system. The va­ rates, poor vapor-liquid contact can result. High liquid rates
por and liquid rates can vary independently over a broad range can cause flooding and dumping as the liquid capacity of the
and the column will operate satisfactorily. At low vapor rates downcomers is exceeded.
unsatisfactory tray dynamics may be characterized by vapor
pulsation, dumping of liquid, or uneven distribution. At high In order to handle higher liquid rates, more downcomer area
vapor rates, the tower will eventually flood as liquid is entrained is required. This is often achieved by using multiple pass trays.
to the tray above or backed-up in the downcomers. At low liquid Multipass trays increase liquid handling capacity for a given
diameter due to increased weir length and reductions in the
weir crest. Fig. 19-12 shows various configurations beyond a
FIG. 19-9 one pass tray where the liquid phase is split into two to four
Flow Through Vapor Passages28 flow paths to increase liquid handling capacity.

­Sizing
“C” Factor Method — Many design methods for sizing
trayed fractionators have been used. Generally these methods
are oriented toward liquid entrainment limitations or correla­
tions for flooding limits. A simple method called the Souders and
Brown equation8 involves using a Stokes’ Law type formula:

ρL – ρv
Vmax = (0.3048) C
√ ρv Eq 19-11
Note that ρL and ρv are at flowing temperature and pressure.
The value of C can be found from Fig. 19-13 based on tray spac­
ing and liquid surface tension. The column diameter is:

Vmax
DT =
√V max (0.7854) Eq 19-12
This method was originally developed for bubble cap trays and
(a) Vapor flow through bubble cap (b) Vapor flow through perforations
gives a conservative diameter, especially for other types of trays.
Nomograph Method — Manufacturers of valve trays have
developed design methods for their trays. Design procedures
are made available9, 10, 11 for preliminary studies. One such pro­
cedure starts with the nomograph in Fig. 19-14.10 This is a
simple relationship of liquid rate (GPM) and a quantity Vload
defined as:

ρv
√(ρ – ρ )
(c) Vapor flow through valves
Vload = CFS
L v Eq 19-13
FIG. 19-10
Valve Types28 FIG. 19-11
Limits of Satisfactory Tray Operation for a
Specific Set of Tray Fluid Properties8

19-9
 Simplified Hand Method — Tray vendors today provide
FIG. 19-12 computer programs to users to size both trayed and packed col­
Alternative Liquid Flow Paths umns. These vendors should be contacted for copies of their pro­
grams for their products. The method included here is a hand
method that can be used for preliminary sizing of trayed col­
umns and to understand the key parameters that affect column
sizing.
Fig. 19-14 is an approximation only and does not take into
account foaming which is a major consideration in many sys­
tems. In order to compensate for foaming, a System Factor is
used to adjust the vapor and liquid capacities (Fig. 19-15).
The downcomer velocity VDdsg *
is found from Fig. 19-16.
VDdsg
*
is corrected by the System Factor:
VDdsg = VD*dsg (System Factor) Eq 19-14
The other factor required for this design method is the vapor
capacity factor CAF.
CAF = CAFo (System Factor) Eq 19-15
CAFo is read from Fig. 19-17. In order to compute the column
cross sectional area, three quantities are needed.

FIG. 19-13
Souders-Brown Correlation for Approximate Tower Sizing8

19-10
 FIG. 19-14
Valve Tray Diameter10

19-11
The flow path length, FPL: From Fig. 19-17:
FPL = 9 DT/NP Eq 19-16 CAFo = 0.126 m/s
DT and NP are found from Fig. 19-14. CAF = (0.126) (0.85) = 0.107 m/s
The active area, AAM: 9 (2.29) from Equation 19-16
FPL = = 10.3 m
Vload + [(GPM (FPL/530)] 2
AAM =
CAF • FF Eq 19-17 0.1889 + [4.5 (10.3/530)]
AAM =
(0.107) (0.82)
FF, the flooding factor commonly used is 0.82 for most systems.
= 3.15 m2 from Equation 19-17
The downcomer area, ADM:
4.5
ADM = GPM/(VDdsg • FF) Eq 19-18 ADM = = 0.85 m2 from Equation 19-18
(6.46) (0.82)
If ADM is less than 11% of AAM, use 11% of AAM or double ATM = 3.15 + 2 (0.849) = 4.85 m2 from Equation 19-19
ADM, whichever is smaller.
0.1889
The tower cross sectional area is then: ATM = = 2.76 m2from Equation 19-20
(0.78) (0.107) (0.82)
ATM = AAM + 2 (ADM) Eq 19-19
4.85
or
ATM =
Vload
0.78 • CAF • FF Eq 19-20
DT =
√ 0.7854 = 2.48 mm = 2480 mm from Equation 19-21

A comparison of the methods (rounded to the nearest 50 mm)

The larger of these two values is used. Then:
C Factor 2550 mm
DT = √ ATM/0.7854 Eq 19-21 Nomograph 2300 mm (2900 mm for single pass)
Example 19-3 — Determine the diameter of a depropanizer Detailed Method* 2500 mm
with the following data:
vapor rate: 1994 m3/h
vapor density: 48 kg/m3 FIG. 19-15
liquid rate: 4.5m3/min System Factors9
liquid density: 461 kg/m3
liquid surface tension: 0.0033 η/m Systems with foaming tendencies are taken into account by using a
factor to derate the capacity of a given tray design. A list of the more
tray spacing: 61cm common foaming systems and their recommended factor is below.
“C” Factor Method
System Factor
From Fig. 19-13: C ≅ 430
Absorbers (over –18°C) 0.85
461 – 48
√
Absorbers (below –18°C) 0.80
vmax = 430 (0.3048) = 384.4 m/h
48 Amine Contactor 0.80
from Equation 19-11 Vacuum Towers 0.85

√
1944 Amine Stills (Amine Regenerator) 0.85
DT = = 2.54 m = 2540 mm
384.4 (0.7854) from Equation 19-12 H2S Stripper 0.85
Furfural Fractionator 0.85
Nomograph (Fig. 19-14) Glycol Contactors 0.50
Glycol Stills 0.65

√
1994 48 CO2 Absorber 0.80
Vload = = 0.1889 m3/s from Equation 19-13
3600 461 – 48 CO2 Regenerator 0.85
Caustic Wash 0.65
then from Fig. 19-14 @ GPM = 4.5 m3/s
Caustic Regenerator, Foul Water, Sour Water 0.60
DT ≅ 290 cm for a 1 pass tray = 2900 mm Stripper
229 cm for a 2 pass tray = 2290 mm Alcohol Synthesis Absorber 0.35
Hot Carbonate Contactor 0.85
Detailed Method Hot Carbonate Regenerator 0.90
VD*dsg = 7.6 m3/min/m2 at ρL – ρv = 413 Oil Reclaimer, 0.70
The capacity of a given tray design used in high pressure
2.93
System Factor = = 0.85 (Fig. 19-15) fractionation service with a vapor density of 28.8 kg/m3 and
413 0.32
 48 
higher should be derated by a system factor calculated by the
following formula:
VDdsg = 7.6 (0.85)
2.93
System factor =
= 6.46 m3/min/m2 (ρv)0.32

19-12
 FIG. 19-16
Downcomer Design Velocity10

FIG. 19-17
Approximate Flood Capacity of Valve Trays10

19-13
*At tray spacing different than 600 mm or for a different flood­ The system in Example 19-2 required 21 theoretical stages
ing factor, the diameter could change considerably. including the reboiler. The total actual trays is:

­Tray Efficiency 21 – 1
≅ 25 trays
0.80
All column design work is performed using theoretical trays.
An actual tray will not achieve equilibrium due to limitations of Typically an extra tray is added to the tray count for each
vapor-liquid contact time. In an actual column, more trays are feed tray and each side exchanger. Using this criteria, this col­
required to obtain the desired separation. This determination is umn should have 26 trays.
usually accomplished by the use of an overall tray efficiency
Typical operating pressures, tray counts, reflux ratio, and
defined as:
tray efficiencies for various gas processing systems are shown
theoretical trays Eq 19-22­ in Fig. 19-19. These are not design values; rather guidelines for
ε= actual trays typical values in previous applications. The actual selection de­
The determination of tray efficiencies from theoretical pa­ pends on many factors such as feed composition, energy cost,
rameters is the topic of numerous technical articles.12, 13, 14 A and capital cost.
detailed discussion of this subject is beyond the scope of this
book. ­HIGH CAPACITY TRAYS
O’Connell15 correlated the tray efficiencies of fractionators The 90s saw the proliferation of High Capacity Trays by dis­
and absorbers. For fractionators, this correlation considered tillation equipment vendors and users. These trays employ un­
thirty-eight systems of which 27 are hydrocarbon fractionators. conventional downcomer and deck configurations to effect va­
The correlation, shown in Fig. 19-18, relates overall tray effi­ por and/or liquid handling capability increases, when used to
ciency to relative volatility computed at average column condi­ revamp distillation columns. High Capacity Trays have been
tions and the feed viscosity at average column conditions. particularly effective in demethanizers, deethanizers, depro­
panizers and butane columns.
Example 19-4 — Evaluate the tray efficiency for the system in
Example 19-2. One such High Capacity Tray, the NYE TRAY®, is shown in
Average column temperature = 85°C Fig. 19-20. This tray increases vapor capacity by raising the
receiving pan and increasing the area available for vapor flow.
Feed viscosity @ 85°C = 0.076 mPa • s This, and similar trays, employ a crossflow arrangement with
Average α = 1.854 liquid traveling horizontally across the decks and vapor bub­
bling up through the liquid, creating a froth where the mass
Solution Steps transfer occurs. Proprietary downcomer designs and deck en­
(α)(µ) = 0.141 hancements such as fixed or mini-valves are often employed.
Examples of other High Capacity crossflow trays include the
From Fig. 19-18, ε ≅ 80% following:

FIG. 19-18
Effect of Relative Volatility and Viscosity on Plate Efficiency of Fractionating Columns

RELATIVE VOLATILITY OF KEY COMPONENT TIMES VISCOSITY OF FEED, mPa•s

(AT AVERAGE COLUMN CONDITIONS)

19-14
 FIG. 19-19
Typical Fractionator Parameters

Operating Number of­ Tray ­

Reflux1 Ratio Reflux2 Ratio
Pressure, kPa (ga) Actual Trays Efficiency, %
Demethanizer 1380 – 2750 18 – 26 Top Feed Top Feed 45 – 60
Deethanizer 2590 – 3100 25 – 35 0.9 – 2.0 0.6 – 1.0 60 – 75
Depropanizer 1650 – 1860 30 – 40 1.8 – 3.5 0.9 – 1.1 80 – 90
Debutanizer 480 – 620 25 – 35 1.2 – 1.5 0.8 – 0.9 85 – 95
Butane Splitter 550 – 690 60 – 80 6.0 – 14.0 3.0 – 3.5 90 – 100
Rich Oil Fractionator (Still) 900 – 1100 20 – 30 1.75 – 2.0 0.35 – 0.40 Top 67
Bottom 50
Rich Oil Deethanizer 1380 – 1725 40 – – Top 25 – 40
Bottom 40 – 60
Condensate Stabilizer 690 – 2750 16 – 24 Top Feed Top Feed 50 – 75
1
Reflux ratio relative to overhead product, mol/mol
2
Reflux ratio relative to feed, m3/m3

NYE Tray (Raschig-Jaeger) There are generally three types of packed columns:
SUPERFRAC ®
(Koch-Glitsch) • Random packing wherein discrete pieces of packing are
dumped in a random manner into a column shell. These
VGPlus (Sulzer ChemTech) packings are of a variety of designs. Each design has par­
Another style of High Capacity Trays employs multiple ticular surface area, pressure drop and efficiency charac­
downcomers to increase weir length and liquid handling capac­ teristics. Examples of various packing types are shown in
ity beyond what conventional multipass trays can provide. Ex­ Fig. 19-21. Random packing have gone through various
amples of such trays include the following: development phases from the first generation packings
which were two basic shapes, the Rashig Ring and the
MD™ (UOP) Berl Saddle. Second generation packings include the Pall
ECMD™ (UOP)
Hi-Fi (Shell Global Solutions) FIG. 19-20

Efficiencies of these counterflow trays are often lower than NYE TRAY Schematic
those of crossflow trays due to the reduced contact time between
the phases. Capacities can be quite high, and tray spacings
quite small, due to the very long outlet weir that these trays are
capable of providing.
Still another tray configuration, called “cocurrent flow,” is
expected to gain greater acceptance in the future. With cocur­
rent flow, the vapor and liquid phases are allowed to flow to­
gether, unidirectionally, for awhile inside contacting elements.
Examples of such trays are ULTRA-FRAC® (Koch-Glitsch),
Simulflow (UOP) and ConSep (Shell Global Solutions). Some
such trays are only capable of functioning at low liquid rates
such as those that are encountered in glycol dehydration col­
umns. Little information is publicly available regarding the ef­
ficiency of cocurrent flow trays. Sizing for these, and in fact
most High Capacity Trays, is regarded as proprietary by their
vendors. Fractionation Research Inc. (FRI) has performed ex­
tensive studies of both trays and packing in many configura­
tions. Companies that are members of Fractionation Research
Inc. (FRI) have access to this extensive set of test data and cor­
relations to predict capacity and efficiency.

­PACKED COLUMNS
Traditionally the majority of fractionation columns in gas
processing plants were equipped with trays. However, an op­
tion to trayed columns is to use packing. With packed columns,
contact between the vapor and liquid phases is achieved
throughout the column rather than at specific levels.

19-15
 Ring and the Inatalox Saddle which are still used exten­ • Grids which are systematically arranged packing which
sively today. Third generation packings come in a multi­ use an open lattice structure. These types of packings
tude of geometries most of which evolved from the Pall have found application in vacuum operation and low
Ring and The Intalox Saddle. (Fig. 19-22 and 19-23). Pop­ pressure drop applications. Little use of these types of
ular third generation packings include IMTP and Nutter packings are seen in high pressure services.
Rings. More recent developments include Raschig Super
Rings and Koch-Glitsch’s Intalox Ultra. Structured packing has found application in low liquid load­
ing applications which are below 49 m3/h/m2. Structured pack­
• Structured packing where a specific geometric configura­ ing has performed very well in extremely low liquid loading ap­
tion is achieved. These types of packing can either be the plications such as glycol dehydration (See Section 20). The high
knitted-type mesh packing or sectionalized beds made of surface tension in glycol dehydrators also helps the structured
corrugated sheets (Figs. 19-24a and 19-24b). There are a packing to perform well. Care is needed in the design to en­
number of commercially available packings which differ sure sufficient liquid to wet the entire packing surface.
in the angle of the crimps, the surface grooves and the
use of perforations. Above 49 m3/h/m2, random packings are more advantageous.
Structured packings have been tried in fractionators with little
success. Numerous case of structured packing failures have
been experienced in high pressure and/or high-liquid rate ser­
FIG. 19-21 vices. Predicting structured packing efficiency has been
problematic and unreliable in high pressure systems. Struc­
Various Types of Packing28
tured packings generally have lower pressure drop per theo­
retical stage than random packings. This can be important in
low pressure applications but not for high pressure NGL frac­
tionators.

­Column Sizing
The Eckert generalized pressure drop correlation (GPDC) is
often used for sizing randomly packed columns. The chart in
Fig. 19-26, which is a modified correlation, can be used to pre­
dict pressure drop for a given loading and column diameter.
Alternatively, for a given pressure drop the diameter can be
determined.
Most packed columns are designed for pressure drop of be­
tween 5 and 15 mm of water per foot of packed depth with
25 mm of water being the maximum.
The packing factors for various packings are shown in Fig.
­ 9-25. Broadly speaking, packings smaller than 25 mm size are
1
intended for towers 300 mm or smaller in diameter, packings 25
mm or 38 mm in size for towers over 300 mm to 900 mm in di­
ameters, and 50 or 75 mm inch packings are used for towers
three or more feet in diameter. This results from tradeoffs of
capacity and efficiency. The designer should select the proper

FIG. 19-23
Nutter Ring™ Metal Packings

FIG. 19-22
Flexipak ®, Cascade Mini Rings ® and Fleximax ® Packings

19-16
 FIG. 19-24a size of packing, and therefore the proper packing factor for cal­
culations.
Structured Packing
The packing factors in Fig. 19-25 are average values which
are sufficient for preliminary sizing but specific packing ven­
dors should be contacted for design applications.
The GPDC has limitations in describing the performance of
packings. Efforts to improve the correlation for specific packing
geometries have led to the development and publication of
charts for each packing which strive to correlate packing perfor­
mance information with the same abscissa and ordinate as the

FIG. 19-24b
Structured Packing

FIG. 19-25
Packing Factors (Fp) (Dumped Packing)

Nominal Packing Size (mm)

Packing Type Material
6 9 12 15 18 25 31 37 50 75 87
IMTP ®
Metal 51 40 24 18 12
Hy-Pak™ Metal 45 29 26 16
Super Intalox Saddles ®
Ceramic 60 30
Super Intalox Saddles® Plastic 40 28 18
Pall Rings Plastic 75 55 40 26 17
Pall rings Metal 70 56 40 27 18
Intalox Saddles ®
Ceramic 725 330 200 145 92 52 40 22
Raschig Rings Ceramic 1600 1000 580 380 255 155 125 95 65 37
Raschig Rings 1/32” Metal 700 390 300 170 155 115
Raschig Rings 1/16” Metal 410 300 220 144 110 93 62 32
Berl Saddles Ceramic 900 240 170 110 65 45
Flexiring Metal 49 23 16
Fleximax Metal 35 26 17 11
Cascade Mini Rings Metal 39 36 33 26 18
Cascade Mini Rings Plastic 44 33 20 17
NOTE: Values in this table are average values for the packing factor (Fp). Fp is actually a function of loading. Specific correlations for each
packing from the vendors should be used for design purposes.

19-17
GPDC chart. Kister31 published a series of 96 charts for a wide (0.024) (461 – 48) (48)
variety of packings. As discussed previously, FRI member com­ G2p = = 43.73
(0.000076)0.1 (26) (1000/461)0.1
panies have access to an extensive set of test data and cor­
relations to predict capacity and efficiency for many pack­ Gp = 6.61 kg/(m2 • s)
ings. 95 712
Ac = = 4.02 m2
Example 19-5 — Determine the packed column diameter for (6.61) (3600)
example 19-3 using 50 mm plastic Pall rings.
DT = 2.26 m
Given: µ = 7.6 (10–5) Pa • s
∆P = 42 mm water/m of packing ­Packing Height
Solution Steps In order to determine the height of a packed column bed, the
height of a theoretical plate, HETP, is required. HETP times the
ML = (4.5) (60) (461) = 124 470 kg/h number of theoretical stages gives the height of the packing.
MG = (1994) (48) = 95 712 kg/h Generally HETPs range from 300 to 900 mm but can be as high
as 1500 mm. Packed columns have found wide usage in cryo­
genic plant demethanizers. Typical HETP’s for demethanizers
√ √ √
Lp ρv ML ρv 124 470 48
= = = 0.42 are 91 cm for the upper section and 76 cm for the lower section.
Gp ρL MG ρL 95 712 461
The prediction of the HETP from theory or empirical rela­
From Fig. 19-26 at ∆P = 42 mm water/m of packing: tions is a complex subject.17 Recent research by Fractionation
Research Inc. has underscored the sensitivity of HETP with a
G2p µL0.1 Fp (ρw/ρL)0.1
= 0.024 number of variables. HETP is a function of flow rates and prop­
ρv (ρL – ρv) erties of the system as well as the specific geometric and me­
chanical factors. In order to determine packing requirements, a
From Fig. 19-25, Fp = 26, then: packing manufacturer should be consulted.

FIG. 19-26
Packed Column Pressure Drop Correlation

19-18
 Fig. 19-27 provides some example HETPs for hydrocarbon ­Dumped Packing Versus Trays
systems in the gas processing industry.18
Packed columns have been used extensively in the chemical
HETP’s are also a function of the packing size. In general, industry for many years. Packings are selected instead of trays
the smaller packings have lower HETP values. Fig. 19-28 shows for several reasons:
an example trend of packing HETP’s for one type of packing.
• Pressure drop — Packed towers usually yield a lower
­Packed Column Internals pressure drop per theoretical stage. This can be impor­
tant for low pressure operations. However, at the elevat­
A critical consideration in packed columns is the control of ed pressures encountered in natural gas processing, col­
the vapor and liquid phases. Fig. 19-29 shows a cross section of umn pressure drop is usually not a major issue.
a packed tower with various internals. Each section of packing
is supported by a support plate or grid whose function is to car­ • High liquid loading — for high liquid-to-vapor ratio sys­
ry the weight of the bed with minimum pressure drop. Hold- tems, a packed column will have more capacity for a giv­
down grids are used at the top to prevent lifting of the bed by en diameter. Some fractionation applications are charac­
the vapor phase. terized by low liquid/vapor ratios and packing has less of
an advantage for these designs.
Liquid distribution is a critical consideration in packed col­
umns. Poor liquid distribution causes dramatic loss of efficien­ • Foaming — in systems prone to foaming, packing is pre­
cy. Various designs have been used to distribute liquid feeds ferred over trays because packing does not promote foam.
and to collect and redistribute the liquid at various points in the Packing is less susceptible to foaming so credit may be tak­
tower. Generally the liquid should be redistributed every 6 m of en in the capacity correlation itself or by applying a less se­
packing height or every 10 column diameters, whichever is vere system factor (than trays in the same service) to the
smaller. Each of the packing vendors has several types of dis­ existing correlation. Designers should consult with their
tributors. The design and configuration of the distributor varies company fractionation specialist and/or the packing vendor
according to tower vapor and liquid loading. to confirm the appropriate system factor to use.

FIG. 19-27
Typical Packing Depths18

Packing Bed System ∆P, in mm

L/G, ­ Diam. HETP, HTU, %
System Size, Size, depth, press., ­ H2O/m
kg/(h • m2) cm Type m m Overhead
in. mm m kPa (abs) pkg
Absorber 350 / 470 91 Pall rings 2 50 7 0.85 – 5964 46 –
L.O.-Top fractionator 150 / 200 91 Pall rings 2 50 5 0.76 – 1082 10 –
L.O.-Bottom 450 / 240 122 Pall rings 2 50 5 0.85 – 1082 25 –
fractionator
Deethanizer top 470 / 190 46 Pall rings 11⁄2 37 6 0.88 – 2068 17 –
Deethanizer bottom 800 / 190 76 Pall rings 2 37 55 1.00 – 2068 25 –
Depropanizer top 240 / 175 59 Pall rings 11⁄2 37 4.9 0.98 – 1862 25 –
Depropanizer bottom 240 / 175 59 Pall rings 11⁄2 37 7.3 0.73 – 1862 25 –
Debutanizer top 80 / 130 50 Pall rings 11⁄2 37 3.7 0.73 – 621 10 –
Debutanizer bottom 80 / 130 50 Pall rings 11⁄2 37 5.5 0.61 – 621 10 –
Pentane-iso-pentane 90 / 80 46 Pall rings 1 25 2.7/2.3 0.46 – Atmos. 33 –

Light and heavy 30 / 50 38 Pall rings 1 25 3 0.61 0.62 100 mm. Hg 92 95.0
naphtha 13 / 25 38 Pall rings 1 25 3 0.99 0.76 100 mm. Hg 17 95.0
26 / 60 38 Pall rings 1 25 3 0.44 0.38 100 mm. Hg 146 97.5
10 / 27 38 Pall rings 1 25 3 0.44 0.40 100 mm. Hg 17 97.5
16 / 36 38 Intalox 1 25 3 0.70 0.60 100 mm. Hg 67 93.0
   9 / 21 38 Intalox 1 25 3 0.82 0.64 100 mm. Hg 18 99.0
14 / 34 38 Raschig rings 1 25 3 0.59 0.43 100 mm. Hg 92 91.6
   9 / 21 38 Raschig rings 1 25 3 0.82 0.60 100 mm. Hg 33 96.5

Iso-octane Toluene 83 / 97 38 Pall rings 1 25 3 0.41 0.41 Atmos. 58 82.0

40 / 90 38 Pall rings 1 25 3 0.58 0.66 Atmos. 8 76.0
90 / 110 38 Pall rings 1 25 3 0.24 0.31 Atmos. 142 84.0
40 / 50 38 Pall rings 1 25 3 0.47 0.43 Atmos. 12 74.0
46 / 55 38 Pall rings 1 25 3 0.41 0.39 100 mm. Hg 90 92.5
21 / 25 38 Pall rings 1 25 3 0.57 0.55 100 mm. Hg 17 87.0
43 / 55 38 Pall rings 1 25 3 0.51 0.49 100 mm. Hg 95 92.0
20 / 55 38 Pall rings 1 25 3 0.63 0.61 100 mm. Hg 17 89.0

Gas plant asborber 200 / 250 122 Pall rings 2 50 7 0.88 – 6205 9 92.00
propane
absorbed

19-19
 • Corrosion — for corrosive systems, packing can be fabri­ tions in each of these areas and a sound design must consider
cated from ceramics or plastics. Trays may have to be many details to ensure proper operation. For additional infor­
fabricated from expensive alloy materials. mation on heat exchangers see Section 9.
Packed columns also have several disadvantages which
must be taken into account in a fractionation design: ­Reboiler Arrangements19
• Liquid distribution — In trayed columns, the liquid phase There are several reboiler configurations which have been
is forced to flow across a tray surface. With gas bubbling used in fractionation service.20 The most common types are:
through the liquid, contact is assured. In packed towers,
• Forced circulation
the liquid and vapor are free to seek their own flow paths,
and channeling can occur. It is critical that the liquid • Once-through natural circulation
phase be properly distributed at the top of the column
and be redistributed at 6 m intervals or every 10 column • Vertical thermosyphon
diameters, whichever is smaller. • Horizontal thermosyphon
• Turndown — Turndown in a packed tower is usually limited by • Flooded bundle (kettle type)
the liquid distributor design. Typical distributor designs can
achieve 50% turndown, but special designs can provide a These types of reboilers are shown in Fig. 19-30 through
wide operating range, e.g. 10:1. Note that the high turndown Fig. 19-35. Modifications of these types are also used; for ex­
distributor often occupies more tower height than a standard
design. Turndown for a tray varies with tray type and design
details. Moveable valve trays typically can be operated at 20- FIG. 19-29
30% of full load, and special valve tray designs can operate at Example Packed Column Internals29
even lower turndown. This can be important in situations where
gas production is phased in and throughput rates build up Vapor Outlet
over time. to Condenser

• Plugging — Packed towers will be more susceptible to

plugging from dirt and other foreign materials.
• Packing Height — The HETP for a packed column is an Manway
uncertain matter for new services. Often they must be Reflu
Con x from
determined by testing or field applications. HETP’s can dens
er
vary from a few inches to several feet.
Liquid Distributor
Hold-Down Grid
• Inspection — It is difficult to inspect internals without
removing all the contents of a column. Structured Packing

­MECHANICAL CONSIDERATIONS
Support Grid
Liquid Collector
Once the diameter and height of a fractionator have been
determined, consideration must be given to the column inter­ Ringed Channel
nals and heat exchanger arrangements. There are many op­ eed Manway
id F
Liqu
Liquid Distributor/
Redistributor
Rings or
Hold-Down Grid Saddles
FIG. 19-28 Random Packing
Example Effect of Packing Size on HETP
Support Plate
Manway

eed
or F
Vap

Liquid Distributor

Structured Grid

turn
ile r Re
R ebo
Circulation Pipe
Skirt to Reboiler

Bottom
Product

19-20
ample, forced circulation reboilers are not necessarily in verti­ there is no opportunity for recirculation in the arrangement in
cal orientation. Also, internal “stab-in” type reboilers have been Fig. 19-31(a), it is called a once-through reboiler arrangement.
used but are not common. Fig. 19-31(b) is referred to as a recirculating reboiler.
Each type of reboiler has its special advantages and disad­ Vertical Thermosyphon — The vertical thermosyphon
vantages. Selection criteria of a reboiler configuration should reboiler shown in Fig. 19-32 is usually a one tube pass-one shell
include: pass exchanger with the channel end up. The upper tube sheet
is placed close to the liquid level in the bottom of the fraction­
• Heat transfer surface required ation column. This type exchanger is capable of high heat trans­
• Space and piping requirements fer rates (minimum area) and requires simple piping. It is not
easily fouled and has generally good controllability. Because of
• Ease of maintenance the vertical orientation, additional column skirt is required and
• Fouling tendency maintenance can be awkward.

• Operating stability Recirculation Ratios — The recirculation ratio is deter­

mined from the difference between hydrostatic head in the col­
• Operating cost umn corresponding to the tube length of the reboiler and the
weight of the vapor-liquid mixture.
• Column and skirt elevation requirements
Recirculation ratios of 4:1 or greater are usually employed.
Forced Circulation — A typical arrangement for forced Referring to the vertical thermosyphon in Fig. 19-32, there are
circulation reboilers is shown in Fig. 19-30. This type is also five principal resistances:
called a pumped-through reboiler. All the liquid from the bot­
tom tray is carried by a downcomer to below the liquid level in • Frictional pressure drop through the inlet piping
the bottom of the column. The liquid can be circulated through
the reboiler as many times as is economically feasible to control • Frictional pressure drop through the reboiler
the percent vaporization. The bottom product is drawn off the • Expansion or acceleration loss due to vaporization in the
line to the reboiler. This type arrangement is usually used in reboiler
installations where piping pressure drop is so high that natural
circulation is not practical. Remotely located reboilers or instal­ • Static pressure of a column of mixed liquid and vapor (Z3)
lations where the reboiler heat is provided by several sources in the reboiler
may require forced circulation. The main advantages of forced • Frictional pressure drop through the outlet piping
circulation are the abilities to closely control circulation rate
and to handle viscous or solid containing fluids. The continuous Detailed calculation of the hydraulics of this system is com­
operating cost of the pump makes natural circulation designs plex due to the two-phase flow involved.20 The simplified meth­
more desirable. od presented by Kern19 can be used to estimate this system:
Natural Circulation — By far the greater number of re­ Expansion Loss Due to Vaporization — This is taken
boiler installations employ natural circulation. This can be as two velocity heads based on the mean of the inlet and outlet
achieved in either of two ways as shown on Fig. 19-31(a and b). densities.
In Fig. 19-31(a) all the liquid on the bottom tray is circulated
directly to the reboiler, where it is partially vaporized. The un­ G2
∆P1 =
vaporized portion, on being disengaged under the bottom tray, 144 gc ρavg Eq 19-23
is withdrawn as bottom product. In Fig. 19-31(b) the liquid
passes through the downcomer below the liquid level of the col­ Particularly where the recirculation ratio and the operating
umn as in forced circulation. The bottom liquid is free to recir­ pressure are great, the difference in the densities between the
culate through the reboiler as many times as the hydrostatic inlet and outlet is not very large and the expansion loss is
pressure difference between Z1 and Z3 will permit. Because negligible.

FIG. 19-30 FIG. 19-31

Forced-Circulation Reboiler Arrangement Natural-Circulation Reboiler Arrangements
Vapor + Liquid

Vapor + Liquid

Vapors Vapors

Z3 Z3
Z1 Liquid Liquid Steam
Steam
Z1
Z2 Z2

Liquid
Bottoms Bottoms

(a) One-through Reboiler (b) Recirculating Reboiler

19-21
 Weight of a Column of Mixed Liquid and Vapor — This Example 19-6 — A vertical thermosyphon reboiler is to provide
is difficult to evaluate if precision is required, since the expan­ 18 507 kg/hr of vapor which is almost pure butane. The column
sion of the vapor is a function of the recirculation ratio, average operates at a pressure of 1896 kPa(ga) corresponding to a near­
specific volume of the vapor, coefficient of expansion of the liq­ ly isothermal boiling point of 109°C. Heat will be supplied by
uid, etc. For nearly all practical cases it may be assumed that saturated steam at 862 kPa(ga). A recirculation ratio of 4:1 or
the variation of the density is linear between the inlet and the greater should be employed. What is the optimum exchanger to
outlet. If v is the specific volume at any height, h, in the vertical fulfill this requirement? (Assume negligible inlet and outlet
tube whose total length is Lt and whose inlet and outlet specific piping pressure drops.)
volumes are vi and vo:
Given:
(vo – vi) h
v = vi + Tube Data:
Lt Eq 19-24
2 cm, 16 BWG, 2.5 cm triangular pitch
If the weight of the column of mixture is m, the change in weight
with height is dm, and if a is the cross-section flow area, I.D. = 1.57 cm
a Surface Area = 0.0598 m2/m
dm = dh Eq 19-25
v Internal Tube Area = 1.948 cm2
If the static pressure of the column of liquid and vapor is desig­ Vapor Density = 36.36 kg/m3
nated by Z3ρavg and the cross-section area, a, is unity, then:
Lt Liquid Viscosity = 0.1 Pa sec
  vi + (vo – vi) x/Lt
dx dx
Z3ρavg = = Eq 19-26
v Liquid Relative Density = 0.43
Integrating and dividing by 100 to obtain the static head per Solution Steps
square decimeter:
Heat balance:
Z3 ρavg 2.3 Lt vo
= log Eq 19-27 Enthalpy of liquid at 109°C and 2000 kPa(abs)
100 100 (vo – vi) vi
= 561 kJ/kg
Rational solutions for the recirculation ratio can be established
Enthalpy of vapor at 109°C and 2000 kPa(abs)
by taking all the heads in the circuit into account as functions
= 786 kJ/kg
of the mass velocity, G, and upon solution for G, the recircula­
tion rate can be obtained directly. Because the gravity of the Q = 18 507 (786 • 561) = 4.16(106) kJ/kg
reboiler outlet mixture also varies with the recirculation ratio,
the expression becomes complex and it is simpler to solve by Steam = 4.160(10)6/2019 = 2060 kg/h
trial and error. If the height of an existing reboiler is given, the Isothermal boiling, ∆T = LMTD = 178 - 109 = 69°C
recirculation ratio can be computed. If the recirculation ratio is
given, the required head, Z1ρL, may be computed. Trial 1:
When establishing reboiler surface the first trial should always
be taken for the maximum allowable flux (136 000 kJ/m2 • h):
Q 4.16 (106)
FIG. 19-32 A = = = 30.59 m2
Q/At 136 000
Vertical Thermosyphon Reboiler Connected to Tower

FIG. 19-33
Thermosyphon Reboiler Driving Force Curve

19-22
 Assume 4877 mm long tubes to reduce the shell diameter The resistances are greater than the hydrostatic head can
and provide the cheapest reboiler. However, the long tubes will provide; hence the recirculation ratio will be less than 4:1. Of
also require the greatest elevation of the column. the resistances, the frictional pressure drop may be reduced by
the square of the mass velocity if the tubes are made shorter.
30.59 (1000) The other alternative is to raise the liquid level in the column
Number of tubes = = 105
(4877) (0.0598) above the upper tube sheet.
Static pressure of reboiler leg (Equation 19-27): Trial 2:
Vapor density, Assume 3658 mm tubes and 4:1 recirculation ratio:
ρv = 36.36 kg/m 3
30.59 (1000)
Number of tubes = = 140
(3658) (0.0598)
1 m3
vv = = .0275
36.36 kg vi = 0.00233 as before
1 m3 vo = 0.00736
v6 = vi = = 0.00233
429.8 kg Static pressure,
Weight flow of recirculated liquid = 4 (18 507) = 74 028 kg/h Z3 ρavg
= 8.19 kPa
Total volume out of reboiler: 100
Liquid (74 028) (0.00233) = 172.5 m3 Frictional resistance:
Vapor (18 507) (0.0275) = 508.9 m3 at = 140 (1.948) = 272.72 cm2
Total = 681.4 m3
74 028 + 18 507 kg
681.4 m3 Gt = = 339.3
vo = = 0.00736 272.72 (h • cm2)
(18 507 + 74 028) kg
kg
Static pressure of leg, (Equation 19-27) = 3.39 (106)
(h • m2)
Z3 ρavg 2.3 (4.88) 0.00736 (0.0157) (3.39) (106)
= log Re = = 147 842
100 (0.00736 –0.00233) 0.00233 0.360
= 10.92 kPa f = 0.000135 (144) = 0.01944 m2/m2

Frictional resistance: (0.01944) [3.39 (106)]2 3.66

∆Pt = = 7.10 kPa
(2.592) (1010) (0.0157) (282.83)
Flow area:
Total resistance = 8.19 + 7.10 = 15.29 kPa
a′t 1.948 Z1 ρL (9.807)
at = Nt = (105) = 0.02045 m2 Driving force, = (3.66) (429.8) = 15.43 kPa
10 000 10 000 100 1000
74 028 + 18 507 Since the driving force is slightly greater than the resistances,
Gt = = 4.525 (106) kg/(h • m2)
0.02045 a recirculation ratio better than 4:1 is assured.
At 109°C, µ = (0.10) 3.60 = 0.360 kg/m • h) If the solution is not reached, a plot of ΔP (driving force – resis­
1.57 tance) can be developed to determine the maximum length to
D′ = = 0.0157 m achieve a pressure balance where the driving force exceeds the
100 resistance. Fig. 19-33 shows a plot for this example at various
D′ Gt (0.0157) (4.525) (106) lengths.
Re = = = 197 340
µ 0.360 Horizontal Thermosyphon — The horizontal thermosy­
f = 0.000127 (144) = 0.01829 m /m 2 2 phon (Fig. 19-34) is similar in nature to the vertical thermosy­
phon reboiler. However, less skirt height is required to provide
1 circulation head. It consists of centrally located inlet and outlet
429.8 + kg nozzles, a vertical support plate between the nozzles, and a
0.00736 = 282.83
ρavg = m3 horizontal longitudinal baffle. The liquid enters the bottom,
2
flowing in two directions around the longitudinal baffle and
f G2t L back together at the outlet. Vapor disengagement takes place in
∆Pt = the column. The horizontal thermosyphon can be natural or
(2.592) (1010) (D) (ρavg)
forced circulation.
(0.01829) [4.525 (106)]2 (4.88)
= = 15.88 kPa Kettle Reboilers — Kettle reboilers are quite different in
(2.592) (1010) (0.0157) (282.83)
their configuration than other types (Fig. 19-35). Liquid from
Total resistance = 10.92 + 15.88 = 26.80 kPa the bottom tray is gravity fed to the reboiler from a draw-off on
the bottom of the column. A weir maintains the liquid level in
Z1ρL (4.88)(429.8)(9.807) the reboiler such that the tube bundle is always submerged.
Driving force, = = 20.6 kPa
100 1000 Vapor disengaging space is provided in the exchanger. The va­
por is piped back to the column to provide stripping vapor for
the bottom tray. Bottom product is drawn from the reboiler.

19-23
 Kettle reboilers are attractive due to the ease of control. No Bottom Vapor Inlet22 — The optimum vapor inlet below
two-phase flow or circulation rate considerations are required. the bottom tray is shown at location A in Fig. 19-40. The vapor
The kettle is also equivalent to a theoretical tray. Due to the is introduced parallel to the bottom downcomer at a recom­
vapor disengagement requirement, kettles are constructed with mended spacing of 61 cm below the bottom tray. A vapor inlet
an expanded shell. The additional cost of this shell is offset to nozzle, causing impingement of the vapor stream against the
some extent by a reduced tower skirt requirement. downcomer and/or liquid overflow as shown by location B,
should be provided with vapor inlet baffle or piping. The vapor
­Column Internals velocity can be controlled by the hood outlet area. For multipass
trays, it is very important to feed each compartment equally
The most common causes of startup and operating problems
and allow for vapor equalization between sections.
are the column internals. These items are usually small details
that are often overlooked and later become operating difficul­ Liquid Outlet — Sufficient residence time must be pro­
ties. Correct location and orientation of inlet and outlet nozzles vided in the liquid draw-off sump. Fig. 19-41 presents recom­
and other internal considerations must be addressed to elimi­ mended residence times for various situations. These guidelines
nate problems. Kister21 presented a series of articles which dis­ are intended to provide sufficient times for vapor disengage­
cussed these areas in detail. ment, to smooth out column upsets, and to give operating per­
sonnel time to correct operating problems. For large residence
Top Feed Nozzles — Fig. 19-36 illustrates various ar­
time requirements, an external vessel should be considered in
rangements used for top-tray feed and reflux nozzles. Fig. ­19-37
lieu of a large sump volume.
lists factors and restrictions in each design. For cost reasons,
arrangements of Fig. 19-36a, b, c, e, and f are preferred. How­ Bottom Sump Arrangements — A common design prac­
ever, for a two-phase stream only b, d, e, and h are suitable. tice is to divide the bottoms sump into a reboiler-feed compart­
Most installations use arrangement a or c for all-liquid feed ment and bottoms-drawoff compartment by installing a prefer­
while b and e are popular for a two-phase feed. ential baffle. Typical arrangements are shown in Fig. 19-42.
The baffle has the advantage of providing an additional theo­
Tray sections and baffles that are contacted by an entering
retical tray, supplying a constant head to the reboiler, and in­
feed should be strengthened. Feed nozzles and internal liquid
creasing the bottoms-outlet sump residence time. The
distributors should be anchored to the tower shell. Feed lines
­installation of such a baffle is recommended when thermosy­
containing two phase flow should be designed to minimize slug­
phon reboilers are used.
ging which causes column instability and possible tray damage.
For liquid feeds, the nozzle velocity should not exceed 0.9 m/s. Each sump must have its own drainage facilities. This can
frequently be achieved by drilling a hole through the baffle, or
Intermediate Feed Nozzles — Fig. 19-38 shows various
by using an external dump line at a low point to interconnect
methods for introducing intermediate column feeds. Fig. 19-39
the liquid outlet lines from each compartment.
summarizes the application area for each design. Fig. 19-38a is
only suitable for subcooled liquids. Vapor containing or hot Either one of the arrangements shown in Fig. 19-42a or b is
feeds would cause flashing in the downcomer and loss of capac­ satisfactory. The arrangement of Fig. 19-42b has slightly better
ity. Fig. 19-38b is only suitable for low-velocity liquid feeds and mass-transfer characteristics; however, it is somewhat more
is seldom recommended. Fig. 19-38c and d show a similar noz­ complicated than that of Fig. 19-42a. A baffle similar to that on
zle location with a baffle to direct the feed stream. These are the left-hand side of Fig. 19-42b can also be incorporated in ar­
both designed for two-phase streams with d being the preferred rangements such as shown in Fig. 19-42c and d.
arrangement. Fig. 19-38c can be used for virtually any feed ex­
cept for high velocity feeds where a baffle plate is added as in The arrangement of Fig. 19-42d is preferable to that of Fig.
Fig. 19-38f. 19-42c for two-pass trays. The latter forces the vapor to flow
through a curtain of liquid while ascending to the first tray,
which may cause entrainment or premature flooding.

FIG. 19-34
Horizontal Thermosyphon Reboiler FIG. 19-35
Kettle Reboiler Arrangement

19-24
 FIG. 19-36
Example Top Feed Nozzles21

FIG. 19-37
Design Parameters for Top Feed Nozzles21

Dimensions for top feed/reflux inlet arrangements

See Fig. 19-36
a b c d e f g h
Maximum nozzle dia., mm 150 – – 150 – – 150 –
Note 3 3 – – 1 1 – 2
Pure-liquid feed
Dimension x, mm Wd hcl dn/2 >300 2dn dn/2 100 >300
Dimension y, mm 100 to 150 2dn – Wd 2dn 2dn Wd Wd
Dimension z, mm – dn – 100 to 150 dn 1.5dn – 100 to 150
Vapor/liquid feed NS NS NS NS
Dimension x, mm 2dn >300 2dn >300
Dimension y, mm 2dn Wd 2dn Wd
Dimension z, mm dn 100 to 150 2dn 100 to 150
dn = Inlet pipe dia.,mm
hcl = Clearance under downcomer, mm
Wd = Downcomer width, mm
NS = Not suitable
Note 1: Drill a 6 mm vent hole on top.
Note 2: Wear plate may be required.
Note 3: Ensure nozzle enters behind the baffle. If it does not hydraulic jump could be a problem.
Internal inlet pipes should be removable for maintenance.

19-25
 Draw-off Arrangements — Total draw-off is normally ac­ most of the liquid on the tray is degasified. The degasified liquid
complished with a chimney tray or draw pan as indicated in on the tray produces a greater hydrostatic head than the col­
Fig. 19-43. The chimney tray has an advantage over the draw umn of aerated liquid in the downcomer. This effect is aggra­
pan because it catches tray weepage during startup and at low vated if two phases are separated. If these effects are not al­
vapor rates. Chimneys are normally sized for approximately lowed for, and sufficient height is not provided, downcomer
15% of tower area. The chimneys should be located or hooded to backup may exceed the spacing between the liquid level and the
prevent liquid flow downward through the chimney. Elevating tray above, and lead to premature flooding.
the draw nozzles flush with the draw tray in many cases elimi­
nates the need for weep holes. A spill-over baffle can be provid­ Fig. 19-44 shows two types of partial draw-off arrange­
ed for the draw pan to maintain tower circulation for cases ments. When a chimney tray is used, a partition (sometimes
where a draw-off may not be required during operation. A vor­ insulated by application of two plates) can be provided to allow
tex breaker is suggested for outlet nozzles. a draw-off and return on the same tray. Elevating the partition
will determine total separation or recycling. The return nozzle
If the liquid on the chimney tray seals the downcomer from should be located above the liquid level if vapor content is ex­
the tray above, particular care must be taken with the design of pected.
this downcomer. The liquid in the downcomer is aerated, while
Partial draw from a recessed pan is frequently used. The
draw pan saves shell length at the sacrifice of surge capacity. It
is advisable to provide a positive downcomer seal.
FIG. 19-38
Water draw-off has been successfully accomplished by using
Example Intermediate Feed Nozzle Arrangements21 the design shown in Fig. 19-45. The perforated plate normally
contains 25% of the pan area as hole area. Water draw pans are
usually sloped for multipass trays in large towers. A weld-in
pan with a flush fitting draw nozzle is recommended.

­Mechanical Design
Special care should be given to designing the trussing struc­
ture at heavily loaded areas, such as draw pan and draw trays
where additional liquid levels are anticipated.
Where total draw-off arrangements are required, it is gener­
ally recommended that seal welding should be applied in lieu of
gasketing, as gasketing may not maintain its sealing effective­
ness at operating conditions. For large towers and higher tem­
peratures, expansion joints should be provided. Good inspection
can, in many cases, detect errors which could lead to column
operation problems. It is, therefore, important that an inspector
be well informed on internal design fundamentals including
feed and draw-off arrangements. The following recommended
inspection check list can be an invaluable tool for discovering
errors and providing a record for future reference:
• Weir levelness
• Tray levelness
• Is hardware properly installed?
• Have correct materials been used (general spot check)?

FIG. 19-39
Intermediate Feed Nozzle Applications21

See Fig. 19-38

a b c d e f
Cold-liquid feed Yes Yes Yes Yes Yes Yes
Vapor/liquid feed No No Yes Yes Yes Yes
Vapor feed No No Yes Yes Yes Yes
Hot feed No Yes Yes Yes Yes* Yes*
High-velocity feed No No No Yes No Yes
High-pressure application No Yes Yes Yes Yes Yes
Downcomer capacity critical No Yes Yes Yes Yes Yes
*Assuming insulation plate is provided.

19-26
 FIG. 19-40 FIG. 19-41
Bottom Vapor Inlet22 Residence Time for Liquid in the Sump21

Minimum
Operating condition residence
time, min
Liquid is withdrawn by level control and feeds 2
another column directly by pressure.
Liquid is withdrawn by level control and pumped 3
away. Spare pump starts manually.
Liquid is withdrawn by level control and pumped 1
away. Spare pump starts automatically.
Liquid is withdrawn by level control and feeds a unit 5-7
that is some distance away or that has its instru­
ments on a different control board.
Liquid is withdrawn by flow control. 3-5
Liquid flows through a thermosyphon reboiler 1
without a level controller, to maintain a level in the
sump.

FIG. 19-42
Example Baffle Arranements for Bottom Sumps for Recirculating Reboilers21

Reboiler Reboiler
Return Return Reboiler
Return
Auxiliary
Baffle

Preferential Reboiler
Baffle Return

To Bottom To Bottom To Bottom To Bottom

Reboiler Product Reboiler Product Reboiler Product Reboiler Product
a. b. c. d.

FIG. 19-44
FIG. 19-43 Example Partial Draw-off22
Example Total Draw-off22

19-27
 FIG. 19-45 FIG. 19-46
22 Example Feed/Product Exchanger
Example Water Draw-off

Weld-In Draw
Pan Perforated
Plate

Flush

• Are downcomers properly sealed to prevent vapor from

escaping into downcomer area?
Bottom
Product

• Are nozzles oriented properly with respect to feed and

draw-off requirements? Feed

• All special internal designs as well as feed and draw noz­

zles should be checked for restrictions
FIG. 19-47
• Are trays clean and clear of obstruction from foreign ma­ Example Side Heater
terial?
• Are all internal parts secured?
Side Heater
­ENERGY EFFICIENT DESIGN­ Draw off
Tray
CONSIDERATIONS
Fractionators require energy input in the form of heat to the
reboiler. Regardless of the exact arrangement, fuel is often re­ Secondary
Heat Source
quired and represents a major operating cost. Minimization of
fuel usage is a common design goal.
To provide reflux for the fractionator, a utility is required to
remove the heat to an appropriate heat sink. For columns uti­ Reboiler
lizing air or water cooling, all the utilities use a common tem­
perature heat sink. However, for columns using refrigeration,
the temperature level is very important. A lower temperature Primary Heat
refrigeration level increases both the capital and operating cost Medium
of a unit. If the condenser duty can be applied to a higher tem­
perature system, considerable savings can be realized.

­Feed/Product Exchangers
To Next Process Unit
One of the simplest ways to reduce the reboiler fuel require­ or to Side Heater
ment is to preheat a liquid feed stream. This can be accom­
plished with a feed/product exchanger as shown in Fig. 19-46. gradient in the column, this heat is applied at a much lower
In general this heat input will decrease the reboiler duty. temperature than the reboiler. The heat source for this side
However, since the feed is now partially vaporized, the over­ heater can be any stream which requires cooling and is at a
head condenser duty will tend to increase (see Equation 19-9). high enough temperature level to be useful. Often, the bottom
This increased condenser duty must be offset by reboiler duty. product is used to side-heat the column. In cryogenic plants, the
The net reboiler savings will be close to, but not equal to, the feed gas often supplies the reboil heat.
heat input to the feed. The net effect will depend on many sys­ One penalty for side heating is the additional column height
tem parameters; but feed/product exchange is generally an at­ required for the liquid draw-off tray and vapor disengagement
tractive heat conservation application. of the two-phase return stream. Typically, this will add about
1.8-2.4 m to the column height.
­Side Heaters
For small amounts of side heat up to 25% of the reboiler
Side heaters (Fig. 19-47) can be used to add heat to a tower duty, the side heater has little effect on the column design and
several trays up from the reboiler. Because of the temperature condenser duty. As this heat is increased, the condenser duty

19-28
will rise, requiring more total heat and/or more trays. In gen­ absorber to remove the heat of absorption from the liquid phase.
eral, a good rule of thumb is to limit the side heat duty to no This helps limit the temperature rise throughout the tower
more than 50% of the total heat requirement. thus increasing absorption recovery.
One other possible benefit of side heating is that tower load­ ­Heat Pumping
ing below the side heat tray is reduced. In many towers the
bottom trays have the greatest loading. Judicious application of One technique for energy conservation in fractionation sys­
side heating can “smooth” the column vapor and liquid profiles tems is the use of a heat pump. Heat pumping usually employs
in the lower section, reducing the required diameter. an external working fluid as shown in Fig. 19-48. Compression
is used to raise the temperature of the working fluid above that
­Side Coolers/Condensers required for the reboiler. The fluid leaving the reboiler is then
flashed and used to condense the reflux. The net result is that
Side coolers can be used to extract heat from a tower at an the heat absorbed in the condenser is used to reboil the column.
intermediate point above the feed tray. Due to the tower gradi­ The main operating cost then becomes the compressor rather
ent this heat removal can be accomplished with a higher tem­ than the normal heating and cooling utilities.
perature cooling medium. This can be particularly attractive if
the condenser uses refrigeration. An alternative to the basic heat pump is to use the column
overhead as the working fluid. This alternative, vapor compres­
For large quantities of heat removal a side condenser can be sion, eliminates the overhead condenser (Fig. 19-49).
effective. However, this is a bit more complicated than a single
phase cooler. A side condenser requires a vapor draw-off and It is often difficult to find a working fluid to reboil and con­
provision for a two phase return and separation section. dense in a single fractionator. However, often plants have sev­
eral fractionators with condensers and reboilers at a variety of
One place where side coolers are used is in absorber applica­ temperatures. It may be possible to link a condenser and reboiler
tions. The cooler draw-off is located in the middle part of the from separate columns to utilize a heat pump configuration.

FIG. 19-48 FIG. 19-49

Heat Pumping Vapor Recompression

Feed Overhead
Feed
Overhead
Column
Column

Compressor
Compressor
Bottoms
Bottoms

19-29
 Absorption
Absorption is one of the oldest unit operations used in the lowers the partial pressure of the light hydrocarbons which is
gas processing industry. Rich gas enters the bottom of the ab­ equivalent to lowering the effective operating pressure.
sorber and flows upward contacting the countercurrent lean oil
stream. The lean oil preferentially absorbs the heavier compo­ Refrigerated lean oil plants normally use direct fired heat­
nents from the gas and is then termed “rich oil”. The rich oil is ers to vaporize a portion of the rich oil in the stripper (still) to
sent to a stripper (or still) where the absorbed components are provide the necessary stripping vapor.
removed by heating and/or stripping with steam. The lean oil is
recycled to the absorber to complete the process loop. ­ABSORBER CALCULATIONS
For a given gas, the fraction of each component in the gas Absorber and stripper calculations, like fractionation col­
that is absorbed by the oil is a function of the equilibrium phase umn calculations, can be accomplished with tray-by-tray com­
relationship of the components and lean oil, the relative flow puter models. However, hand calculations can be performed to
rates, and the contact stages. The phase relation is a function of estimate the absorption of components in a lean oil absorber.
pressure, temperature, and the composition of the lean oil. No­ The stripping operation is essentially the reverse of absorption
menclature for an absorber is shown in Fig. 19-50. and can be handled in a similar fashion.
As components are absorbed, the temperature of the gas and Many attempts have been made to define an “average” ab­
oil phases will increase due to heat of absorption. The heat re­ sorption factor method to short-cut the time consuming rigor­
leased is proportional to the amount of gas absorbed. In many ous calculation procedures. The sole restriction of such a meth­
cases, side coolers are used on the absorber to limit the tem­ od is how well the average factor, as it is defined, will represent
perature rise and aid in absorption. the absorption that actually occurs. One of the simplest defini­
tions of an average absorption factor was by Kremser and
Lean oil will have a molecular weight in the 100 to 200 range. Brown.23, 24 They defined it as:
For ambient temperature absorbers, a heavy lean oil of 180 to
200 molecular weight will normally be used. For refrigerated ab­ A = Lo/(Kavg Vn+1) Eq 19-28
sorbers, a lighter lean oil of 120 to 140 molecular weight is used.
or Lo = A (Kavg) (Vn+1) Eq 19-29
A lower molecular weight lean oil will contain more moles per
gallon resulting in a lower circulation rate. However, lower mo­ Using an average absorption factor, the extraction of any com­
lecular weight lean oil will have higher vaporization losses. ponent from a rich gas can be described by:
The stripping column is operated at low pressure and high Yn + 1 – Y1 An + 1 – A
temperatures. In older plants, “live” steam is injected into the = = Ea Eq 19-30
Yn + 1 – Yo An + 1 – 1
bottom of the column to strip the NGL components. The steam
Fig. 19-51 provides a graphical solution of Equation 19-30.
Example 19-7 — Oil absorption is to be used to recover 75 per­
FIG. 19-50 cent of the propane from 100 moles of the rich gas stream shown
below. The absorber is to have six theoretical plates. What oil
Absorption Nomenclature circulation rate is to be used if the average temperature and
pressure of the absorber are 40°C and 6895 kPa? The entering
lean oil is assumed to be completely stripped or denuded of rich
Vn+1Y1 gas components. What will be the composition of the residue
Lean Gas gas leaving the absorber?

L0Xo ­Solution Steps

Lean Oil Using the equilibrium ratio charts (Section 25), obtain the
K-value for each component at 40°C and 6895 kPa.
From Fig. 19-51 at Ea = 0.75, A = 0.80
Using Equation 19-29:
Lo = (0.8) (0.37) (100) = 29.6 moles/hr
(based on 100 moles of gas)
Using Equation 19-28, the oil rate calculated and the com­
ponent K-values determine the absorption factor “A” for the re­
maining components.
For example, for methane:
Vn+1Yn+1 29.6
Rich Gas A= = 0.091
3.25 (100)
Using the absorption-factor values read values of Ea for each
L0Xn component (Fig. 19-51).
Rich Oil

19-30
 FIG. 19-51
Absorption and Stripping Factor Correlation

19-31
 Solve Equation. 19-30 for each component to determine the ­SOUR WATER STRIPPERS
moles of components in the residue gas, Y1.
Sour water is a term used for water containing dissolved
For example, for methane: hydrogen sulfide. Facilities for processing sour gas may have
Yn + 1 – Y2 90.6 – A1 several sources of sour water. These include water from inlet
= = 0.091 separators, water from compressor discharge scrubbers, quench
Yn + 1 – Yo 90.6 – 0 water from certain Claus unit tail-gas cleanup processes, and
Y1 = 82.36 water from the regeneration of solid bed product treaters or
dehydrators. In some plants it is possible to dispose of this wa­
Note: For this example, Yo = 0 since entering lean oil is as­ ter by using it for makeup to the gas treating solution. Howev­
sumed completely stripped of rich gas components. This as­ er, most sour gas plants have an excess of water and the hydro­
sumption will not be true for all cases. gen sulfide must be removed to a level of 1 to 2 ppmw before
Calculate the moles of each component in the rich oil. For disposing of the water. Sour water strippers are used for this
example, for methane: purpose.

l = Yn + 1 – Y1 + Yo = 90.6 – 82.36 + 0 = 8.24 Sour water strippers commonly have 10 to 15 trays or 6 to 9

m of packing. The feed enters at the top and heat is ­supplied
Comp Mol % K A Ea Y1 l either by a reboiler or by steam injection directly below the bot­
tom tray. Typical operating conditions are:
C1 90.6 3.25 0.091 0.091 82.36 8.24
Pressure, kPa   69 – 103
C2 4.3 0.9 0.329 0.329 2.89 1.41
Feed Temperature, °C 93 – 110
C3    3.2 0.37 0.80 0.75 0.80 2.40 Bottom Temperature, °C 115 – 121
iC4    0.5 0.21 1.41 0.96 0.02 0.48 Reboil Heat, kW 223 – 446
nC4    1.0 0.17 1.74 0.985 0.015 0.985 Residual H2S, mg/kg 0.5 – 2.0
C6    0.4 0.035 8.46 1.0 0.0 0.40 Overhead vapors from sour water strippers contain hydrogen
Total 100.0 86.085 13.915 sulfide, steam, trace amounts of hydrocarbons and, in some
plants, carbon dioxide. These vapors are usually sent to the re­
The use of an average absorption factor, as defined in Equa­ generator (still) condenser in plants using aqueous treating so­
tion 19-28, ignores the change in gas volume from inlet to out­ lutions. Alternatively, the vapors may be sent directly to the
let. Also, the assumptions of average temperature and K-values sulfur recovery unit, or incinerated if emission standards are
can cause significant errors in the preceding calculation meth­ not exceeded.
od.
Foaming occurs in sour water strippers and the tower diam­
Fig. 19-51 can also be used to determine the trays required eter should be based on operation at 50 to 70 percent of the
for a given lean oil rate or to calculate recoveries with a given flooding loads for a non-foaming system.25
oil rate and tray count. Fig. 19-51 shows that oil rate declines
with increasing number of trays and that beyond about eight The required number of theoretical trays and stripping va­
theoretical trays little increase in efficiency is achieved. por quantity can be calculated as shown in the following ex­
ample. However, the results of such calculations must be used
Since higher oil rates require more energy for heating, cool­ only as a guide to the relative effects of changing vapor rates
ing, and pumping, the optimum design is usually one that uses and trays. This is because tray efficiencies or packing HETPs
the minimum possible oil rate with a reasonable size absorber. are not known accurately and the effects of other components
The lowest molecular weight lean oil should be used. This in the sour water change the apparent solubility of hydrogen
will be fixed by oil vapor pressure and absorber operating tem­ sulfide. Ammonia, which is common in refinery sour waters,
perature. Most problems in absorber operation center around can increase the hydrogen sulfide solubility by a factor of 10 or
oil quality and rates. Proper stripping of the oil is necessary to more. A more detailed design procedure is then required for
minimize lean oil losses to the gas and to maximize absorption refining sour water strippers than that given in the following
capacity. example.26
Example 19-8 — Sour water containing 2500 ppmw of hydro­
­STRIPPER CALCULATIONS gen sulfide is to be stripped to 1.5 ppmw. Enough indirect re­
boiler heat is provided to allow 90 kg of steam to leave the top
In a calculation sense, a stripper is simply an upside-down tray for each cubic meter of feed. The feed rate is 0.038 m3 /min
absorber. For hand calculations, a stripping factor is defined as and the tower top is to operate at 145 kPa. Determine the num­
KV ber of theoretical trays required.
ST = Eq 19-31
L Set desired overall material balance:
then: Feed = (0.038) (1000) (60 min/h) = 2280 kg/h
Xm + 1 – X1 SmT + 1 – ST Overhead steam = (90) (60) (0.038) = 205 kg/h
= = Es Eq 19-32
Xm + 1 – Yo SmT + 1 – 1
Fig. 19-51 can be used to perform stripper calculations in a
similar manner to absorber calculations.

19-32
 Feed Bottoms Overhead Two theoretical trays would be required for the stated condi­
kg/h kg/h kg/h tions. Since tray efficiencies or packing HETPs are not predict­
able, 10 actual trays or 6 m of packing would be used.
H2S     5.7     0.003    5.697
The relative effect of various operating pressures and re­
Water 2274.3 2069.300 205.0 boiler heat rates can be estimated by the above method.
2280.0 2069.303 210.697
Required fraction of H2S to be stripped:
­REFERENCES
1. Chien, H. H. Y., “A Rigorous Method for Calculating Minimum
5.697/5.7 = 0.99945 Reflux Rates in Distillation,” AIChE Jour. 24, July, 1978.
Estimate top temperature: 2. Chien, H. H. Y., “A Rigorous Calculation Method for the Mini­
mum Stages in Multicomponent Distillation,” Chem. Eng. Sci.
Fraction water vapor in overhead 28, 1967-74, 1973.
205/210.697 = 0.973 3. Fenske, M. R., “Fractionation of Straight-Run Pennsylvania
Partial pressure water in overhead Gasoline,” Ind. Eng. Chem. 24, 482-5, 1932.
4. Winn, F. W., “New Relative Volatility Method for Distillation
0.973 (145) = 141 kPa Calculations,” Pet. Refiner. 37(5), 216-218, 1958.
Temperature (from steam table, Fig. 24-29) 5. Underwood, A. J. V., “Fractional Distillation of Multicomponent
at 141 kPa = 109°C Mixtures,” Chem. Eng. Prog. 44, 603-14, 1948.
Estimate the K-value for H2S at top conditions: 6. Erbar, J. H., and Maddox, R. N., “Latest Score: Reflux vs. Trays,”
Petr. Refiner 40(5), 183-188, 1961.
K = Henry’s Constant/Total pressure
7. Fair, J. R., and Bolles, W. L., “Modern Design of Distillation Col­
Henry’s Constant for H2S at 109°C = 1.41 (105) kPa (Fig. 19-52) umns,” Chem. Engr. 75(9), 156-178, April 22, 1968.
K = 1.41 (105)/145 = 976.2 8. Katz, D. L., et al., “Handbook of Natural Gas Engineering,” Mc­
Graw-Hill, 1959.
V = mols vapor leaving top tray
9. Koch Engineering Co., “Flexitray Design Manual,” 1982.
5.697 205
= + = 11.556 10. Glitsch, Inc., “Ballast Tray Design Manual,” Third Edition.
34 18
11. Nutter Engineering, “Float Valve Design Manual,” Aug., 1981.
L = mols liquid to top tray
12. AIChE, “Bubble-Tray Design Manual,” New York, 1958.
5.7 2274.3
= + = 126.518 13. Smith, B. D., “Design of Equilibrium Stage Processes,” McGraw-
34 18
Hill, 1963.
Use Equation 19-31 to calculate fraction H2S stripped:
14. Vital, T. J., et al., “Estimating Separation Efficiency,” Hyd. Proc.
(976) (11.556) 63, 147-153 Nov., 1984.
ST = = 89.2
126.518 15. O’Connell, H. E., “Plate Efficiency of Fractionating Columns and
(m + 1) Absorbers,” Trans. AIChE 42, 741-755, 1946.
ST – ST
Es = (m + 1) 16. Eckert, J. S., “Selecting the Proper Distillation Column Pack­
ST –1 ing,” Chem. Eng. Prog. 66(3), 39, 1970.
Assume various values for “m” and calculate “Es”: 17. Vital, T. J., et al., “Estimating Separation Efficiency,” Hyd. Proc.
Results are: 63, 75-78 Dec., 1984.
18. Eckert, J. S., “Tower Packings . . . Comparative Performance,”
m Es Chem. Eng. Prog. 59(5), 76-82, 1963.
1 0.98891 19. Kern, D. Q., “Process Heat Transfer,” McGraw-Hill, p. 453-491,
1950.
2 0.99988
20. Fair, J. R., “What You Need to Design Thermosiphon Reboilers,”
3 0.99999 Petr. Refin. 39(2), 105-123, 1960.
21. Kister, H. Z., “C. E. Refresher: Column Internals,” Chem, Engr.,
May 19, 1980, p. 138-142; July 28, 1980, p. 79-83; Sept. 8, 1980,
p. 119-123; Nov. 17, 1980, p. 283-285; Dec. 29, 1980, p. 55-60;
Feb. 9, 1981, p. 107-109; Apr. 6, 1981, p. 97-100.
FIG. 19-52 22. Jamison, R. H., “Internal Design Techniques,” Chem, Eng. Prog.
Henry’s Constants for H2S in Water 65(3), 46-51, 1969.
23. Kremser, A., “Theoretical Analysis of Absorption Process,” Na­
Temp, °C H (H2S), kPa (abs) tional Petro. News, 22(21), 48, 1930.
38 7.58 (104) 24. Brown, G. G., and Souders, M., “Fundamental Design of Absorb­
ing and Stripping Columns for Complex Vapors,” Ind. Eng.
93 12.55 (10 )
4
Chem. 24, 519, 1932.
149 17.93 (104)

19-33
25. Walker, G. J., “Design Sour Water Strippers Quickly,” Hyd. 28. Van Winkle, M., “Distillation,” McGraw-Hill, p. 480-645, 1967.
Proc., June 1969, pp. 121-124.
29. Chen, G. K., “Packed Column Internals,” Chem. Engr., March 5,
26. Beychok, M. R., “Aqueous Wastes from Petroleum and Petro­ 1984, p. 40-51.
chemical Plants,” John Wiley & Sons, Ltd., 1967.
30. Campbell, J. M., “Gas Conditioning and Processing Vol. 2,”
27. Gillespie, P. C., and Wilson, G. M., “Vapor-Liquid and Liquid- Campbell Petroleum Series, p. 4, 1978.
Liquid Equilibria,” GPA-RR-48, 1982.
31. Kister, Henry, “Distillation Design,” McGraw-Hill, Chapter 9,
1992.

19-34