SOUR WATER ANALYSIS USING A DUAL-

COLUMN HEADSPACE SAMPLING SYSTEM

Andy Burgess, David Greaves, Daniel Murphy and Sam Rolley

Applied Analytics, Inc.
29 Domino Drive, Concord, MA 01742

KEYWORDS

Sour Water, Headspace, H2S, NH3, Spectroscopy

ABSTRACT
In the petroleum refining process one of the main byproducts is wastewater rich in
hydrogen sulfide and ammonia, commonly referred to as sour water. These chemicals
must be removed to prevent contamination and corrosion. A process called Sour Water
Stripping accomplishes this removal. Accurate measurement of hydrogen sulfide and
ammonia concentrations at the outlet of the stripper provides critical process validation
information and allows for corrective action in the event of stripper failure. To carry out
this measurement in real time, a dual-column headspace sampling system is utilized in
conjunction with a UV-diode array based spectrometer. This system allows for the
simultaneous measurement of hydrogen sulfide and ammonia with a single device. A
detailed discussion of sour water measurement benefits, the unique features of headspace
sampling, and experimental results are presented.

INTRODUCTION

Sour water is defined as any refinery process water that contains hydrogen sulfide (H2S).
With such a broad definition, sources of sour water vary significantly. Atmospheric
crude columns and vacuum crude towers produce sour water from condensed stripping
steam removed by overhead condensing systems. Steam cracking and fluid catalytic
cracking units produce sour water as condensates from steam used in injection stripping
and aeration [1].

Refineries create sour water as a byproduct in many of their processes. Unfortunately,

sour water contains ammonium salts such as ammonium bisulfide and ammonium
Session 7.3: – Page 1 –
© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
carbonate. These ammonium salts can cause several problems. In an extensive study of
corrosion rates in sour water, a number of variables were considered including
ammonium bisulfide concentration, velocity, H2S partial pressure, temperature, chloride
concentration, hydrocarbon content, and chemical treatments. It was confirmed that the
concentration of ammonium bisulfide is the most influential variable contributing to
corrosion [5]. Ammonium salts can also settle on orifices causing plugging and increase
maintenance concerns.

As a result, refineries need to remove the hydrogen sulfide and ammonia (NH3) from the
water to avoid the formation of these ammonium salts when the water is reused for other
plant processes. To remove the H2S and NH3, a sour water stripper is often utilized. The
plant operators need to know how efficient the stripping process is and if the water is
suitable for reintroduction.

TECHNOLOGY

To address the need brought to light in the introduction, Applied Analytics developed a
system utilizing the OMA-300 for the measurement of H2S and NH3 in sour water. It
utilizes an ultraviolet-visible full-spectrum spectrometer and a dual column headspace
sampling approach.

The combined system offers many benefits. It allows a large dynamic range in the
measurement of the analytes due to flexibility in wavelength selection. Cross
interference from moisture is avoided, because the measurement is being done in the UV-
VIS region where water doesn’t absorb light. Finally, consumables and/or reagents are
not necessary for operation.

The technology is broken down into two sections for ease of explanation: an analytical
section and a sample conditioning section.

ANALYTICAL

The analytical configuration consists of the following equipment: a xenon flash lamp, an
ultraviolet-visible full-spectrum spectrometer, collimating lenses, fiber optic cables, and a
flow through cell. This layout can be seen in Figure 1.

Absorbance is defined using the following equation:

A = log10 (I0/I) (1)

Where: A = Absorbance
I0 = Light In
I = Light Out

Session 7.3: – Page 2 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
FIGURE 1. ANALYTICAL HARDWARE

The measurement algorithm used for analysis is a classical least squares regression of
Beer’s Law [2]:

Al = elbc (2)

Where: A = Absorbance
l = Wavelength
e = Molar Absorptivity
b = Path Length
c = Concentration

The spectrometer measures I0 during its zero calibration and I when the analyte is in the
flow through cell of path length b. el is found by introducing a known concentration of
the analyte into the flow-through cell and is saved for use in subsequent readings to
determine unknown concentrations of the analyte.

This would be adequate if the absorbance at a given wavelength could be attributed to

only one component. However, this isn’t the case with H2S and NH3 in the ultraviolet
region. At each wavelength in the UV where NH3 absorbs, so does H2S. This is
illustrated in the following spectra (Figures 2 & 3).

Session 7.3: – Page 3 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
 H2S (gas) Absorbance Curve

1.2

) 0.8
U
A
(
e
c 0.6
n
a
b
r
o 0.4
s
b
A
0.2

0
215 220 225 230 235 240 245 250
-0.2
Wavelength (nm)

FIGURE 2. H2S ABSORBANCE CURVE

NH3 (gas) Absorbance Curve

0.9

0.8
)
U 0.7
A
(
e 0.6
c
n
a
b
r 0.5
o
s 0.4
b
A
0.3

0.2

0.1

0
215 220 225 230 235 240 245 250
Wavelength (nm)

FIGURE 3. NH3 ABSORBANCE CURVE

Session 7.3: – Page 4 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
Therefore, in this case the equation must be extended to the following:

Alx = e1lx* b * c1 + e2lx* b * c2 (3)

Aly = e1ly* b * c1 + e2ly* b * c2 (4)

Where lx could be 217nm and ly could be 220nm, for example (the over determination
is omitted for simplicity). The final concentration readings are determined by solving for
c1 and c2 using the previous equations. Accordingly, before the gas can be analyzed it
must be coerced into the vapor phase. This is accomplished by the headspace sample
conditioning system.

SAMPLE CONDITIONING

The heart of the headspace system is the stripping column (Figure 4). Its dimensions are
24” in height by 2” in diameter. The column is sealed with gaskets that are held together
by sanitary clamps. The column is filled with Pall rings for efficient mass transfer
between the liquid sample and the carrier gas.

The column is heated with three band heaters and controlled with a thermocouple and a
temperature controller. The flow rates of the carrier gas and the sample liquid are
controlled by variable area flow meters with integrated flow control valves. A
backpressure regulator regulates the pressure in the stripping column and the flow
through cell.

During operation, the liquid sample enters through a fitting on the top of the column and
exits through a fitting on the bottom of the column. A carrier gas is introduced through
another fitting on the bottom of the column and exits through a fourth fitting on the top of
the column. This establishes a countercurrent flow between the liquid sample and the
carrier gas. The Pall rings allow for efficient mass transfer between the liquid sample and
the carrier gas by greatly increasing the total internal surface area of the column. After
the carrier gas leaves the column it flows to the flow through cell for measurement.

The headspace system is based on the principle of Henry’s Law:

C = k*Pgas (5)

Where: P = Partial Pressure

k = Henry’s Law Constant
c = Concentration

Henry’s Law states that the amount of a gas dissolved in a solution at a given temperature
is directly proportional to the partial pressure of the gas above the solution [3].

Session 7.3: – Page 5 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
FIGURE 4. HEADSPACE SAMPLE CONDITIONING SYSTEM

Therefore, given constant conditions such as temperature, pressure, carrier gas flow rate
and liquid sample flow rate the headspace gas being measured can be correlated to the
concentration of the liquid sample.

Although the headspace sampling method can be used in a variety of applications, it is

most advantageous in measurements of components with high volatility in opaque
samples that have interfering components with low volatility. Sour water is a good
example of such a stream, since it can often contain phenols that were formed from
reactions between steam and cyclic hydrocarbons.

Phenols have a very strong absorbance in the low UV region (1500 L/(mol*cm) @
228nm), which would make measurement of the H2S or NH3 in the liquid phase very
difficult. Fortunately, phenols also have a much higher boiling point (see Table I) than
H2S or NH3.

As a result, only very low phenol concentrations can exist in the headspace gas, thus
making the measurement easier by minimizing the interfering component.

Session 7.3: – Page 6 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
TABLE I. BOILING POINT COMPARISON

Phenol H2S NH3

Boiling Point © 181.7 -60.28 -33.34

EXPERIMENT
The objective of the experiment is to quantitatively demonstrate the typical accuracy and
stability that the headspace sampling system offers when correlated with a liquid sample
of H2S and NH3 in water.

PARAMETERS

A number of parameters need be determined before starting the experiment.

The proper liquid sample flow rate shall be constant. The liquid sample flow rate was
experimentally determined by filling the column with water and draining it with just
gravity as a motive force. The column was packed with Pall rings, as it would be in
normal operation, during this test. The time in which it takes the water to fully evacuate
the column was recorded. The volume of water as it leaves the column was captured in a
graduated beaker, thus allowing the determination of the volume of the column with Pall
rings installed. That information was then used to find the volumetric flow rate (V/t).

Since Henry’s Law is dependent upon temperature, it was controlled and kept constant at
a temperature that promotes low solubility of the gas in the water. The dependence of the
solubility on temperature is described by the following equations [4]:

k = kt * exp [ (-enthalpy/R) * ((1/T) – (1/Tt)) ] (6)

(-d*ln(k))/(d*(1/T)) = enthalpy/R (7)

Where k = Henry’s constant of a specific gas (M/atm)

kt = Henry’s constant of a specific gas at a temperature of 298.15 K (M/atm)
R = universal gas constant (8.314 (J/mol*k))
T = System temperature (K)
Tt = 298.15 K
(-d*ln(k))/(d*(1/T)) = Temperature Dependence (K)

TABLE II. HENRY’S LAW CONSTANT & TEMPERATURE DEPENDENCE

H2S NH3
Henry's Law Constant (M/atm) 0.1 61
Temperature Dependence (K) 2200 4200

Session 7.3: – Page 7 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
Table II gives a value for temperature dependence, which according to Equation 7 is
equal to enthalpy/R. That value was substituted into Equation 6 to find the Henry’s Law
Constant at different temperatures. Then Equation 5 was used to predict the
concentration of the gas in the water.

TABLE III. HENRY’S LAW CONSTANT FOR H2S AT DIFFERENT

TEMPERATURES

k (M/atm) kt (M/atm) E/R (K) T (K) Tt (K)

0.0787 0.1 2200 308.15 298.15
0.0628 0.1 2200 318.15 298.15
0.0509 0.1 2200 328.15 298.15
0.0417 0.1 2200 338.15 298.15
0.0346 0.1 2200 348.15 298.15
0.0290 0.1 2200 358.15 298.15
0.0245 0.1 2200 368.15 298.15

TABLE IV. HENRY’S LAW CONSTANT FOR NH3 AT DIFFERENT

TEMPERATURES

k (M/atm) kt (M/atm) E/R (K) T (K) Tt (K)

38.6184 61 4200 308.15 298.15
25.1617 61 4200 318.15 298.15
16.8277 61 4200 328.15 298.15
11.5251 61 4200 338.15 298.15
8.06690 61 4200 348.15 298.15
5.75998 61 4200 358.15 298.15
4.18873 61 4200 368.15 298.15

Tables III and IV show that as temperature increases, the Henry’s Law constant and the
concentration of the dissolved gas in the water decreases. Therefore, it is beneficial to
choose a high temperature for the column set point because it forces more of the
dissolved gas from the liquid phase into the vapor phase. A larger concentration in the
vapor phase results in more absorbance in the path length of the flow cell. Furthermore,
the accuracy of the analyzer increases as absorbance increases. Keeping this in mind,
along with practical limitations of the setup, 50 degrees C was chosen for the set point of
the column.

The carrier gas flow rate was also kept constant. The ideal carrier flow rate was set based
on the highest flow rate that was achieved without decreasing the concentration of the
analyte in the stripping carrier gas. This was determined experimentally by analyzing the
carrier gas at several carrier gas flow rates and recording the results.
Session 7.3: – Page 8 –
© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
The analyzer (UV absorbance) was calibrated on gas standards of H2S in N2 and NH3 in
N2. These spectral standards were used to establish an extinction coefficient for each gas
at each wavelength in the range of interest. This calibration was done prior to water
testing.

It is also important to note that pH plays an important role in sour water stripping. At
low pH levels near 5.5, NH3 tends to be held in solution and H2S tends to be more
volatile. However, at higher pH values near 9.0, NH3 is more volatile and H2S tends to
stay in solution [1]. For the purposes of these experiments a constant pH was assumed.

PROCEDURE

A sample of water without H2S or NH3 was introduced into the column. The temperature
of the column, the carrier gas flow rate and the liquid sample flow rate were held constant
at their optimal values. This sample was used as a zero reference by the analyzer.

A sample of water with a known concentration of H2S was prepared in a reservoir large
enough to hold the volume required to run a 1-hour stability test. The temperature of the
column, the carrier gas flow rate and the liquid sample flow rate were held constant at
their optimal values.

The concentration log function on the analyzer was activated and the system was left to
run for 1 hour. The start time and the end time were recorded. At the end of the 1-hour
run, a reservoir of clean water was connected to wash the system.

Second, third, and fourth samples of water with different known concentrations of H2S
were prepared and introduced to the system in the same manner as the first. A 1-hour run
was not required as these samples were for the purpose of testing accuracy, not stability.

After the system was washed with clean water again, a sample of water with a known
concentration of NH3 was prepared in a reservoir large enough to hold the volume
required to run a 1-hour stability test.

The concentration log function on the analyzer was activated and the system was left to
run for 1 hour. The start time and the end time were recorded. At the end of the 1-hour
run, a reservoir of clean water was connected to wash the system. Second, third, and
fourth samples of water with different known concentrations of NH3 were prepared and
introduced into the system in the same manner as the first.

RESULTS

The data retrieved from the concentration log file was used to show the stability and
accuracy of the headspace system using H2S and NH3 concentrations in water.

Session 7.3: – Page 9 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
TABLE V. H2S STABILITY TABLE

Test Duration (hours) 1

Reading Stored (1/sec) 60
Average Value (PPM) 980.54
Maximum Value (PPM) 987.75
Minimum Value (PPM) 968.87
Standard Deviation (PPM) 6.54
Slope (PPM/reading) .36
Slope * 60 (PPM/hour) 2.35

H2S Stability

2000.00
Concentration (ppm)

1500.00

1000.00

500.00

0.00
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58

Time (seconds)

FIGURE 5. H2S STABILITY GRAPH

The first set of data is the stability of H2S in water over 1 hour. The average value over
the 1-hour run was determined to be 980.54, with the maximum reading at 987.75 and the
minimum reading at 968.87. The standard deviation of the data set is 6.54 and it has a
slope of .36.

TABLE VI. NH3 STABILITY

Test Duration (hours) 1

Reading Stored (1/sec) 60
Average Value (PPM) 999.99
Maximum Value (PPM) 1009.82
Minimum Value (PPM) 990.73
Standard Deviation (PPM) 4.09
Slope (PPM/reading) 0.02
Slope * 60 (PPM/hour) 1.10

Session 7.3: – Page 10 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
 NH3 Stability

2000.00
)1800.00
m
p1600.00
p
(1400.00
n1200.00
o
ti1000.00
a
tr 800.00
n 600.00
e
c 400.00
n
o 200.00
C 0.00
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Time (seconds)

FIGURE 6. NH3 STABILITY GRAPH

The second set of data is the stability of NH3 in water over 1 hour. The average value
over the 1-hour run was determined to be 999.99, with the maximum reading at 1009.82
and the minimum reading at 990.73. The standard deviation of the data set is 4.09 and it
has a slope of 0.02.

TABLE VII. H2S ACCURACY TABLE

Liquid Concentration – Sample 5 (% of Cal Sample) 43.5

Liquid Concentration – Sample 6 (% of Cal Sample) 61.25
Liquid Concentration – Sample 7 (% of Cal Sample) 91.25
Liquid Concentration – Sample 8 (% of Cal Sample) 100
Gas Concentration – Sample 5 (PPM) 1300
Gas Concentration – Sample 6 (PPM) 2600
Gas Concentration – Sample 7 (PPM) 3800
Gas Concentration – Sample 8 (PPM) 4200
Linearity – R2 Value .9996

The third set of data is the accuracy of H2S in water. The gas concentration reading for
Sample 1 was 1300ppm. The gas concentration reading for Sample 2 was 2600ppm. The
gas concentration reading for Sample 3 was 3800ppm. The gas concentration reading for
Sample 4 was 4200ppm. The linearity (R2 value) of the gas concentration readings was
.9996.

Session 7.3: – Page 11 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
 H2S Accuracy
R² = 0.9996
4500
Gas Concentration (PPM)

4000
3500
3000
2500
2000
1500
1000
500
0
0 20 40 60 80 100 120

Liquid Concentration (% of Cal Sample)

FIGURE 7. H2S ACCURACY GRAPH

TABLE VIII. NH3 ACCURACY TABLE

Liquid Concentration – Sample 5 (% of Cal Sample) 25

Liquid Concentration – Sample 6 (% of Cal Sample) 50
Liquid Concentration – Sample 7 (% of Cal Sample) 75
Liquid Concentration – Sample 8 (% of Cal Sample) 100
Gas Concentration – Sample 5 (PPM) 115.80
Gas Concentration – Sample 6 (PPM) 184.15
Gas Concentration – Sample 7 (PPM) 245.10
Gas Concentration – Sample 8 (PPM) 314.06
Linearity – R2 Value 0.9994

The fourth set of data is the accuracy of NH3 in water. The gas concentration reading for
Sample 5 was 115.80ppm. The gas concentration reading for Sample 6 was 184.15ppm.
The gas concentration reading for Sample 7 was 245.10ppm. The gas concentration
reading for Sample 8 was 314.06. The linearity (R2 value) of the gas concentration
readings was .9994.

Session 7.3: – Page 12 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
 NH3 Accuracy
R² = 0.9994
350
Gas Concentration (PPM)

300
250
200
150
100
50
0
0 20 40 60 80 100 120
Liquid Concentration (% of Cal Sample)

FIGURE 8. NH3 ACCURACY GRAPH

CONCLUSIONS
Refineries create sour water as a byproduct in many of their processes. Sour water has
been proven to cause corrosion and plugging due to the presence of ammonium salts,
such ammonium bisulfide. Therefore, it is beneficial for refineries to remove the
hydrogen sulfide and ammonia so the water can be reused or discarded. This is often
done through a stripping process.

Measurement of the H2S and NH3 in sour water occurs afterwards to ensure the
effectiveness of the stripper before the water is sent for reuse. The measurement is not
particularly easy without advanced sampling techniques.

Therefore, a sampling technique was developed that used some of the same principles of
stripping that the refineries themselves use to remove H2S and NH3 from the sour water
stream. After the concentrations of each of these chemicals are measured in the gas
phase, they are correlated back to a liquid concentration.

In conclusion, the results from the stability test demonstrate a drift of less than 0.5% full
scale. Also, the headspace system yields a linearity of 0.9994 or better when comparing
carrier gas to sample stream analyte concentrations. Coupled with the benefits mentioned
in the introduction, these analytical results make the UV-VIS full-spectrum spectrometer

Session 7.3: – Page 13 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.
with the headspace sample conditioning system an ideal solution for the measurement of
H2S and NH3 in sour water.

REFERENCES
1. Armstrong, Tim, Scott, Bruce, Taylor, Kin, and Gardner, Art, “Sour Water
Stripping”, Today’s Refinery, June, 1996.

2. Mark, Howard, Workman, Jerome, Classical Least Squares, “Part 1:

Mathematical Theory”, Spectroscopy, May, 2010.

3. Zumdahl, Steven S., Zumdahl, Susan A., Chemistry – 6th Edition¸ Houghton
Mifflin, Boston, Massachusetts, 2006.

4. Sander, Rolf, Compilation of Henry’s Law Constants for Inorganic and Organic
Species of Potential Importance in Environmental Chemistry, Max Planck
Institute of Chemistry, April 1999.

5. Horvath, Richard J., Lagand, Vishal V., Srinivasan, Sridhar, Kane, Russel D.,
Prediction and Assessment of Ammonium Bisulfide Corrosion Under Refinery
Sour Water Service Conditions – Part 2., NACE International, 2010.

Session 7.3: – Page 14 –

© Copyright 2011, International Society of Automation. All rights reserved.
ISA 56th Analysis Division Symposium 2011, League City, TX.