Two examples of steady state simulation with HYSYS at

GALPenergia Sines Refinery

Jos Egdio Fernandes Inverno1*, Eurico Correia2, Pablo Jimnez-Asenjo3 and
Josep A. Feliu3
1
Galpenergia, Refinaria de Sines, 7520-952 Sines, Portugal
2
Intituto Superior Tcnico, Lisboa, Portugal
3
Aspentech Inc, Passeig de Grcia 56, 08007 Barcelona, Spain

Abstract
This document shows two examples of aplication of a simulation software, in this case
HYSYS, to a real situation in an oil refinery: a crude destillation model and a sour
water stripper model.
Keywords: simulation software, oil refining, crude destillation, sour water stripping

1. Introduction
The purpose of this paper is to present the work developed at Sines Refinery in
cooperation with the University (Instituto Superior Tcnico, Lisbon) regarding the
development of models of some process units of GALPenergia SINES Refinery. The
work was done during six months and HYSYS was used as simulation tool.
Sines Refinery has been using this software for the last 4-5 years with success in solving
some small operational problems. The chalenge here was to model a complete process
unit in order to allow the optimization of the operation.

2. Crude destillation Unit

The crude destillation unit at Sines Refinery is the process unit where the crude oil is
first processed in the refinery and is composed by the following equipment:
2 pre-heat trains (each one composed by several heat exchangers, a desalter and a
furnace)
1 pre-flash drum
1 main destilation column (a very complex column with several pumparounds and side
strippers)
4 secundary fraccionation columns (Deetanizer, debutanizer, naphta splitter and a
deisopentanizer)

Author to whom correspondence should be adressed: jos.inverno@galpenergia.com

The original model was initially developed in close collaboration with Hyprotech (now
Aspentech) some years ago. The objective of the present works was to validate the
model, checking all the settings and calibrations, to verify the answer of the model to
real operating conditions.
The first step for a successfull simulation is a correct choose of the thermodynamic
method that will be used in the calculations of the state variables and the physical
properties. For the family of compounds that are present in the crude distillation unit
process streams, the Peng-Robinson equation of state is normally accepted as adequate.
The second step is to correctly characterize the feed composition. Refineries can be
feeded with a multiplicity of crude oils and usuallly the feed is not a crude of only one
origin but a crude mix that can vary from 3 to 10 diferent crude types. Simulation
packages are able to properly characterize petroleum fluids if data from laboratory
assays (distillation curves, light end analysis, bulk properties, etc.) is available. How to
do it if we dont have this type of data available?
Ready-to-use characterized oil crudes from a BP crudes database are available in
another commercial software (KBCs Petrofine with CAMS) leased by GALPenergia.
This software doesnt use rigorous thermodynamic models but is used to simulate all the
refinery for schedulling and blending proposes.The results of generating a crude slate
with this software is a huge table with all types of data (SG, S, V, Ni, UOPK,
viscosities, etc.) for 75 different components (22 pure light components and 53 heavier
pseudo components separated trough different cut points). All this data can be used to
feed the model developed in HYSYS and check the correct characterization of the
quality of the feed. However, using a database and not current crude data, has the
inherent problem of ignoring the properties variability of the real crude extracted from
fields. When observing mismatches between simulated and real plant data there will
always remain the doubt if the cause is a modelization mistake in the process or a
wrong (outdated) feed characterization.
A back-blending procedure would have been as well possible in case assays from the
distillation column products whould have been available. Using the distillation curves,
the bulk properties and the light end analysis obtained from the column products,
HYSYS could have back-generated the original crude entering the column by
leveraging the properties as a function of the products flows leaving the column.
In any case, using the crude database, we have obtained a good agreement between the
crude column simulation and the real data obtained from the process.
At the end of the paper, the Figure 1 shows the general arrangement of the unit
operations in the simulation of this unit.
In the Table 1 we can compare the results of the simulation with the real data obtained.
For the case study we have used the 10 of June 2003 operating data. In this day the
refinery treated a crude mix composed by 4 different crudes (Es Sider, Qua Iboe, Sahran
Blend and Marlin). We can compare the column temperature profile and also some
products properties (SG Specific gravity and 5 an 95 v/v % ASTM D86 distillation
temperatures). From the table we can see that we have obtained a good agreement
between the crude column simulation and the real data obtained from the process.

Table 1 Differences between the real operation and the simulation.

Real

Simulation

(1)

(2)

C
C
C
C
C
C
C
C
C
C
C
C
C

120.7
137.7
189.9
203.0
245.1
265.6
300.6
311.7
348.8
343.7
-72.21
-91.05
-73.01

120.3
140.8
190.5
213.0
243.2
272.8
294.7
319.7
355.0
351.6
-68.7
-97.7
-71.4

0%
-2%
0%
-5%
1%
-3%
2%
-3%
-2%
-2%
5%
-7%
2%

C
kg/m3
C
C
kg/m3
C
C
kg/m3
C
kg/m3

163.7
717.6
310.5
241.2
848.4
376.5
306.5
883.3
339.6
978.8

161.5
718.7
325.0
235.5
848.1
384.0
296.2
884.3
348.5
977.1

1%
0%
-5%
2%
0%
-2%
3%
0%
-3%
0%

Units

(1)-(2)

Column temperature profile

Overhead Temp
Tray 03 Temp
Tray 15 Temp
Tray 19 Temp
Tray 26 Temp
Tray 30 Temp
Tray 35 Temp
Tray 39 Temp
Tray 43 Temp
Bottoms Temp
T Pumparound Naphtha
T Pumparound LGO
T Pumparound HGO
Products properties
Naphtha 95%
Naphtha SG
Light Gasoil 95%
Light Gasoil 5%
Light Gasoil SG
Heavy Gasoil 95%
Heavy Gasoil 5%
Heavy Gasoil SG
Atmospheric Residue 5%
Atmospheric Residue SG

3 . Sour Water stripper

In many refinery units water or steam is injected to facilitate the process. This water is
usually separated and collected in the boots of the overhead receivers of the destillation
columns or in other type of receivers where the separation between water and
hydrocarbons can be enhanced.The drawback of injecting water into the refinery
processes is that it strips the acid soluble gases, NH3 and H2S, generating sour water
streams
The objective of the sour water stripper is to reuse the recovered water in other refinery
processes and to remove almost all the H2S and NH3 present, mainly because
environmental reasons.

If by any reason the water cannot be reused, and these two gases are not substantially
removed the sour water needs to be sent to a wastewater treatment plant where the
treatment to remove this type of contaminants can be finally performed but with a very
high cost.
With the information that in the existing installation the NH3 removal was not effective,
we used the simulation software to simulate the column (Sour Water Stripper) to find
out what would be the best operating conditions in terms of column temperature,
pressure, and stripping steam in order to maximize the removal of NH3 from water.
For this simulation, and because the type of compounds present is not the same than in
the crude distillation column, it is necessary to use another thermodynamic method. For
sour systems, due to the presence of H2S and NH3 there are some adequate
thermodynamic methods present in HYSYS database (the Sour options of the SRK and
PR equations of state as well as the Stryjek Vera modification of the PR). Finally, for
data availability, it was decided to use the Peng Robinson Stryjek Vera [PRSV]
equation of state
The tests performed using the column model in HYSYS have concluded that to have a
better NH3 removal the operating temperatures should be increased, pressures decreased
and the stripping steam flow optimised.
Graphs 1 and 2 show the relationship between the pressure, temperature and steam flow
in the level of NH3 in the stripped water.

160

Overhead P: 2bar

140

Overhead P: 2.5bar

137

Ammonia (ppm)

120
100
79

80
60
51

40

45
33

26

21

20

15
10

9
4

0
6

6.5

7.5

8.5

Steam flow (t/h)

Graph 1- Ammonia in stripped water as a function of stripping steam flow

120

Overhead P: 2bar

114

Overhead P: 2.5bar

98

100

Ammonia (ppm)

84

80
71
61

60

53

40

40

34
29

26

23
21

20

0
75

80

85

90

95

100

105

Column overhead temperature (C)

Graph 2 - Ammonia in treated water as a function of the overhead temperature

The real plant column has been adjusted using the results obtained trough HYSYS
simulation. The increase in column performance obtained has supposed company
savings both in raw water consumption and in the waste water treatment plant.
Figure 2 shows the general arrangement of the unit operations in the simulation of this
unit

4. Conclusions
Simulation software is a very good tool for the process industry, not only at the level of
conceptual design but also during the entire lifecycle of the equipment, were it can be
very useful for performance, debottlenecking and process studies.
The presented results have been obtained using the steady state version of the software,
but all the simulations done in steady state can have an evolution to a dynamic
simulation (for example to build process simulators for operator training or to study the
behaviour of the process units in transient conditions), or to a real time optimisation
system where, together with the advanced process control tools, can be very profitable
in the optimisation of the operation in real time.

Figure 1 Crude distillation column diagram

Figure 2 Sour water stripper diagram