Stripping sour water: the effect of heat
stable salts
Mass transfer rate-based simulation models sour water stripping and assesses
the distribution of ammonia in conventional amine treating systems
Ralph Weiland and Nathan Hatcher Optimized Gas Treating, Inc
n a recent article dealing with a live steaminjected stripper (PTQ, Q3 2012), we discussed
how steam usage affected the stripped water
residual ammonia and hydrogen sulphide levels,
how ammonia distributed itself within the stripper in an unexpected way, and how Murphree
(1925) vapour efficiencies varied with location
within the tower at various stripping steam
rates.4 The discussion was limited to nonphenolic sour water. In the present article, we
examine the effect of heat stable salts (HSS) on
sour water stripper (SWS) performance, how the
injection of caustic soda can spring ammonia
from the sour water, and how caustic injection
can worsen H2S stripping if it is injected at the
wrong place, or too much caustic is injected. The
analysis uses a mass transfer rate-based simulation model for sour water stripping and for
assessing the distribution of ammonia in conventional amine treating systems.
Sour water sources
The sour water generated in refineries comes from
numerous sources. Most refinery sour water
systems contain very little CO2, but H2S levels
can become very high. The capacity of ammonia
solutions for H2S is a direct result of the weak
acid-weak base reactivity between H2S and
ammonia. The potentially high H2S content can
make sour water extremely foul, and H2S
removal from the sour water to quite low levels
is mandatory to avoid unacceptable pollution
levels. Many sour water sources have been noted
in the excellent review article by Asquith and
Moore.1
Sour water is generally classified as phenolic or
non-phenolic. Non-phenolic water contains
www.digitalrefining.com/article/1000543
almost exclusively NH3, H2S and possibly a trace
of CO2. It is generated by refinery hydrotreating
(hydrodesulphurisation, or HDS) units. When
stripped of contaminants, non-phenolic water can
typically be recycled for reuse in the HDS unit as
wash water, or it can be used as make-up water to
the crude desalting process. Phenolic (or more
broadly, non-HDS) water typically contains HSS,
phenols and caustic.
Finally, it may be useful to point out that
ammonia and hydrogen sulphide have almost
unlimited solubility in water when they are
present together. This is an interesting consequence of the fact that the reactive component
of the solvent, ammonia, is volatile and, if
present in the gas phase, it will continue to
absorb as long as it becomes protonated as a
result of H2S co-absorption. Thus, it is conceivable that a particular sour water stream may be
a lot more concentrated than the solubility of
either ammonia or H2S by itself might suggest.
SWS uses either steam generated by a reboiler,
directly injected steam, or even a hot hydrocarbon stripping vapour to shift chemical reaction
equilibria by applying heat. Stripping vapour is
the gaseous solvent used to remove and carry
the ammonia and H2S out of the system. It functions by:
Heating the sour water feed to the boiling
point
Reversing chemical reactions
Diluting the partial pressure of the gases
stripped by furnishing excess vapour.
Figure 1 shows a typical SWS column with
heating by the injection of live steam and with
the possibility of injecting caustic soda to one of
the trays in the column. Typical energy usage in
PTQ Q4 2012 1
Cooler
Stripped
water
11
8
7
Sour
water
9
1
SWS
Feed/
bottoms
Pump-1
Figure 1 SWS using live steam and caustic injection
the stripping process is in the range 1.0-1.5 lb of
50 psig equivalent saturated steam per gallon of
sour water.
To minimise heat exchange surface, an external reboiler often uses higher pressure
(temperature) steam than is typical in an amine
regenerator because amine thermal degradation
is not a limiting factor. However, there is a practical limit of 400-450F, where coking heavy
hydrocarbons can lead to fouling and solids
deposition in the reboiler and, of course, corrosion is always a concern.
Stripped sour water specifications for NH3 and
H2S can be highly dependent on local requirements. Typical targets for NH3 are 30-80 ppmw
in the stripped water and undetectable to less
than 0.1 ppmw for H2S. Typical recent installations involve 35-45 actual trays.1,3
It is common in refinery cracking units (FCCs
and cokers) for the sour water generated to
contain organic and inorganic acid impurities
from HSS precursors and, just as for amine
units, ammonia partially in the protonated form.
It cannot be thermally regenerated because the
HSS responsible for the protonation is completely
non-volatile and cannot be removed by boiling it
into the stripping steam. In such cases, it is quite
common to inject a small amount of strong base
(NaOH) to shift the pH into a range where
ammonium ion (NH4+) shifts back to NH3. Spent
caustic from Merox-type units is commonly used
for this purpose, but care must be taken to
ensure that disposal of the spent caustic is not
completely reliant on this destination, or the tail
will begin to wag the dog.
2 PTQ Q4 2012
When adjusting the pH of the
water to spring ammonia chemically, the adjustment is usually
10
SWS AG
made by metered injection of
caustic onto a tray far enough
down the column that most of
the H2S has already been
stripped
out
and
NaOH
ammonia is the main remaining
injection
component. The metering rate is
normally controlled to a set
Live
steam
point on the pH measurement in
the stripped water after it has
been cooled. Caustic injection on
a lower tray generally works
better than injection directly
into the SWS feed itself because
the H2S concentration is already small on lower
trays. However, pH is extremely responsive to
caustic addition, so the measuring and control
elements should be as close together in time as
possible if rather large fluctuations in pH are to
be avoided. As we shall see, no more caustic than
is absolutely necessary should be injected
because excess caustic can permanently bind H2S
into the solution and eventually this will find its
way into biological treatment ponds, either
reducing the efficacy of the microbial population
or unnecessarily increasing the biological oxygen
demand.
Traditionally, SWSs have been modelled as a
series of equilibrium stages, with stage efficiencies being quoted anywhere in the range from
15% to 45%, that is ranging over a factor of three.
However, since the mid to late 1980s, the mass
transfer rate-based approach to simulating amine
contactors and extractive, azeotropic and reactive distillation has been in successful commercial
use. The extension to sour water stripping is a
natural progression and, in December 2011, the
ProTreat simulation package saw the addition of
a commercial mass transfer rate-based SWS
model. The remainder of this article uses this
model to explore how HSS affect the performance of SWSs and how caustic addition can be
tailored to provide the optimum amount of stripping of H2S and ammonia from HSS-laden sour
water.
Case study
Figure 1 shows the simplest possible configuration of a SWS with caustic injection. For this
www.digitalrefining.com/article/1000543
case study, the same stripper as outlined previously was used.4 It contained 40 one-pass valve
trays on 2ft spacing with 2-inch weirs. Sour
water was fed at 235F to tray 6 (from the top),
live steam saturated at 50 psig entered below the
bottom tray and, in all cases, the column was
sized for 70% of jet and downcomer flood.
Caustic could be injected on any tray in the
column. Table 1 gives the conditions of the sour
water used for this case study.
Sour water feed conditions
Total flow, lb/h
Temperature, F
Pressure, psia
Water, mol%
Hydrogen sulphide, mol%
Carbon dioxide, mol%
Ammonia, mol%
Formate, ppmw
Thiocyanate, ppmw
150 000
135
70
96.4
1.5
0.1
2.0
200
100
Table 1
Effect of HSSs on stripped water quality
H2S in stripped water, ppmwt
NH3 in stripped water, ppmwt
Figure 2 shows the effect of the stripping steam
rate and the absence versus presence of 300 indeed what happens, but for accurate answers
ppmw HSS (composition indicated in Table 1 on one must account for the pKa of the weak acids
the simulated stripped water quality with respect and bases involved, as well as considering the
to residual ammonia and H2S content). It is finite mass transfer rates of ammonia and H2S
readily apparent from the upper plot that the by using a genuine mass transfer rate model of
presence of HSS forces the
retention of a corresponding
10000
concentration of ammonia, and
0 HSS
no matter how much steam is
300 ppmwt
injected into the stripper there
1000
is a residual ammonia level that
simply cannot be removed when
there are HSS in the sour water.
100
HSS have a negative effect on
ammonia stripping. However,
they have a beneficial effect on
10
H2S removal because they are
generally stronger acids. A two
order-of-magnitude reduction in
1
0.75
1.00
1.25
1.50
residual H2S is possible with
Steam,
lb
/
Sour
water,
gallons
only a modest amount of HSS
present.
One way to improve ammonia
103
removal is to inject caustic soda,
0 HSS
102
a stronger base than ammonia,
300 ppmwt
10
onto a tray somewhere in the
1
column. At the risk of oversim1
plification, if the right amount is
10
added, the caustic will then bind
102
all the HSS that would other103
wise
be
neutralised
with
104
ammonia, without negatively
105
affecting H2S stripping. If too
106
much is used, the excess would
be expected to permanently
107
0.75
1.00
1.25
1.50
bind H2S; if too little, then less
Steam,
lb
/
Sour
water,
gallons
than the full potential for
ammonia release would be realised. In a broad sense, this is Figure 2 Effect of HSS on stripped water quality
www.digitalrefining.com/article/1000543
PTQ Q4 2012 3
optimal degree of improvement.
However, H2S removal is
1.48 lb/gal
improved only if caustic is
1.2 lb/gal
120
injected below tray 30. Indeed, if
caustic is not injected at all
100
(below tray 40 is the equivalent
80
of the tower sump and no stripping is assumed to take place
60
there) the best possible H2S
removal is obtained, but no
40
improvement to ammonia strip20
ping results. Interestingly, and in
contradiction to conventional
0
wisdom, caustic might as well be
10
20
30
40
injected right into the sour water
Caustic feed tray
feed itself, as there is no real
benefit from injecting it onto a
101
mid-tower tray, at least not in
this case study. The benefit to
102
H2S from injecting below tray 30
103
is probably because by that tray
most of the H2S has been
104
removed anyway, and there is
too little left in the water to be
105
held by the caustic. On the other
106
hand, the further below tray 30
one injects caustic, the fewer
107
1.48 lb/gal
stripping trays that are left to
1.2 lb/gal
remove the ammonia released;
108
0
20
30
40
thus, in this case, spiking the
Caustic feed tray
water with caustic is increasingly
pointless the further below tray
Figure 3 Effect of caustic injection on stripped water quality
30 one injects. We note in passing that some of this discussion
the tower and its trays. There are two questions becomes increasingly more academic as the H2S
to be answered: where to inject ammonia, and levels being considered fall below detectable limits
how much optimally to inject.
(<<1 ppmw).
H2S, ppmwt
NH3, ppmwt
140
Injecting neutralising caustic on various trays
Effect of caustic injection rate
As a first approximation, one might expect that
the optimal caustic injection rate will be just
sufficient to completely neutralise the HSSs in
the feed. This is a neutralisation ratio of NaOH
to HSS of unity. Less than unity is under-neutralised, greater than unity is over-neutralised.
Figure 3 shows the effect on the stripped water
residual ammonia and H2S levels of injection of
an exactly neutralising amount of caustic soda
onto various trays. Two live stripping steam rates
are considered.
As far as ammonia removal is concerned, caustic injection at or above tray 30 results in the
The effect of the extent of HSS neutralisation on
ammonia stripping is fairly clear cut. As the upper
plot of Figure 4 shows, caustic injection at a rate
just sufficient to neutralise the HSS springs the
maximum possible amount of ammonia, although
at low stripping steam rates, a slightly larger
amount helps. With HSS, however, the effect of
under- versus over-neutralisation is spread over a
wide range of neutralisation ratios. When the
neutralisation ratio reached 0.5, the H2S residual
had already started to rise and, by the point where
a ratio of about 1.25 was reached, the residual H2S
had attained its highest level. However, if exact
4 PTQ Q4 2012
www.digitalrefining.com/article/1000543
140
1.48 lb/gal
1.2 lb/gal
120
NH3, ppmwt
neutralisation is not exceeded by
much, quite satisfactory residual
H2S levels in the stripped water
can be achieved. Note, though,
that these results pertain only to
the conditions presented in this
case study they should not be
generalised; rather, the ProTreat
SWS model should be run for the
conditions pertinent to the existent installation.
100
80
60
40
20
Set point for pH control and optimal
stripping
0
0.0
0.5
1.0
1.5
2.0
H2S, ppmwt
When using caustic injection
Neutralisation ratio, NaOH/HSS
with phenolic sour waters, it is
common practice to control the
1
caustic flow to a pH set point.
1.48 lb/gal
The question is, what should the
1.2 lb/gal
1
10
set point value be? The answer
102
almost certainly depends on the
amounts and kinds of HSS
103
present and probably on the
104
levels of ammonia and H2S in
the sour water as well. For the
105
case being examined here, and
106
injection at tray 30, the stripped
107
water pH at 120F is shown as a
function of neutralisation ratio
108
in Figure 5 for two stripping
0
0.5
1.0
1.5
2.0
steam rates. If the target caustic
Neutralisation ratio, NaOH/HSS
injection rate is 100% neutralisation of the HSS, the treated Figure 4 Effect of extent of neutralisation on stripped water quality. Caustic
water pH at 120F is 9.0-9.1. If injection was onto tray 30 from the top of the column
the set point in actual use were
8.0, say, no caustic at all would be injected at a ising ratio of 1.0, essentially all the MDEA
steam rate of 1.2 lb/gal and only 70% neutralisa- reported to the stripped water. However, the
tion would be achieved at 1.48 lb/gal, leaving 35 higher consequent pH (9.48 versus 9.08) caused
ppmw ammonia in the stripped water rather less ammonia (5.5 versus 7.8 ppmw) but more
than the 5 or so ppmw actually achievable. The H2S (0.0000548 versus 0.0000235 ppmw) to
value of the pH set point to achieve the best remain in the stripped water.
ammonia removal without needlessly poor H2S
stripping is rather an important parameter. It Conclusion
does not seem to depend to any significant extent Perhaps the most important message here is that
on the stripping steam rate, but it is a parameter genuine mass transfer rate-based modelling
whose value can be predicted and verified accu- allows the construction of a virtual plant on a
rately only by using a genuine mass transfer rate computer. There is no reliance on what might be
model for SWS.
termed fudge factors to achieve agreement
An additional observation from the simulations between calculations and reality. The power of the
was that when 500 ppmw of MDEA was added ProTreat SWS model is the ability to analyse a
to the sour water (steam rate of 1.48 lb/gal) and given tower under specified conditions in such
the caustic was added to tray 30 with a neutral- detail that the optimal caustic injection rate and
www.digitalrefining.com/article/1000543
PTQ Q4 2012 1
Treated water pH at 120F
12
11
1.48 lb/gal
1.2 lb/gal
10
9
8
7
6
0.0
0.5
1.0
1.5
2.0
in higher pH and better ammonia stripping, but slightly poorer
H2S removal. All amine remains
in the stripped water.
In conclusion, this article has
revealed several aspects of stripping phenolic sour water that do
not seem to have been recognised
heretofore.
It
also
highlights the tremendous benefits that can be obtained from
simulating SWSs the right way
using a genuine mass transfer
rate model.
Neutralisation ratio, NaOH/HSS
ProTreat is a mark of Optimized Gas
Treating, Inc.
Figure 5 Setting the pH control point for optimal stripping
even the pH control point can be very accurately
assessed beforehand.
A mass transfer rate-based SWS model
provides a virtual SWS on a computer. Engineers
can design, analyse and optimise SWS units with
unprecedented accuracy and reliability. Plant
operations personnel can answer a wide range of
what-if questions to troubleshoot operations and
to optimise existing units with considerably more
confidence in the results. Some possibly valuable
lessons include:
The optimal caustic injection rate is primarily
dictated by the kind and concentration of HSS in
the sour water
No matter how much caustic is injected to
spring ammonia, it will have a deleterious effect
on stripping H2S
The tray selected for caustic injection is unimportant as long as it is far enough up the stripper
for there to be sufficient trays below the injection
point to strip the additional ammonia
As long as caustic is added to the sour water a
reasonable number of trays up from the bottom,
the precise tray for injection is immaterial to the
outcome. In other words, even adding caustic
directly to the feed water itself produces stripped
water of the same quality as obtained by injection
into the column. Some practitioners advise to add
caustic to the feed water; others say always to add
it to a tray. In fact, it makes no difference, so one
might as well save a feed nozzle
The presence of amine in the sour water results
6 PTQ Q4 2012
References
1 Asquith J, Moore A, Sour water processing balancing needs,
Proceedings of the 2000 Brimstone Sulphur Recovery Symposium,
Vail, CO.
2 Murphree E V, Graphical rectifying column calculations, Ind.
Eng. Chem., 17, 747-750 and 960-964, 1925.
3 Quinlan M P, Hati A A, Processing NH3 acid gas in a sulphur
recovery unit, Proceedings of the 2010 Laurance Reid Gas
Conditioning Conference, Norman, OK.
4 Weiland R H, Hatcher N A, Reliable design of sour water
strippers, PTQ, Q3 2012.
Ralph Weiland began working in gas treating in 1965. He taught
chemical engineering for 30 years at universities in Canada,
Australia and the US, and spent 10 years doing mass transfer
tray development with Koch-Glitsch LP. He established Optimized
Gas Treating in 1992. He holds BASc, MASc and PhD degrees in
chemical engineering from the University of Toronto.
Email: [email protected]
Nathan Hatcher joined Optimized Gas Treating, Buda, Texas,
as Vice-President, Technology Development, in 2009. He has
spent most of his career in amine treating and sulphur recovery.
He holds a BS in chemical engineering from the University of
Kansas and is currently a member of the Amine Best Practices
Group. Email: [email protected]
Links
More articles from: Optimized Gas Treating
More articles from the following category:
Corrosion/Fouling Control
Water/Wastewater Management
www.digitalrefining.com/article/1000543
