Optimization of an Existing Habshan II Amine

Sweetening Unit
A. Alkasem; Y.A. Al Zarouni, Abu Dhabi Gas Industries Ltd - GASCO; J.C. Slagle, Bryan Research &
Engineering, Inc.; A.S. Berrouk; D. Satyadileep, The Petroleum Institute
Copyright 2015, Society of Petroleum Engineers
This paper was prepared for presentation at the Abu Dhabi International Petroleum Exhibition and Conference held in Abu Dhabi, UAE, 912 November 2015.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
As part of its ongoing optimization effort, Abu Dhabi Gas Industries (GASCO) is working with the Petroleum
Institute (PI) and Bryan Research and Engineering Inc. (BR&E) to identify opportunities for optimization of
the Habshan II amine sweetening unit for a wide range of gas throughput.
In a previous study, ProMax, a process simulation package, was first verified by comparing the model results
to operating data for about 300 days and was then used to optimize the facility operating near maximum
capacity. Recently, most of the gas has been diverted to another facility resulting in Habshan II operating in
the 30-40% capacity range.
In the present study, ProMax, is used to re-optimize the facility at the current low throughput. The model
results for the new optimum are implemented in the plant and compared to operating data to confirm the
predictions. In addition, the plant plans to increase throughput in the coming months. Therefore, the plant has
been optimized at various gas flowrates to provide operators with set points at any given gas flowrate. In
addition to preparing operators for throughput fluctuations, the results show a reduction of operating costs that
amount to roughly 800,000 USD/yr.

SPE-Error! Reference source not found.-MS

Introduction
GASCO operates several methyl diethanolamine (MDEA) units, such as the Habshan II unit shown in Figure
1. Unlike primary and secondary amines, MDEA is commonly used for selective removal of H 2S while slipping
some CO2 into the sweet gas.
Sweet Gas
Acid Gas
Pump
Water Wash

Condenser

Air Cooler
1

Absorber

Flash Absorber

Sour Feed

Stripper

Flash Gas

21

Rich Amine
Rich Flash

25

Reboiler

Lean/Rich Exchanger
Lean Amine
Steam

Figure 1: Typical GASCO MDEA Sweetening Unit

The Habshan II amine sweetening unit may treat up to 500 MMSCFD of sour gas containing roughly 12%
acid gas. The plant has been optimized to meet the sweet gas specifications of 9 ppm H2S and 2% CO2 at 230
MMSCFD, as described in a previous work (1). However, the plant conditions have changed and the sweet
gas specification has been updated. Now the plant is meeting a sweet gas specification of 20 ppm H 2S and
3.3% CO2 while operating near 200 MMSCFD.
Conveniently, the model infrastructure is already in place to optimize the plant once again. This work addresses
each of the variables optimized previously and updates the optimized case to represent current and future
operations. The results from the ProMax(2) model are applied in the plant and observed.

Model Validation
Before optimizing the plant, it is important to have confidence the model is predictive and robust. If the model
is able to predict the plant performance, it should be able to match past operating conditions. A robust model
will be able to accurately represent a wide range of values. To validate the model, data from 2013-2015 were
modeled using the ProMax Scenario Tool. If the model accurately represents the data from these dates, it can
be said the model is predictive and robust due to the wide range of operating conditions.
There are two distinct ranges for data. The first 300 days are from 2013 when the plant had relatively high
flow. The average results for the first 300 days are shown in Table 1.

SPE-Error! Reference source not found.-MS

Table 1: Average Results First 300 Days at High Gas Flow

Plant Data
ProMax
H2S Sweet Gas (ppm)
10.93
10.92
CO2 Sweet Gas (mol%)
1.72
1.67
The next 180 days are from 2014 and 2015 when the plant reduced its gas flow to less than half the previous
300 days. The average results for the last 180 days are shown in Table 2.
Table 2: Average Results Last 180 Days at Low Gas Flow
Plant Data
ProMax
H2S Sweet Gas (ppm)
13.8
12.6
CO2 Sweet Gas (mol%)
0.75
0.72
The model matches the data very well at both high and low gas flow rates. Therefore, the ProMax model may
be seen as predictive and robust, as confirmed in many other studies and publications (1) (3) (4) (5) (6).

Optimization
The optimization of this plant starts with the previous work completed at higher gas flow rates. The variables
used to optimize the plant previously are seen in Table 3.
Table 3: Summary of Optimization Study at High Gas Flow
Reduction in Steam
Optimization Variable
Rate
Reduction in Solvent Circulation Rate
10 tonnes/hr
Increasing MDEA Concentration
2.5 tonnes/hr
Increasing Regenerator Inlet T and P
3 tonnes/hr
Since this study is focusing solely on optimizing the operating costs, optimization that includes capital
expenditures are avoided. It is important to note that while the lean amine temperature is often a variable to
optimize, Abu Dhabis ambient temperature is over 40 C for more than 40% of the year. Therefore, as with
most amine sweetening plants in the Middle East, the lean amine temperature is as low as an air cooler can
achieve. For this reason, the lean amine temperature is not included in the optimization.
Upon deeper evaluation, the inlet temperature and pressure to the regenerator are constrained due to the plate
and frame lean/rich heat exchanger design. Therefore, the latest study considers these variables constant, to be
revisited at a later date when capital cost expendetures are evaluated.
The optimization of the plant is now primariliy considering the solvent concentration, circulation rate and
steam consumption.
Currently, the plant is operating near 200 MMSCFD. The model shows good representation of the plant data
at these conditions, as shown in Table 4.
Table 4: Plant Data Comparison at Current Operating Conditions
Plant Data
ProMax
H2S Sweet Gas (ppm)
17.8
16.3
CO2 Sweet Gas (mol%)
1.4
1.5

SPE-Error! Reference source not found.-MS

Using methods described in previous work, the plant is optimized (1) (7). As demonstrated in the previous
work, CO2 absorption decreases as the MDEA concentration increases. Due to project constraints, the
maximum MDEA concentration the operators are able to achieve is 48 wt% in this plant. Therefore, the MDEA
concentration is set at 48% to maximize CO2 slip into the sweet gas.
The model will calculate the solvent circulation rate and steam rate based on maintaining less than 3.3% CO2
and 20 ppm H2S in the sweet gas. The optimized case is compared to the plant operating conditions prior to
optimization in Table 5.
Table 5: Plant Operating Conditions Prior to Optimization Compared to Optimized Case
Plant Data
Optimized
H2S Sweet Gas (ppm)
17.8
18
CO2 Sweet Gas (mol%)
1.4
3.3
3
Circulation Rate (m /h)
510
350
MDEA Concentration (wt%)
44
48
Reboiler Steam Rate (tonnes/h)
58
44

The steam rate may be reduced by roughly 25%, or 14 tonnes/hr, when compared to the plants current steam
consumption. To do so, CO2 slip is maximized by reducing the solvent circulation rate. If this level of steam
reduction is maintained throughout the year, the optimization may amount to a savings of roughly 800,000
USD/yr (8).

Application
While it is beneficial to perform the above exercise, the operators see a wide range of inlet gas flowrates.
Therefore, the optimization needs to be done at various throughputs. To optimize the plant in a way that is easy
for an operator to control, the model is executed at hundreds of conditions with constraints placed on the
process to always maintain less than 20 ppm H2S and less than 3.3% CO2. The cases traverse operating
conditions between 150 and 450 MMSCFD at step sizes of 5 MMSCFD.
With the optimized cases modeled at various plant flowrates, the operator finds the current gas throughput in
a table or graph and applies the changes the ProMax model proposes for the plant. This strategy is followed in
the plant to establish its viability. The operator graph for this plant is shown in Figure 2. The operator only
needs to find the current inlet gas flowrate on the graph to find the corresponding solvent circulation rate and
steam rate needed to maintain optimum performance.

120

1000

110

900

100

800
700

90

600

80

500
70

400

60

Steam Rate

300

50

200

Circulation Rate
40

Solvent Circulation Rate (m3/h)

Steam Rate (tonne/h)

SPE-Error! Reference source not found.-MS

100

30

0
125

175

225

275

325

375

425

475

Inlet Gas Flow (MMSCFD)

Figure 2: Operators Reference Guide to Find Steam Rate and Solvent Circulation Rate
On August 4th, 2015, the plant implemented the set points created by the model. After about 10 hours, the plant
was able to achieve the desired goals and the results agreed very well with the model. In fact, the reboiler duty
was reduced in this period by an average of 12%, all while the plants throughput increased by 10%.
Not only is the plant able to decrease the reboiler duty, but it is also able to increase throughput.
The application of the plants optimum is summarized in Table 6.
Table 6: Summary of Plant Savings
Before Optimization

Optimized

Plant Data Average Sweet Gas CO2 (mol%)

1.23

2.93

Plant Data Average Sweet Gas H2S (ppm)

8.9

15

Plant Data Average Circulation Rate (m /h)

515

419

Plant Data Average Reboiler Steam Rate (tonnes/h)

61

53

To see the potential savings, past performance of the plant is studied. When evaluating the savings over time,
roughly 500 days of operations are compared to the optimum values found by the model. The plant data before
and after optimization are shown in Figure 3.

SPE-Error! Reference source not found.-MS

120
110

Steam Rate (tonne/h)

100
90
80
70
60

Optimum

50

Plant Data After Optimization

40

Plant Data Before Optimization

30
125

175

225

275

325

375

425

475

Inlet Gas Flow (MMSCFD)

Figure 3: Plant Data Comparison Before and After Optimization

As shown above, this methodology leads to consistently optimum conditions for the plant. While more data
will be collected as this methodology is used through the remaining life of the plant, the 14 data points collected
after the optimization was applied in the plant show near perfect results. In contrast, only 7% of the data points
collected before optimization fall along the optimum line.
If the set points generated by the model had been applied for these 500 days, an average reduction of steam of
28% would likely have been realized, a savings of 1.1 MM USD.
Therefore, applying this methodology through the life of the plant could save roughly 25 MM USD.

Conclusion
The Habshan II amine sweetening plant has experienced large fluctuations in gas throughput and is expected
to see large fluctuations again in the future. The optimization applied through this study has helped the
operators navigate large changes in the plant conditions by rigorously modeling the plant with ProMax at all
the possible plant throughputs. The operators are now adequately prepared to run the plant at optimum
conditions at any sour gas flowrate.
In addition, the operators are able to realize large savings in the plant. The next step is to apply this optimization
for all of GASCOs amine sweetening units. If similar savings are found, GASCO may save close to 14 MM
USD every year.

References
1. Optimization of the Habshan II Amine Sweetening Unit. Satyadileep, Dara, Berrouk, Abdallah and
Slagle, Justin C. Abu Dhabi : Sour Oil & Gas Advanced Technology, 2015.
2. Bryan Research & Engineering, Inc. ProMax 4.0. Bryan, Texas, United States of America : s.n.,
2015.
3. Simulation of the Benfiels HiPure Process of Natural Gas Sweetening for LNG Production and
Evaluation of Alternatives. Ochieng, R, et al. Abu Dhabi : s.n., 2012. Proceedings of Sour Oil and Gas

SPE-Error! Reference source not found.-MS

Advanced Technology.
4. Comparison of Ideal Stage and Mass Transfer Models for Separation Processes With and Without
Chemical Reactions. Skowlund, Christopher, et al. New Orleans : s.n., 2012. Proceedings of Gas
Processesors Association.
5. Converting to DEA/MDEA Mix Ups Sweetening Capacity. Spears, Michael L., et al. 1996, Oil &
Gas Journal August 12, pp. 63-67.
6. Why Not Optimize Your Amine Sweetening Unit? Bullin, Jerry A. October 2003, Hydrocarbon
Engineering.
7. Improve Your Gas Plant's Performance in the Middle East. Slagle, Justin C. Dubai : Gas Processor's
Association GCC, 2013.
8. US Energy Information Administration. [Online] http://www.eia.gov/naturalgas/weekly/.