Crude

Didier Lambert, Claude Saint-Martin, Bernard

Ribero and Michel Sanchez, Topnir Systems,
France, Gustavo Lopez, Luz Angela Novoa, Luz Myriam Guerra and Jorge
control
Huertas, Ecopetrol, Colombia, and Diego Martinez and Lina Lozano,
Insurcol Ltda, Colombia, discuss systems for crude management and
their application at the Ecopetrol crude network.

C
rude diet characterisation is crucial for refinery optimisation. It affects the planning and scheduling activities and
the optimisation of the individual units as well as the blending of final products. This is particularly important
when special products are produced, for instance paraffinic and naphthenic lubricants and bitumen. A good
understanding of the crude in the refinery has an impact on the planning and affects all the refinery operations.
Real time knowledge of crude properties is the only way to ensure that full value is extracted from the crude in the
refining activity, based on simulated plant operation and product quality. This affects both the allocation
of the crude to the appropriate crude distillers, and also the setting of target yields and
qualities for process units.
Short term fluctuation of feed properties introduces instability
into the unit, necessitating larger offsets from constraints. In
terms of days, changes in the crude mix necessitate
operational changes: steady operation over a
longer period enables better optimisation
of the unit by pushing against the
appropriate constraints.
A huge project has been
completed at Ecopetrol in
2005 aimed at identifying
and characterising the
whole crude production
network, including
 salt content and volumetric yield, which consequently
varies the quality of the feedstock to the various crude
units at the refinery. The following four segregations
occur:
l Cusiana: for low sulfur diesel grades (ACPM) and
reduced crude is fed direct to the cracking unit.
l Cupiagua: used for paraffinic lubricants.
l LCT: used for naphthenic lubricants.
l Mezclado: a mixture of more than 50 crude oils for
the Demex-Unibon-VRII scheme and bitumen.
Although the types of crude to be segregated were
well established, before the launching of the crude
project the refinery did not have systems that allow
characterisation of the quality (assays) of the crude
Figure 1. Crude mixture monitoring scope.
arriving at the refinery. The only properties that were
known were salt and BSW. Moreover, there was no
control on the quality of the crude fed to each one of
the process units, since the only consideration was an
average API. On occasion, the homogeneity of these
feeds was not good due to differences between the top,
middle and bottom samples, which led to instability in
the operation of the topping units.
There were no tools in place to allow prediction of
the composition of the prepared blend, therefore it was
impossible to know the yield of products in the topping
units. Also no tools were available to prepare feeds to
maximise distillate yield, since the only consideration
was an average API. It is worth noting that two crude oils
with the same API can have different yields in one crude
unit. Finally, with the processing of light crude, topping
units were constrained in their overhead systems,
Figure 2. Crude terminology.
resulting both in constrained throughput and increased
gas to fuel gas and to the flares.
both pumping stations linked to inland wells and the two One example is that some Cusiana crude was
refineries located in Cartagena and Barrancabermeja. purchased in excess of the plan to ensure that low sulfur
This unique identification and characterisation of crude diesel could be produced and to meet the cetane index
mixture obtained from Topnir and Simcrude led to in the diesel pool. This resulted in sulfur and cetane
annual benefits of several million US$, not only for crude giveaway. This was attributed to uncertainties in the
operation itself but also for downstream activities such crude/component logistics systems.
as diesel blending. This article describes the project as For Cartagena the refinery operations were reactive
well as the technology and real case results. to the crude fed to the refinery as they did not know
the crude that was in the blend. Crude was segregated
Objective of project according to API and sulfur, not allowing for a proper
The project, named the crude diet and property assessment of product yields and quality prior to feed
monitoring project, is dedicated to monitoring the to the CDU. This situation produced a suboptimal
composition of crude mixtures: operation of the refinery, where plant capacities were not
l At two key pumping stations (Vasconia and maximised.
Ayacucho) of the inland crude pipe network.
l At the entry points of the two refineries New situation after project
(Barrancabermeja and Cartagena). completion
An integrated system to control, optimise and schedule
This information feeds the SENSOR monitoring the preparation of crude blends that are received at
system, constituted by the network of Topnir analysers, Barrancabermeja and Cartagena Refineries allows for,
the planning and scheduling and IT systems. The project among other things:
also includes the characterisation (monitoring of crude l Automation of transport and storage operations
properties) of the crude mixture feed to the topping for the various crude that are received in the
units at Barrancabermeja and Cartagena. The process is Barrancabermeja and Cartagena refineries.
shown in Figure 1. l Preparation of feeds for each of the topping
units that are sufficiently homogeneous to allow
Previous situation greater stability in these units and for the type of
Barrancabermeja acquires its crude from various blend defined as appropriate for the operational
locations around the country. The composition of this requirements at the refinery.
crude varies greatly in properties such as density, sulfur, l Preparation of feeds that allow maximising the yield

Reprinted from HydrocarbonEngineering October2007 www.hydrocarbonengineering.com

 of valuable products, such as
Jet A1 and/or ACPM.
l Preparation of homogeneous
and optimum feeds for the
preparation of bitumen and
lubricants.
l Forecasting the properties
for the various intermediate
or finished products that are
produced both in the topping
units as well as in the remaining
process units in the refinery,
to direct production towards
valuable products.
l To ensure raw materials
regarding quality and quantity
for process units downstream
from topping units.

The crude composition is known Figure 3. Crude flowsheet origins.

before arrival. A combination of
better information in the Ecopetrol
transportation pipeline and the use
of Topnir/Simcrude technology Figure 4. Cartagena crude application
to predict the composition of the context (types).
crude blend in the pipelines to the
pumping stations and crude tanks
at Barrancabermeja and Cartagena
now addresses this issue.
In total there are several days
crude stocks between the pumping
stations and the time when the
crude is fed into the CDU when
Topnir is available at the pumping
stations, giving a good lead time for
crude and product scheduling.
Once blend composition is
known, product quantities and
qualities of all streams in the
refinery configuration can be
calculated such that refinery
throughput is maximised and
product qualities are known.
A second stage, phase 2,
involves installing APC and
optimisation on the crude oil
blender to trim the blend within
constraints given by scheduling.
This requires an online NIR located
either on the component streams
or the blend headers to the topping
units. Phase 2 allows the feeds
to the individual topping units to
be better optimised to take into
account transient effects. It will
therefore give additional benefit in
the areas of:
l Preparation of feeds that allow
maximising the yield of valuable
products such as Jet A1 and/or
ACPM (diesel grade).
l Preparation of homogeneous
and optimum feeds for bitumen
and lubricants preparation. Figure 5. Barrancabermeja crude application context (types).

www.hydrocarbonengineering.com Reprinted from October2007 HydrocarbonEngineering

 Table 1. Monitored cutpoints
Crude Nafta 1 Nafta 2 Nafta 3 Nafta 4 Medios 1 Medios 2 Medios 3 Gasoleo 1 Gasoleo 2 Gasoleo 3 CRED FVAC
˚C 15 - 60 60 - 107 107 - 152 152 - 199 199 - 249 249 - 315 315 - 371 371 - 427 427 - 482 482 - 565 371+ 565+

The terminology applied to crude monitoring has

been defined as follows:
l Crude identification to measure the percentage of
pure crude in any crude mixture.
l Crude characterisation to measure the exact
properties of the crude itself.

Crude identification applies mainly at crude

mixtures at the various pumping station but also
at the crude arriving at the refinery gate, aiming at
Figure 6. NIR system hierarchy. facilitating the crude allocation/segregation, as well
as E&S inputs.
Crude characterisation is mainly applied at the
refinery on the CDU feed to monitor the quality of the
blend feeding the various CDUs/topping units. Such
data are used for crude blending control/optimisation;
this application will therefore be limited to the two
refineries.
Figures 3, 4 and 5 show an overview of the
project complexity. Figure 3 describes the complexity
of the crude oil network in Ecopetrol operations.
Figure 4 details the application at Cartagena Refinery
while Figure 5 shows the crude flowsheet to be
considered in Barrancabermeja Refinery.

Modelling software/NIR
Figure 7. Topnir crude characterisation screen display. calibration
NIR calibration is done through Topnir and Simcrude
technology using the topology method and computer
l Forecasting of the properties for the various database densification. It integrates automatic outlier
intermediate or end products that are produced detection and prediction from the online Topnir database
in the topping units, to direct production towards
in real time densification.
valuable products and to ensure raw materials for
the process units downstream from said process.
Topnir
Topnir is a chemometrics software that exploits the NIR
Project overview spectra to determine a full vector of properties of the
The crude diet and property monitoring system is
hydrocarbon product. The key features of the Topnir
composed of:
technology package are the near infra red (NIR) hardware
l Two new at-line NIR systems (Topnir Systems AS200 required to carry out the analysis. The software enables
analyser type using NetworkIR FTIR Spectrometer the information provided by the analyser to be modelled
integrated in a cabinet) located in the two key and the multiple properties determined. It will then
pumping stations (Ayacucho and Vasconia), named provide system integration to allow the properties to be
Slave 1 and 3. transferred to the process advanced control system.
l Two existing offline NIR systems upgraded at the Topnir, based on topology, works through pattern
refineries. recognition and databank densification. It does not use
l One new offline NIR system at Instituto Colombiano a linear model for a set of properties (one model per
del Petroleo (ICP), from Ecopetrol, for development property). Topnir instantly delivers all the properties
purposes. required for a given application. Moreover, it offers the
l Simcrude model and software to provide ‘crude possibility of extrapolation from the initial calibration range.
identification’ application to deliver crude diet/ratio Concerning the treatment of outliers, the classical
of pure crude. linear methods require a very heavy amount of work to
l Topnir model and software to provide ‘crude maintain the models. It can rapidly become a bottleneck
characterisation’ application to deliver CDU feed when the application involves a lot of properties, every
properties (density, distillation (12 points), sulfur and one requiring a specific calibration treatment.
TAN). The system takes into account any outlier and does
l Additional online NIR systems planned in phase 2 on not require any new calibrations of the models. It is a
the feed inlet of the CDU plants at Cartagena and self learning method designed for the reality of industrial
Barrancabermeja. operations.

Reprinted from HydrocarbonEngineering October2007 www.hydrocarbonengineering.com

 l Five blended crude, mixture of five crude preblends,
at Barrancabermeja refinery.
l One blended crude, mixture of five crude preblends,
at Cartagena refinery.

Simcrude
Simcrude has been developed, as an extension to the
Topnir modelling technique, to satisfy the following
requirements:
l Predict the type of crude oil and the properties from
an unknown sample.
l Predict the formulation of an unknown crude oil
blend and its associated properties.
l Perform as a crude oil tank management tool.

Figure 8. Simcrude crude identification screen display. Thus, Simcrude is used as shown in Figure 3 to give
the following information:
l Identification of the composition in pure crude:
s For 15 families divided in eight sampling points at
Vasconia pumping station site.
s For seven families divided in four sampling points
at Ayacucho pumping station site.
l Identification of the composition in pure crude:
s For 15 families divided in seven sampling points at
the Barrancabermeja refinery.
s For 11 families divided in two sampling points at
the Cartagena refinery.
l Identification of the composition in pure crude:
s For six families in crude pre-blends of the
Barrancabermeja refinery.
s For four families in crude pre-blends and one
imported crude family of the Cartagena refinery.
Figure 9. List of crude oils identified in mixtures and characterised.
Indeed, Simcrude monitors the crude percent from
The principle of topology is to represent spectra of n samples collected at various sampling points of the
wavelengths to one point in a set of planes. These planes two major pumping stations (Ayacucho and Vasconia)
are defined by some aggregates and are representative as well as the crude percent from samples collected at
of particular chemical characteristics (aromatic, Cartagena at the pipe entrance (one line) and at the three
unsaturated ratio). By this concept the spectral vision is tanks and the crude percent from samples collected at
reduced to some planes that allow the quick and precise Barrancabermeja at the pipe entrance (seven lines) and
visualisation of the quality of the analysis products. at the ten tanks.
Only one model is required to determine the The two programs, i.e. Topnir™ and Simcrude have
full set of properties for one or several grades. been delivered at Ecopetrol sites:
Moreover the technique is self learning, as the model
is automatically updated every time a new point l Topnir™ installed on the laboratory PC at ICP,
Barrancabermeja, Cartagena and on CDU online
is entered in the database. Thus, Topnir offers the
analysers at Barrancabermeja, Cartagena on
opportunity to anticipate feed quality changes and
phase 2.
to optimise process CDU based on real time quality
determinations. l Simcrude installed on the laboratory PC at ICP,
One example is the ability to predict true boiling Barrancabermeja, Cartagena, and on the at-line
point (TBP) of a crude mix to the topping. The incentive analysers PC at Vasconia, Ayacucho.
is particularly important when refineries are switching s One Simcrude database for circa 100 in-land
crude qualities several times per week and have difficulty +30 import crude oils for determination of percent
tracking crude mixture in feed tanks. It is also possible in mixtures (identification).
by using Topnir prediction on the TBP curve to calculate s One Topnir database/models for approximately
the CDU product yields at given cut points. 100 in-land +30 import crude oils for determination
The benefits from crude analysis are a function of of stream properties such as distillation (12 points)
the economics, but the reduction of feed transitions can i.e. percent distillate at each cut point, degree
yield a magnitude of gains, keeping the distillation unit API, sulfur and TAN (characterisation). Regarding
closed to the optimum set of cut points and throughput. the distillation points, it should be noted that the
Thus, Topnir is used to provide characterisation of the points corresponding to the 12 cuts in Table 1 are
properties of the topping units feed for: monitored.

www.hydrocarbonengineering.com Reprinted from October2007 HydrocarbonEngineering

 NIR systems/functions
Five off/at-line NIR analysers are used. The database/
models have been built on the master NIR to be installed
in ICP while four ‘slave’ analysers are installed, one
in each of the two pumping stations (PS) and the
two refineries, to monitor both crude diet and crude
properties (only in the refineries). Online NIR analysers
on refinery CDU plant feed lines will be used for crude
characterisation in phase 2. The databases have been
Figure 10. Topnir/Simcrude system is able to predict outlier.
then transferred to these four ‘slave’ spectrometers as
depicted in Figure 6.
The transfer occurred in three steps according to Refinery a 10% of blending product optimisation
complexity of the crude diet. The four slave apparatus leading to an annual benefit of several million US$. For
for phase 1 are actually used: Cartagena Refinery, the estimated benefit for better
l For monitoring the crude percent from samples crude characterisation and optimisation is an increase
collected at various sampling points of the two major in crude throughput from additional crude processed
pumping stations (Ayacucho and Vasconia). in the refinery, leading also to an estimated amount of
l For monitoring the crude percent from samples several million US$/yr. Crude management takes half of
collected at Cartagena at the pipe entrance (one line) this benefit and the other half is attributed to scheduling
and at the three tanks, and monitor the properties of tools. 20% of the product giveaway is also attributed to
the blended stream (one line). better crude management.
l For monitoring the crude percent from samples The second stage, phase 2, involves installing
collected at Barrancabermeja at the pipe entrance APC and optimisation on the crude oil blender to trim
(seven lines) and at the 10 tanks, and monitor the the blend within constraints given by scheduling. This
properties of the blended stream (four lines). requires an online NIR located either on the component
streams or the blend headers to the topping units.
Results and benefits Phase 2 allows the feeds to the individual topping
TBP prediction of a pure crude or a crude blend is units to be better optimised to take into account
achieved by Topnir which procures a unique advantage transient effects. It will therefore give additional benefit in
in daily operations. Indeed, advanced control strategies the areas of:
have been constrained mainly by poor availability of l Preparation of feeds that allow maximisation of the
online analysers and dubious robustness of inferential yield of valuable products such as Jet A1 and/or
models for stream qualities, especially during ACPM (Diesel Grade).
transients. l Preparation of homogeneous and optimum feeds for
It is also possible by using Topnir prediction on the bitumen and lubricants preparation.
TBP curve, to calculate the CDU product yields at given
l Forecasting of the the properties for the various
cut points. Topnir, Toplab or Topwin are gathered under
intermediate or end products that are produced
the same software. The different names are given to in the topping units, to direct production towards
highlight specific functionalities. valuable products and to ensure raw materials for
l Toplab is a Topnir module used to allow Topnir to the process units downstream from said process (i.e.
collect a spectrum with the spectrometer. It is used cracking, demex, unibon and lubes).
only offline.
l Topwin stands for Topnir Windows version It is currently estimated that this phase gives an
(laboratory and at-line/online). additional benefit of about 1% throughput on the topping
units.
All the database work (update, compilation, studies,
check up) is done offline, on the laboratory PC. A typical Conclusion
example of screen display is shown in Figure 7. Topnir/Simcrude technology is able to provide two key
Simcrude is a chemometrics software that exploits pieces of information:
the NIR spectra to determine the percent of the pure l Direct identification and quantification in ratio of pure
crude in a crude blend. Typical Crude identification crude composition in an unknown crude mixture.
obtained from Simcrude is shown in Figure 8. l Real time measurements of full property vector for
The complexity of the crude oil in-land and import any pure crude and crude mixture, including full
at ECP is very high. The list of crude oils that have been TBP curve, TAN, sulfur and degree API within ASTM
identified in mixtures and characterised is shown in warranty, in less than one minute.
Figure 9.
Moreover, the Topnir/Simcrude system is able to The unique system installed at Ecopetrol sites is
predict outlier as shown in Figure 10 where the spectra, managing a network involving more than 100 crude oils
the location of sample compared to the box is displayed throughout the country.
as well as the prediction obtained. The identification and characterisation of crude
mixture obtained from Topnir and Simcrude led to
Benefits annual benefits of several million US$, not only for crude
It is expected that the implementation of phase 1 (at-line operation itself but also for downstream activities such
Topnir/Simcrude systems) allows the Barrancabermeja as diesel blending.

Reprinted from HydrocarbonEngineering October2007 www.hydrocarbonengineering.com