You are on page 1of 10

Visbreaking Unit Simulation Model for the Prediction of

Process Performance and Visbreaker Residue Stability


Sara Sousaa, Filipa Ribeiroa, Ana Rita Costab
a
Departamento de Engenharia Química, Técnico Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisboa, Portugal
b
Galp, Rua Tomás da Fonseca - Torre C, 1600-209 Lisboa, Portugal
Sulfur content in bunker fuels has been reduced in order to decrease emissions of sulfur oxides into
the atmosphere. In January 2020, the sulfur content in bunker fuels will be limited to 0.5%wt. The
production of this fuel requires processing sweet crudes in Galp refineries. This is a challenging situation
because these crudes have a higher tendency to instability.
In this work, a rigorous simulation model of the visbreaking unit was developed, using the Petro-
-Sim™ software, which allows a good prediction of yields, product properties, including visbreaker
residue stability. This is the main component of fuel.
The simulation model was solved through two methods for maximum visbreaker conversion
calculation, KBC 1985 and KBC 2005. KBC method 1985 presents a good prediction of visbreaker
residue stability. For KBC 2005 method, simulation model predictions are more dependent on feedstock
quality and more representative of reality.
A delta-base vector structure was created and allows the implementation of the data generated by the
simulation model in Galp linear programming model. This structure was validated using a linear
representation model. The implementation in the linear programming model of the visbreaker unit
simulation model will allow an efficient and optimized selection of crudes that will maximize the refining
margin, ensuring the production of low sulfur fuel oil.

KEYWORDS: Visbreaker, Petro-Sim™, Delta-Base Vectors, Thermal Cracking, Visbreaker Residue


Stability, Refining

Fuel oil produced in Galp refineries can be


1 Introduction composed by atmospheric, vacuum and
visbreaker residues. The visbreaker residue
Fuel oil, used as a bunker fuel, causes large (RVB) is the most commonly used in the
emissions of SOx (Sulfur Oxides) into the production of fuel oil, since its use brings an
atmosphere, which are harmful to human health economic advantage.
and to the environment.
IMO (International Maritime Organization), a
1.1 Objectives
specialized agency of the United Nations, is the
This work addresses the development of a
global standard-setting authority for the safety,
simulation model of Sines visbreaking unit in
security and environmental performance of
Petro-Sim™ software. This model will predict
international shipping. In order to reduce SOx
unit performance and RVB stability depending
emissions from ships, the sulfur content limit in
on the type of the unit feedstock. This model will
fuel oil has been progressively reduced. As
be implemented in Galp linear programming
from January 2020, there will be a substantial
(LP) model in order to select the crudes that
cut in the sulfur content limit, from 3.5%wt (High
Sulfur Fuel Oil – HSFO) to 0.5%wt (Very Low maximize refining margin ensuring the
Sulfur Fuel Oil – VLSFO) [1]. production of VLSFO.
In order to meet the new IMO regulations,
ships can limit the air pollutants installing 2 Visbreaking Unit
exhaust gas cleaning systems, also known as
scrubbers. 2.1 Process Description
The alternative that should be adopted by The refining process starts by feeding the
Galp to supply the VLSFO market consists in crude oil, or usually a mixture of crudes (crude
mixing residues produced from sweet crudes mix), into an atmospheric distillation unit. Here
(sulfur content below 1%wt) with other crude are separated into several fractions
components with very low sulfur content (cutter (defined by cut point – the temperature on the
stocks). The production of VLSFO is a major whole crude TBP (true boiling point) curve that
challenge due to the fuel high tendency for represents the upper and lower limits [2]).
instability as well as sediments and coke In Sines refinery, atmospheric distillation has
generation. a top gas stream, a naphtha stream, a kerosene

1
stream, two streams of gasoil (light gasoil and severity depends on the operating temperature
heavy gasoil) and a bottom stream – and residence time.
atmospheric residue (RAT). The RAT yield Variations in feedstock quality will have
depends considerably on the crude mix fed to impact in the conversion level obtained at a
the atmospheric distillation unit. The heavier the given severity [5]. Thus, the visbrekaer
crude oil mixture fed, the higher the RAT yield. operations severity is generally limited by the
RAT is then fed to a vacuum distillation unit, visbroken product stability.
where is produced a vacuum distillate stream,
a vacuum gas oil (VGO) and a vacuum residue
3 RVB Stability
(RV) stream. RV has high viscosity, which
decreases its commercial value. Therefore, RV
is fed to a visbreaking unit. A fuel is stable if there is no asphaltenes
The main purpose of visbreaking unit (VB) is flocculation. On the other hand, instability may
to convert heavy, high viscosity feedstocks, like be irreversible, i.e. precipitated asphaltenes
RV, to lower viscosity products suitable for use may not be re-dissolved [6]. To ensure fuel oil
in fuel oil. This reduction is obtained by stability it is necessary that RVB is also stable.
decomposing heavy molecules into lighter Generally, cutter stocks are paraffinic (gasoil
molecules through thermal cracking reactions. and kerosene) and, therefore, RVB stability
Since VB operates at low conversions (low degree must be high enough to prevent
extensions of thermal cracking reactions), its asphaltene flocculation during blending with
main product is visbreaker residue (RVB), with cutter stocks. RVB stability is controlled by
a yield of about 80-90%wt. Another benefit from performing a peptizing-value (p-value) test [7].
the visbreaking operation is the production of The p-value provides the peptization state of
gas, naphta and gas oil (GOVB) streams that the asphaltenes in residues. A residue is then
usually have higher product values than RVB considered stable for a p-value greater than
[3]. 1.2. The p-value is obtained by adding cetane
Sines refinery has a soaker visbreaker, (paraffinic compound, once paraffins flocculate
where the bulk of the cracking reactions occurs the asphaltenes) to the sample under study
not in the furnace but in a soaker drum until the asphaltenes begin to flocculate. This
downstream the furnace. In the soaker drum property is determined by equation (1), where
the heated feedstock is held at high Xmin is the critical dilution in cetane, i.e. the
temperature for a predetermined period of time amount of cetane (ml) that can dilute 1g of the
allowing cracking to occur. Then soaker effluent sample until asphaltenes flocculation begin.
goes to a fractionator column where RVB is This method has a high repeatability, about
separated from lighter fractions [3]. 0.07 [7].

2.2 Process Reactions


𝑉𝑎𝑙𝑜𝑟 − 𝑝 = 1 + 𝑋𝑚𝑖𝑛 (3)
Thermal cracking is the decomposition of
hydrocarbons at elevated temperatures,
resulting in the formation of lower molecular There are methods of evaluating the stability
weight products and is a first-order reaction [4]. of fuel using other solvents. However, the
p-value is a widely used stability indicator and
𝑣 =𝑘∙𝑐 (1) is also used in Galp refineries.

𝑎 3.1 Asphaltenes
− ln ( )=𝑘∙𝑡 (2) Asphaltenes are defined as the insoluble
𝑎−𝑥
fraction in n-heptane and soluble in toluene and
represent the heavier and more complex
During thermal cracking, the saturated
fraction of crude oil [8] [9]. Asphaltenes are high
(paraffinic) compounds are transformed into
polarity compounds composed by a very
saturated compounds of lower molecular
condensed aromatic and naphthenic structure.
weight. In high severity conditions (high
Asphaltenes present short paraffinic side
temperature) polymerization reactions are
chains and also heteroatoms, such as sulfur,
favored leading to resins and asphaltenes
nitrogen, oxygen and metals. The large number
cracking and to coke formation.
of aromatic rings gives to asphaltenes a flat
structure [10].
2.3 Operating Conditions
The extent of cracking reactions, i.e. GOVB
conversion, and RV visbreaking are controlled
by the severity of the unit. The visbreaker

2
4 Visbreaker Unit Simulation Model 4.4 Simulation Model Construction
and Calibration
In this work it was developed a rigorous The construction of VB unit simulation model
simulation model of the VB unit in Petro-Sim ™ requires real unit design and operation data.
software, version 7.0. The objective of calibration is to ensure that
The visbreaker simulation model currently model represents the real unit performance in
used by Galp is a simple planning model that terms of yields and product properties. In
calculates yields and product properties calibration the model automatically calculates a
according to the average reality of the unit's set of calibration factors to achieve the best
operating cycle. This model does not predict the match between the predicted properties values
RVB stability produced. and the real properties values.
For the VB unit simulation model, In order to ensure that the simulation model
Petro-Sim ™ has VIS-SIM ™ technology that mimics reality, the data used for calibration
simulates the thermal cracking, based on should be representative of the visbreaking
reaction kinetics. operation.
The data required for model calibration are
4.1 Maximum Visbreaker Conversion real operating conditions, product yields and
Maximum Visbreaker Conversion (MVC), is feed and products characterization in terms of
a feedstock characterization parameter and density, viscosity, distillation curve, sulfur
depends on the asphaltene content and the content, RVB p-value, wax and asphaltene
nature of the feedstock (aromatic or paraffinic). content in the feed stream. These data are
The MVC is defined as the conversion at which obtained through laboratory analyzes of
a visbroken residue is produced at the stability samples collected directly from the unit running
limit (ABN = 46, p-value = 1.15). Conversions in a steady state.
above the MVC value do not guarantee a stable The days selected were November 11, 2017
RVB. Petro-Sim™ allows the calculation of this (calibration 11), March 1, 2018 (calibration 1),
property using two methods: KBC method 1985 and March 8, 2018 (calibration 8). In calibration
(KBC 1985) and KBC method 2005 (KBC procedure it was used the three data sets
2005). corresponding to the three different operating
By the KBC 1985 method the Petro-Sim™ days.
model calculates the MVC taking into account In order to choose the calibration that best
RV asphaltenes content and unit feed BMCI represents the unit real performance,
(Bureau of Mines Correlation Index – fuel predictions of the three days were made using
aromaticity indicator [11]). each one of the three calibrations.
KBC 2005 is a more accurate method that
takes into account RV asphaltenes and wax 5 Linear Programming Model
contents.
The LP model is a fundamental tool in
4.2 Severity planning refining activity. The LP model
Petro-Sim™ calculates severity using the determines monthly the refineries production,
equation ( 4 ). the raw materials to be processed and the
𝐺𝑂𝑉𝐵 𝐶𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 operating conditions of the process units that
𝑆𝑒𝑣𝑒𝑟𝑖𝑡𝑦 = (4) allow the optimization of the refining margin.
𝑀𝑉𝐶 The LP model is also used by Galp to carry out
The RVB stability is ensured if severity value economic studies and to plan annual budgets.
is less than 100%. Galp’s LP model uses the software GRTPMS
(Generalized Refining Transportation
4.3 P-Value Marketing Planning System).
Petro-Sim ™ calculates the p-value based
on the ABN (Aromatic Blending 5.1 Delta-Base Structure
Number wich is a fuel aromaticity and stability The implementation of VB unit model
indicator [12]) value. P-value is given through simulation data in the LP model is performed
equation ( 5 ). through a delta-base structure. The delta-base
structure determines the impact that linear
𝐴𝐵𝑁 variations on feedstock properties have on a
𝑉𝑎𝑙𝑜𝑟 − 𝑝 = (5)
40 given product property or yield. Thus, linear
variations (deltas) are applied to feedstock
properties and operating conditions (vectors)

3
and the effect of these variations on the 12
performance of the unit is observed.

GOVB Yield (%wt)


10
It is required to do a sensitivity analysis in
order to select the vectors, i.e. the feedstock 8
properties and operation conditions, that affect 6
the most the VB unit performance. In this 4
analysis, deltas are defined to ensure that they 11-11-2018 01-03-2018 08-03-2018
represent linear impacts on products yields and
properties. Calib11 Calib1 Calib8 Real
A case, that corresponds to a feedstock and
products yields and properties is then defined. Figure 2 – Prediction of visbreaker gasoil yield
This case has to be representative of the VB with each of the three calibrations using the
maximum visbreaker conversion calculation method
unit operation. Deltas are applied to each vector
KBC 1985.
of this case, which produce a linear impact on
the yields and properties of the products. The
case chosen will serve as the basis for the 1.04
implementation of the delta base structure.

RVB Specific Gravity


The impacts that deltas caused in each 1.02
vector on the properties and yields of the 1.00

(g/cm3)
products of the process unit are generated in
0.98
the simulation model in Petro-Sim™ using the
Linear Programming Utility (LPU) tool and are 0.96
exported to a delta-base template file which will 11-11-2018 01-03-2018 08-03-2018
be provided to LP. Calib11 Calib1 Calib8 Real

5.2 MRL Predictions Figure 3 – Prediction of visbreaker residue


In order to evaluate the delta-base, that will specific gravity with each of the three calibrations
be implemented in LP, model a linear using the maximum visbreaker conversion
representation model (MRL) is developed in calculation method KBC 1985.
Excel using the data generated by LPU.
0.850
6 Results 0.840
Gravity (g/cm3)
GOVB Specific

6.1 KBC 1985 Method 0.830


6.1.1 Calibration Results 0.820
Figure 1 to Figure 7 compare the real yields
0.810
and properties with predicted results for each
11-11-2018 01-03-2018 08-03-2018
model calibration corresponding to the three
different days (November 11, March 1 and 8). Calib11 Calib1 Calib8 Real

Figure 4 – Prediction of visbreaker gasoil


95 specific gravity with each of the three calibrations
RVB Yield (%wt)

using the maximum visbreaker conversion


90 calculation method KBC 1985.
85

80 400
RVB Viscosity 100oC

11-11-2018 01-03-2018 08-03-2018 300

Calib11 Calib1 Calib8 Real 200


(cSt)

Figure 1 – Prediction of visbreaker residue yield 100


with each of the three calibrations using the 0
maximum visbreaker conversion calculation method 11-11-2018 01-03-2018 08-03-2018
KBC 1985.
Calib11 Calib1 Calib8 Real

Figure 5 – Prediction of visbreaker residue


viscosity 100 with each of the three calibrations
using the maximum visbreaker conversion
calculation method KBC 1985.

4
1.8 model and the rigorous simulation model
developed in this work.
RVB P-Value
1.6
1.4 92

RVB Yield (%wt)


1.2 90
88
1.0
86
11-11-2018 01-03-2018 08-03-2018
84
Calib11 Calib1 Calib8 Real
82
Figure 6 – Prediction of visbreaker residue
p-value with each of the three calibrations using the
maximum visbreaker conversion calculation method
KBC 1985.
Actual Prediction Real
Figure 8 – Prediction of visbreaker residue yield
30 for different days of operation with the actual model
and with the new model using the maximum
Isoconversion (%wt)

20 visbreaker conversion calculation method KBC


1985.
10

0 12
11-11-2018 01-03-2018 08-03-2018 GOVB Yield (%wt)
10
Calib11 Calib1 Calib8 Real
8

Figure 7 – Prediction of isoconversion with 6


each of the three calibrations using the maximum
visbreaker conversion calculation method KBC
1985.

Actual Prediction Real


As can be seen from the results presented Figure 9 – Prediction of visbreaker gasoil yield
above, calibration 8 presents prediction values for different days of operation with the actual model
with the greatest deviation from reality. and with the new model using the maximum
Calibrations 1 and 11 present good visbreaker conversion calculation method KBC
predictions with similar deviations, 1985.
comparatively to the reality. However, on
11-11-2017 is processed a crude mix with
Isthmus crude oil, which has a high asphaltenes 1.05
1.04
RVB Specific Gravity

content. Therefore, in order to produce a stable


RVB, the VB unit has to operate with a severity 1.03
lower than usual. As this is a less typical 1.02
(g/cm3)

situation, the calibration factors selected 1.01


correspond to calibration 1 since in this day it 1.00
0.99
was processed a more common crude mix.

6.1.2 Crude Mix Predictions


In order to validate the model, it is important
to check model performance, not only from real Actual Prediction Real
RV characterized in the lab, but also when
processing VB unit feed obtained through Figure 10 – Prediction of visbreaker residue
specific gravity for different days of operation with
simulation of the crude mix. These feedstocks
the actual model and with the new model using the
are obtained simulating RAT and RV streams maximum visbreaker conversion calculation method
from crude assays fed to simulation models. KBC 1985.
The LP model works this way.
Figure 8 to Figure 14 present the real and
predicted yields and product properties. All
figures show the prediction using the actual

5
0.85 25

Isoconversion (%wt)
23
GOVB Specific Gravity
0.84
21
0.83
19
(g/cm3)

0.82 17
0.81 15

Actual Prediction Real


Actual Prediction Real
Figure 14 – Prediction of isoconversion for
Figure 11 – Prediction of visbreaker gasoil
different days of operation with the actual model
specific gravity for different days of operation with
and with the new model using the maximum
the actual model and with the new model using the
visbreaker conversion calculation method KBC
maximum visbreaker conversion calculation method
1985.
KBC 1985.

By analyzing the predictions from the crude


600
mix it is possible to verify that the new
RVB Viscosity 100oC

400 simulation model built presents much more


rigorous predictions than the current model,
200 mainly for the yields and isoconversion, i.e. the
(cSt)

new simulation model is sensitive to the


0 feedstock load to the unit. The new model is
also able to predict the stability of RVB, which
is crucial for the production of VLSFO.

Actual Prediction Real 6.1.3 MRL predictions


The implementation of Petro-Sim™
Figure 12 – Prediction of visbreaker residue simulation model in LP model was validated
viscosity 100C for different days of operation with using MRL.
the actual model and with the new model using the Figure 15 to Figure 20 compare the MRL
maximum visbreaker conversion calculation method predictions with the simulation model
KBC 1985. predictions in Petro-Sim™. The figures present
yields and product properties predictions and
real values.
1.7 92
RVB Yield (%wt)
RVB P-Value

1.5 90
1.3 88
86
1.1
84
0.9 82

Actual Prediction Real PS Prediction MRL Real


Figure 13 – Prediction of visbreaker residue p- Figure 15 – Prediction of visbreaker residue
value for different days of operation with the actual yield by Petro-Sim™ model and MRL for different
model and with the new model using the maximum days of operation using the maximum visbreaker
visbreaker conversion calculation method KBC conversion calculation method KBC 1985.
1985.

6
12 1.7
GOVB Yield (%wt) 10 1.5

RVB P-Value
1.3
8
1.1
6
0.9

PS Prediction MRL Real


PS Prediction MRL Real
Figure 16 – Prediction of visbreaker gasoil yield
by Petro-Sim™ model and MRL for different days of Figure 19 – Prediction of visbreaker residue
operation using the maximum visbreaker p-value by Petro-Sim™ model and MRL for different
conversion calculation method KBC 1985. days of operation using the maximum visbreaker
conversion calculation method KBC 1985.

1.06
RVB Specific Gravity

600
1.04

RVB Viscosity 100oC (cSt)


500
1.02 400
(g/cm3)

300
1.00
200
0.98 100
0

PS Prediction MRL Real


PS Prediction MRL Real
Figure 17 – Prediction of visbreaker residue
specific gravity by Petro-Sim™ model and MRL for Figure 20 – Prediction of visbreaker residue
different days of operation using the maximum viscosity 100C by Petro-Sim™ model and MRL for
visbreaker conversion calculation method KBC different days of operation using the maximum
1985. visbreaker conversion calculation method KBC
1985.

0.85
GOVB Specific Gravity

The developed MRL allows to obtain results


0.84 very similar to those obtained in the simulation
model built in Petro-Sim™. As can be seen
(g/cm3)

0.83 above, the predictions by the MRL and the


Petro-Sim™ model are practically coincident.
0.82 These results validate the vectors, bases and
deltas chosen.

6.2 KBC 2005 Method


PS Prediction MRL Real According to KBC, the most recent method,
KBC 2005, is more accurate than KBC 1985
Figure 18 – Prediction of visbreaker gasoil method. In order to evaluate if the KBC 2005
specific gravity by Petro-Sim™ model and MRL for method leads to RVB stability with greater
different days of operation using the maximum adherence to reality, a model was solved using
visbreaker conversion calculation method KBC
1985.
this method.

6.2.1 Real Data Predictions


Figure 21 compares the real RVB p-value
with RVB p-value predicted using KBC methods
1985 and 2005.

7
1.8 Regarding the RVB p-value, the model
solved using the KBC 2005 method, shows
1.6
RVB P-Value predictions with a greater amplitude of values
1.4 that better mimic the real values trend. These
1.2 predictions show a slight improvement
compared to the predictions made with the KBC
1.0 1985 method.
0.8
11-11-2017 01-03-2018 08-03-2018 6.2.3 MRL Predictions
KBC 2005 Prediction KBC 1985 Prediction Real The implementation of the simulation model
using KBC 2005 method in LP model was
Figure 21 – Prediction of visbreaker residue studied using an MRL.
p-value using laboratory analysis data as input Figure 23 to Figure 28 compare the MRL
using the two maximum visbreaker conversion predictions with the simulation model
calculation methods KBC 1985 and 2005. predictions in Petro-Sim™, when using KBC
2005 method. The figures present yields and
product properties predictions and real values.
This figure shows that predictions using KBC
2005 method present lower deviations from real
values than predictions using KBC 1985 92

RVB Yield (%wt)


method. This shows that using KBC 2005 90
method the predictions are more accurate. 88
In what concerns to yields and other product 86
properties the VB unit simulation model results 84
with the two methods are very similar. 82

6.2.2 Crude Mix Predictions


In order to verify the performance of the
model using the calculation method KBC 2005,
predictions were made from the crude mix, PS KBC 2005 MRL KBC 2005
using crude assays, as it was done for the Real
model with the KBC 1985 method.
Figure 22 present the real and predicted Figure 23 – Prediction of visbreaker residue
yields and product properties. The figure shows yield by Petro-Sim™ model and MRL for different
the prediction using KBC 1985 and 2005 days of operation using the maximum visbreaker
methods. conversion calculation method KBC 2005.

1.6
1.5 12
RVB P-Value

GOVB Yield (%wt)

1.4 11
1.3 10
1.2 9
1.1 8
1.0 7
6

KBC 2005 KBC 1985 Real


PS KBC 2005 MRL KBC 2005
Figure 22 – Prediction of visbreaker residue
Real
p-value from crude mix using the two maximum
visbreaker conversion calculation methods KBC
Figure 24 – Prediction of visbreaker gasoil yield
1985 and 2005.
by Petro-Sim™ model and MRL for different days of
operation using the maximum visbreaker
conversion calculation method KBC 2005..
Like it was observed with real data
predictions, the crude mix predictions using the
two MVC calculation methods are very similar
too with respect to yields and properties.

8
RVB Viscosity 100oC (cSt)
1.08 1000

RVB Specific Gravity (g/cm3)


1.06 800
1.04 600
1.02
1.00 400
0.98 200
0.96 0

PS KBC 2005 MRL KBC 2005 PS KBC 2005 MRL KBC 2005
Real Real

Figure 25 – Prediction of visbreaker residue Figure 28 – Prediction of visbreaker residue


specific gravity by Petro-Sim™ model and MRL for viscosity 100C by Petro-Sim™ model and MRL for
different days of operation using the maximum different days of operation using the maximum
visbreaker conversion calculation method KBC visbreaker conversion calculation method KBC
2005. 2005.

0.850 With the simulation model built with the KBC


GOVB Specific Gravity (g/cm3)

2005 calculation method, the MRL developed


0.840 can mimic the behavior of the model in Petro-
-Sim™ when predicting the actual performance
0.830 of the VB unit.
Therefore it is possible to state that the delta-
0.820 -base structure developed using the KBC 2005
method would present good results when
implemented in the LP model, achieving a good
optimization in the crudes selection for the
production of VLSFO.
PS KBC 2005 MRL KBC 2005
Real

Figure 26 – Prediction of visbreaker gasoil 7 Conclusions


specific gravity by Petro-Sim™ model and MRL for
different days of operation using the maximum
visbreaker conversion calculation method KBC The simulation model of the VB unit with the
2005. KBC method 1985 presents a very good
prediction of the reality with respect to yields
and product properties. The predicted
performance is highly dependent on the
1.7 feedstock quality. Regarding the RVB stability
the p-value predictions present a good
RVB P-Value

1.5
adherence to the real values meaning that the
1.3 new simulation model is able to predict RVB
1.1 stability which is crucial for VLSFO production.
It should be noted that the stability prediction is
0.9
not ensured by the simulation model that Galp
currently use. The new simulation model
corresponds to a great improvement in the Galp
simulation models as a whole.
The simulation model predictions using KBC
PS KBC 2005 MRL KBC 2005 2005 method are very similar to those obtained
Real by the simulation model using the KBC method
1985. The MVC values obtained by the
Figure 27 – Prediction of visbreaker residue p- simulation model when using the KBC 2005
value by Petro-Sim™ model and MRL for different method are in a wider range than the values
days of operation using the maximum visbreaker
obtained by the KBC method 1985. This is also
conversion calculation method KBC 2005.
observed in p-value predictions of the
simulation model with the KBC 2005 method. In

9
contrast to KBC 1985, the p-value predictions [6] A. Breneol, "Marine Fuel Stability
with the KBC 2005 method are more dependent and Compatibility - Issues, Tests and
on the feedstock quality. This method presents Management".
an improvement in the p-value prediction [7] Shell Global Solutions International
comparatively to the old method. B.V., "SMS 1600-01 - Determination of
In this work it was carried out the State of Peptization of Asphaltenes in
development of a delta-base structure, in order Heavy Oil Streams," 2001.
to provide the simulation data to the LP model.
The representation of the delta-base structure [8] E. Y. Sheu and D. A. Storm,
in the VB unit is pioneering in Galp. Asphaltenes - Fundamentals ans
To generate the delta-base representation, Applications, New York: Plenum Press,
was used the Linear Programming Utility of the 1995.
Petro-Sim™ software to obtain the impacts that [9] A. Pina, P. Mougin and E. Behar,
variations in the feedstock properties have on "Characterisation of Asphaltenes and
given products properties. This is the Modelling of Flocculation – State of the
information required to implement the Art," Oil & Gas Science and
simulation model in the LP model. Technology – Rev. IFP, vol. 61, no. 3,
The delta-base structure was validated using pp. 319-343, 2006.
the linear representation model that allows the [10] H. Groenzin and O. C. Mullins,
evaluation of the new VB unit representation in "Molecular Size and Structure of
the Galp LP model. Both linear representation Asphaltenes," Petroleum Science nd
models, constructed from the simulation model Technology, vol. 19, no. 1-2, pp. 219-
with the KBC1985 method and with KBC 2005, 230, 2001.
are robust and reproduce with very low [11] J. H. Gary, G. E. Handwerk and M.
deviations the simulation model predictions. J. Kaiser, Petroleum Refining -
The delta-base structure constructed for the Technology and Economics, CRC
implementation of the simulation model in the
Press, 2007.
LP model is validated and is representative of
reality, being sufficiently robust to be [12] KBC Andvanced Technologies Ltd,
implemented in the LP model. This allows to Petro-Sim Help.
perform good optimizations in the selection of
crudes for the production of VLSFO.
The simulation model of the VB unit with the
KBC 2005 method makes good predictions of
yields and product properties and its delta-base
representation will be implemented in the PL
model.

References

[1] IMO, "Sulphur 2020 – cutting sulphur


oxide emissions," [Online]. Available:
http://www.imo.org. [Accessed Fev
2019].
[2] D. S. J. Jones and P. P. Pujadó,
Handbook of Petroleum Processing,
Springer, 2006.
[3] J. G. Speight, Handbook of
Petroleum Refining, CRC Press, 2017.
[4] J. Singh, S. Kumar and M. O. Garg,
"Kinetic Modelling of Thermal Cracking
of Petroleum Residues: A Critique,"
Fuel Processing Technology, vol. 94,
pp. 131-144, 2012.
[5] Encyclopaedia of Hydrocarbons, vol.
II/Refining and Petrochemicals, Istituto
della Enciclopedia Italiana, 2005.

10

You might also like