Stripping of Volatile Organic Compounds From Refinery Waste Water Objective The major purpose of this writing is to provide a quick but reasonably accurate method for sizing a stripper to reduce the concentration of a volatile organic compound (VOC) in a typical refinery waste water stream down to some specified or acceptable level. The procedure presented here employs a simple BASIC computer program to compute the number of theoretical stages using Kremser’s equation. Then certain correlations and methods are used to determine stripper diameter for cases where sieve trays are to be used and also cases where random packing might be considered. The actual number of trays (sieves) or height of packing needed are determined by applying an appropriate tray efficiency or recommended HETP (height equivalent to a theoretical plate). Background Volatile organic compounds (VOC’s) generally exhibit limited solubility with water but still to the extent that they can be a dangerous contaminant. Common examples of VOC’s would be benzene, toluene, xylenes, n-hexane (benzene precursor), cyclohexane (benzene precursor), naphthalene, acetone and the chlorinated hydrocarbons. The vapor stripping agent is generally steam or basically an inert gas stream such as refinery tail gas (H 2 + CH 4 ) or air. Steam stripping is generally conducted at higher temperatures than inert gas stripping, usually close to the boiling point of water. Since the volatility of the organic contaminants is a strong function of temperature, stripping can be done most efficiently and effectively. This includes the removal of the heavier, more soluble organics such as phenol that are not readily strippable with inert gas. Many times steam is not economically available and another source of stripping gas is sought such as fuel gas (H 2 + CH 4 ). The stripper is generally operated at low pressure and close to ambient temperature. Under these conditions, the removal of the lighter VOC’s such as benzene is quite effective. Kremser Equation Strippers for removing VOC’s from waste water are generally conducted at conditions of nearly constant pressure and temperature and also constant liquid and vapor flow rate (equal molal overflow). These conditions are closely met because the VOC’s present in the liquid streams are at the ppm level of concentration. The top half of Figure 1 shows a simple sketch of a typical stripping column. The constant liquid and vapor flow rates are designated by the symbols L and V in mols/hr. x and y denote the mole fractions of the solute species (VOC) in the liquid and vapor streams. The subscript a refers to conditions around the top of the stripper and subscript b to conditions around the bottom. y a * is the vapor composition in equilibrium with the liquid feed stream, and y b * is the vapor composition in equilibrium with the liquid exiting the bottom.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Stripping of Volatile Organic Compounds
From Refinery Waste Water
Objective The major purpose of this writing is to provide a quick but reasonably accurate
method for sizing a stripper to reduce the concentration of a volatile organic compound
(VOC) in a typical refinery waste water stream down to some specified or acceptable
level. The procedure presented here employs a simple BASIC computer program to
compute the number of theoretical stages using Kremser’s equation. Then certain
correlations and methods are used to determine stripper diameter for cases where sieve
trays are to be used and also cases where random packing might be considered. The
actual number of trays (sieves) or height of packing needed are determined by applying
an appropriate tray efficiency or recommended HETP (height equivalent to a theoretical
plate).
Background Volatile organic compounds (VOC’s) generally exhibit limited solubility
with water but still to the extent that they can be a dangerous contaminant. Common
examples of VOC’s would be benzene, toluene, xylenes, n-hexane (benzene precursor),
cyclohexane (benzene precursor), naphthalene, acetone and the chlorinated hydrocarbons.
The vapor stripping agent is generally steam or basically an inert gas stream such as
refinery tail gas (H2 + CH4) or air. Steam stripping is generally conducted at higher
temperatures than inert gas stripping, usually close to the boiling point of water. Since
the volatility of the organic contaminants is a strong function of temperature, stripping
can be done most efficiently and effectively. This includes the removal of the heavier,
more soluble organics such as phenol that are not readily strippable with inert gas.
Many times steam is not economically available and another source of stripping gas is
sought such as fuel gas (H2 + CH4). The stripper is generally operated at low pressure
and close to ambient temperature. Under these conditions, the removal of the lighter
VOC’s such as benzene is quite effective.
Kremser Equation Strippers for removing VOC’s from waste water are generally
conducted at conditions of nearly constant pressure and temperature and also constant
liquid and vapor flow rate (equal molal overflow). These conditions are closely met
because the VOC’s present in the liquid streams are at the ppm level of concentration.
The top half of Figure 1 shows a simple sketch of a typical stripping column. The
constant liquid and vapor flow rates are designated by the symbols L and V in mols/hr. x
and y denote the mole fractions of the solute species (VOC) in the liquid and vapor
streams. The subscript a refers to conditions around the top of the stripper and subscript
b to conditions around the bottom. ya* is the vapor composition in equilibrium with the
liquid feed stream, and yb* is the vapor composition in equilibrium with the liquid exiting
the bottom.
-2-
The bottom portion of Figure 1 shows some typical equilibrium and (material balance)
operating lines for the VOC stripper. They are both essentially linear because the
equilibrium constant for the VOC and the liquid and vapor rates are basically constant
throughout the entire column. The liquid feed rate, its composition xa and the bottom
terminal compositions (xb, yb, specs) will all be fixed. The inlet vapor composition is
generally very low (yb 0). The minimum possible vapor rate is determined from the
slope of the operating line when it intersects the equilibrium line at the top i.e. when ya =
ya* (a pinch point). By overall solute material balance we would then have,
*
min
(1)a b
a b
x xV
L y y
Theoretically an infinite number of equilibrium stages would be required at this
condition. The next logical step is to increase the vapor rate to some reasonably higher
level than the minimum in order to provide an economic balance between stream flow
rates and some finite number of stages. This is achieved by dropping the top end of the
operating line vertically to a lower vapor composition ya < ya*. This is achieved by
simply increasing the total vapor rate to a desired level of operation. Then the desired
operating V/L ratio is given by,
(2)a b
oper a b
x xV
L y y
For an absorption or stripping column operating at essentially fixed P and T and for
conditions of equal molal overflow, the Kremser equation provides a precise value for the
number of equilibrium (theoretical) stages, N . A detailed derivation of this equation is
presented in Appendix I. The final version of the Kremser equation written in a format
which permits direct evaluation of N is:
*
*
* *
(3)
; ; (4 , , )
b b
a a
b b a a
y yLog
y yN
Log A
Lwhere y K x y K x A a b c
K V
Here the quantity A in the denominator is called the “absorption factor” where K is the
vapor-liquid equilibrium ratio of the VOC in question (K = y/x). Normally, in treating
-3-
or dealing with stripping applications we calculate the stripping factor S which is merely
related to A via the relationship,
1
(5)V
S KA L
Vapor-Liquid Equilibria The level of VOC concentration in either the liquid or vapor
phase is very low and generally in the range of parts per million (ppm). As a result,
Henry’s law can be used to relate the equilibrium concentration between the vapor and
liquid streams. The precise definition of Henry’s law is,
0
(6)i
Si ii solvent
xi
y PH Lim P P
x
Here i is the vapor fugacity coefficient of species i (VOC) in the vapor mixture, and
PsolventS is the vapor pressure of the solvent. At low to moderate pressure and especially
at higher temperatures, the vapor phase can justifiably be treated as being ideal, and i
can be taken to be one. In this case Henry’s law can be expressed as,
(7)i i iy P H x
and the vapor-liquid equilibrium ratio for VOC species i becomes simply,
(8)i ii
i
y HK
x P
In 1983 Tsonopoulos and Wilson (1) studied the mutual solubilities of three C6
hydrocarbons (benzene, cyclohexane and n-hexane) and water at conditions basically at
the three-phase equilibrium pressure. They performed a thermodynamic analysis of their
own experimental measurements plus carefully selected literature data. That portion of
the data which provided the solubility of the hydrocarbons in water was used to calculate
and correlate Henry’s law constants. These values were fitted to an analytical expression
of the form:
2/ (9)
( . deg. ) / .
i
i
Ln H A B T CT D LnT
where T abs temperature is in K and H in MPa mole fr
-4-
Table 1 lists the coefficients to Equation 9 for benzene, cyclohexane and n-hexane in
water for a temperature range extending from the melting point of the VOC hydrocarbon
to the three-phase critical end point. In Table 2 of his paper on the design of steam
strippers for removing VOC’s, Bravo (2) lists some estimates of the values of Hi for
several common VOC’s in water at around ambient temperature (20 deg. C). They are
listed below:
VOC Hi, atm/m.f.
Carbon tetrachloride 1,183
Chloroform 180
Methylene chloride 125
Trichloroethylene 500
Perchloroethylene 800
1,1,1-trichloroethane 200
1,1,2,2-tetrachloroethane 20
Benzene 240
Ethylbenzene 389
Dichlorobenzene 71
Methyl-isobutyl ketone 7.1
Methyl-ethyl ketone 1.7
Another important VOC that can be found in refinery sour water streams is phenol
(C6H5OH). In 1976 Dr. John Lenoir did a comprehensive literature survey of the
solubility of phenol in water. He came up with his best listing of smoothed data
consisting of the Henry’s law constant of phenol in water as a function of temperature
and liquid phase concentration. We took his listing of the Henry’s law constant at infinite
dilution and fitted those values to an equation of the same form as Equation 9 over the
temperature range of 100 to 300 deg. F. The resulting equation fit which predicts the
tabulated Henry’s constants with a standard deviation of better than one percent is,
2168.458 1158.55 / 0.0000162 27.76904 (10)
deg. / . .
i
i
Ln H T T LnT
where T is in R and H is in psia m f
-5-
BASIC Program Table 2 lists the BASIC program called VOCSTR.BAS which is used
to calculate the number of equilibrium stages for a typical VOC stripper employing either
an inert gas or steam as the stripping agent. Lines 17-25 are reserved for input
information. The definition of these inputs is as follows:
A$ = VOC identity (e.g. benzene, phenol, etc.)
B$ = Liquid phase solvent (water in this study)
P = Average stripper column operating pressure, psia
T = Average stripper column operating temperature, deg. F
GPML = liquid flow rate, gals/min (GPM)
SCFMV = Vapor flow rate, SCFM at 60 deg. F and 1 atm
PPMWI = VOC concentration in the feed liquid, ppmw
PPMWO = VOC concentration in the exit liquid, ppmw
MWG = Average molecular weight of the stripping agent (vapor)
DL = Liquid density, lbm/cuft
MWVOC = molecular weight of the VOC
In lines 30-75 the program computes the molar liquid and vapor flow rates and also the
inlet and exit liquid and vapor VOC mole fractions. In lines 85-105 are calculated the
Henry’s law constant, equilibrium ratio and the absorption factor (AF) for the VOC
component. Lines 110-125 compute ya*, yb
* and finally the number of equilibrium stages
N employing the Kremser equation (Eqn. 3 above). The balance of the program is
devoted to printing out pertinent output and the computation of some other important
quantities.
In lines 160-172 are computed and printed the actual cuft/min (ACFM) of vapor flow at
the operating P and T of the stripper and also the vapor density (ideal gas). Then in lines
195-225, the minimum vapor to liquid molar ratio (V/L)min, minimum vapor rate
(SCFM), actual or operating V/L, and the ratio over minimum V/L are all computed and
printed. The rest of the printout (lines 235-270) consists of data that were inputted
directly. A more detailed description of the output will be covered when we look at two
subsequent illustrations to be provided later.
Sieve Tray Design The first consideration in the design of the VOC stripper is to
determine an appropriate column diameter. The diameter must be sufficient in order to
insure that column flooding or entrainment will not occur. McCabe, Smith and Harriott
(3) provide an excellent discussion dealing with operating vapor velocity limits for sieve
tray columns. In fact, they recommend the specific correlation of J. R. Fair (4) for
calculating flooding conditions for sieve and bubble cap plates. A brief discussion of
flooding (excessive entrainment) phenomena and a summary of the Fair correlation are
presented below.
-6-
The upper limit of velocity in a sieve-tray column is determined by the flooding point or
by the velocity at which entrainment becomes excessive. Flooding will occur when the
liquid in the downcomer backs up to the next plate. This is determined primarily by the
observed pressure drop across the plate and the plate spacing.
A long established empirical correlation for predicting maximum permissible vapor
velocity (at the flood or entrainment point) is expressed by the equation,
, / sec (11)L Vv v
V
u K ft
where uv is the maximum permissible velocity based on bubbling or active area, and Kv
(ft/sec) is an empirical coefficient. Kv is usually evaluated from plant data and correlated
with such typical parameters as plate spacing and the liquid/vapor flow rates. Fair (4) has
amended Equation 11 to include the effect of liquid surface tension. This modified
version is,
0.2
; ( / ) (12)20
L Vv v
V
u K surfacetension dynes cm
A value of = 20 dynes/cm is typical of organic liquids. For water ( = 72 dynes/cm)
the correlation would predict about a 30 percent higher flooding velocity than for the
organics. The correlation is not recommended for liquids of very low surface tension or
for systems which foam easily. Fair and co-workers have developed a graphical
correlation which expresses the coefficient of Equation 12 as a function of tray spacing
and the flow parameter F.P. defined below.
0.5
. . (13)V
L
LF P
V
where L/V is the ratio of mass flow rate of liquid to vapor. L and V are the liquid and
vapor densities generally expressed in units of lb/cu ft. The graphical correlation is
applicable to both sieve and bubble cap plates. This graph appears as Figure 18-30 on
Page 515 of the textbook by McCabe et. al. (3).
For safe column operation it is best to use a value of Kv that is around 75 percent of the
value determined from the above correlation. This will then provide an operating
superficial vapor velocity (Eqn. 12) that is safely below that which would initiate
flooding or entrainment. This velocity in turn will then permit us to determine an
appropriate column diameter.
-7-
A very typical overall tray efficiency that if often employed for stripping applications
such as the removal of dilute VOC concentrations in water is around 25 percent. This is
basically the same efficiency value that is used for the design of sour water strippers in
general.
Random Packing Design For illustrating the use of random packing in the sizing and
design of a VOC stripper, we have chosen Intalox Metal Tower Packing ( IMTP)
developed by the Norton Company (5). The computation procedure for maximum vapor
flow capacity using the IMTP packing was essentially developed by Strigle (6,7).
The correlation uses the same flow parameter F.P. (Eqn.13) as is used for the sieve tray
correlation. The vapor load in a column is expressed in terms of the C-factor Cs and is
defined by the expression,
, / sec (14)Vs s
L V
C v ft
The C-factor is basically the superficial vapor velocity corrected for the gas and liquid
densities. In another sense, it is a column capacity factor, in that it is used to correlate
what vapor flow rates or vapor velocities will initiate flooding or liquid entrainment.
In the work that led to the development of high void fraction packing e.g. IMTP packing,
it was found that vapor velocities could be increased to such a high rate that significant
liquid entrainment was produced without producing a corresponding high pressure drop.
This entrained liquid was carried up the column resulting in a drop of separation
efficiency at vapor velocities that were still below the packing’s maximum hydraulic
capacity (flood point). The Maximum operational Capacity (MOC) is defined as the
point at which the separation efficiency starts to decline appreciably. The MOC is not
identical to flood, but it commonly occurs at about 90 percent of the flood point.
For this correlation we replace the C-factor Cs with the term Cmoc in Equation 14. By
experimentation, Strigle (6,7) reported the MOC of IMTP packing for many systems. He
prepared a plot of a factor identified as Cuncorr.,moc as a function of the flow parameter for
IMTP packings of sizes # 25, #40, #50 and #70. We cannot present this graph here
because of proprietary reasons. Cuncorr.,moc represents the capacity factor for the specific
conditions where:
, 20 /
cos , 0.20L
Surfacetension dynes cm
liquid vis ity cps
-8-
Strigle’s data indicates that the maximum increase of Cmoc is 8 percent for liquids with
higher surface tension. Hence for water which has a surface tension in the range 60-73
dynes/cm range, Cmoc is increased by 8 percent. In addition he found the Cmoc value at
fixed flow parameter varies as the liquid viscosity to the –0.11 power. Upon correcting
the C-factor at MOC for liquid viscosity and for the maximum effect of surface tension
(aqueous systems), we get:
0.11
.,
0.20(1.08) (15)moc uncorr moc
L
C C
For a safe design, it is recommended that the vapor rate at the highest loaded point in the
packed bed should not exceed 75 percent of the predicted Maximum Operational
Capacity (Eqn. 15). Therefore the following equation was recommended for sizing the
tower diameter when IMTP packing is employed:
0.75 (16)L Vs moc
V
v C
Illustration 1 We are asked to provide an appropriate design of a stripper for reducing the
concentration of benzene in a waste water stream from 700 ppmw (parts per million by
weight) down to 1 ppbw (parts per billion by weight). The liquid flow (feed) rate is to be
constant throughout the stripper at 300 gpm. The average operating pressure and
temperature of the stripper will be 70 psia and 85 deg. F respectively. The stripping gas
to be employed for this service is a benzene-free hydrogen-methane rich ethylene plant
tail gas with a molecular weight of 10.28. We are specifically asked to choose a
reasonably economic gas rate (SCFM) and look at the following two options as far as the
column (stripper) internal hardware is concerned:
a. Sieve trays with an 18-inch tray spacing and 25 % overall tray efficiency
b. No. 50 intalox metal tower packing (IMTP)
Sieve tray design: In order to be able to select an appropriate design gas rate, we need to
first establish a relationship between the number of theoretical (equilibrium) stages and
gas rate. Program VOCSTR.BAS (Table 2) was run for a host gas rates. A summary of