Example 1 - Oil Page 1 of 54

Example 1 - Oil
This guide is intended to demonstrate the thinking process behind characterising a fluid in PVTp - in this case an oil.

It should be noted that this is not a guide of the essential steps for matching any fluid - this is because every fluid is different, has its own challenges and by
definition there are multiple solutions as well as pathways to a particular solution.

In order to begin this example, open an instance of PVTp and Create a New File:

Scenario
An oil reservoir in the extensive Logie Mill field has been discovered. A fully integrated model is to be built to come up with a field development strategy.
This integrated model is to be comprised of a numerical reservoir model, well and surface network models as well as a process model. Additionally, the
likelihood of flow assurance issues such as hydrate or wax formation is to be investigated and therefore the use of an Equation of State (EOS) rather than a
Black Oil (BO) description of the reservoir fluid is required.

A fluid sample was taken and analysed in a laboratory. The engineer has been tasked with using the lab data to develop an EOS description of the reservoir
fluid using PVTp.

Workflow Summary
QC Lab Report:
The first step in building an EOS description of a fluid is to quality check the laboratory results. This can be done by establishing whether the data available
is sufficient to describe the fluid’s behaviour over the range of conditions it will be exposed to over the field’s life time. With this in place, for example plotting
the composition as a function of carbon number can show whether the determined composition is physical.

Lab Data Entry:

In order to continually quality check the validity of the EOS, we will want to be able to reproduce the laboratory experiments, hence the lab data needs to be
entered.

Define Pure Components:

With the Lab Report quality checked, the engineer then needs to decide which components to model explicitly as pure components and which components
to lump together as a pseudo component. Once this has been completed, the fluid composition can be entered.

Define Pseudo Components:

To start performing flash calculations with the EOS at this stage, the critical properties of the pseudo component need to be defined. Correlations may be
used to estimate these properties from the pseudo’s molecular weight (MW), specific gravity (SG) and boiling point temperature (BPT).

The aim at this stage is to get the EOS to roughly reproduce the liquid phase density and GOR at standard conditions as well as the saturation pressure at
reservoir conditions. Splitting pseudo and introducing binary interaction coefficients (BICs) can also be used to achieve this.

Regression:
The EOS can then be fine-tuned in an attempt to reproduce the lab experiments by regressing on the critical properties, acentric factor, volume shift and
BICs. Multiple regressions can be performed in series, but the results must always be validated against simulated lab experiments and possess a natural
distribution of critical properties etc. Weightings can be applied to the entered data in order to steer the regression towards more important parameters and
to reflect uncertainties associated with some lab measurements. Some data can also be excluded from the regression, to simplify the regression.

If an unnatural distribution is determined, or the EoS insufficiently describes the fluid behaviour, then the user should return to their pseudo split definition
and regression strategy and repeat the process.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 2 of 54

Validation:
Once the EOS reproduces the lab experiments and the critical properties exhibit natural trends, additional validation against real-life process data is
required. The reason for this is that the fluid used to perform the experiments in the laboratory may in fact not be fully representative of the actual reservoir
fluid. To understand whether the fluid sample was representative of the reservoir fluid a more "holistic" approach needs to be takes a wider range of data
into consideration, for example:

· Historical Well Tests

· Repeat Formation Tests (RFTs)

EOS Application:
The validated EOS model can then be used for flow assurance calculations, e.g. hydrate and wax formation, to build the integrated production model, i.e.
deriving a lumping/delumping rule to meet the requirements of the various models within the integrated model (simulator, well, surface network and process
model).

The general workflow is summarised in the graphic below.

For the purposes of this example, we will use the Characterisation Wizard to help guide us through this workflow.

Data Available
The following data is at the engineer’s disposal:

· Fluid composition
· Constant Composition Expansion (CCE) experiment
· Differential Liberation (DiffLib) experiment
· Separator (SEP) experiment

Please note, all of this data can be found below and in the Excel spreadsheet provided within the PVTp sample files.

Reservoir Conditions
Units
Reservoir Pressure 2339 psia
Reservoir Temperature 185 °F
Reservoir Depth 5949 ft

Recombined Composition
Component Mol%
N2 0
CO2 0.8
C1 36.22
C2 8.65
C3 7.78
iC4 1.94
nC4 4.58
iC5 2.26
nC5 2.42
C6 3.25
C7 2.96
C8 2.69
C9 2.44

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 3 of 54

C10 2.22
C11 2.02
C12 1.84
C13 1.67
C14 1.52
C15 1.38
C16 1.26
C17 1.14
C18 1.04
C19 0.95
C20+ 8.96

Mean Molecular Weight of C7+ fraction = 220 g/mol

Mean Specific Gravity of C7+ fraction = 0.87

CCE Experiment
Pressure, Relative Volume, Oil Density, Oil Viscosity,
(psia) (V/Vsat) (kg/m3) (cp)
4000 0.97 661 0.49
3600 0.976 657 0.48
3200 0.982 653 0.47
2800 0.987 649 0.46
2339* 0.994 644 0.44
2200 0.997 643 0.43
2091** 1 641 0.42
2000 1.021 - 0.43
1750 1.093 - 0.44
1500 1.199 - 0.45
1250 1.36 - 0.46
1000 1.621 - 0.47
750 2.084 - 0.48
500 2.92 - 0.51
200 7.691 - 0.61

*Reservoir Pressure
**Saturation Pressure

DiffLib Experiment
Pressure, Oil Density, Solution GOR, Oil FVF,
(psia) (kg/m3) (scf/bblresidual) (bbl/bblresidual)
4000 661 996 1.636
3600 657 996 1.638
3200 653 996 1.655
2800 649 996 1.665
2339* 644 996 1.677
2200 643 996 1.680
2091** 641 996 1.686
1757 655 842 1.611
1408 669 700 1.538
1067 681 571 1.477
711 696 445 1.412
370 707 311 1.310
15 789 1.057

*Reservoir Pressure
**Saturation Pressure

SEP Experiment
SEP Experiment 1:

Pressure, Temperature, Gas Oil Ratio, Oil FVF, Gas Gravity

(psig) (°F) (scf/stb) (bbl/stb)
80 75 661 1.088 0.842
0 75 110 1.007 1.33

Stock Tank Oil Gravity 0.8145 or 42.2°API

SEP Experiment 2:

Pressure, Temperature, Gas Oil Ratio, Oil FVF, Gas Gravity

(psig) (°F) (scf/stb) (bbl/stb)
40 75 735 1.088 0.888
0 75 29 1.007 1.447

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 4 of 54

Stock Tank Oil Gravity 0.8140 or 42.3°API

SEP Experiment 3:

Pressure, Temperature, Gas Oil Ratio, Oil FVF, Gas Gravity

(psig) (°F) (scf/stb) (bbl/stb)
0 75 754 1.007 0.961

Stock Tank Oil Gravity 0.8139 or 42.35°API

Import the Lab Report

This step is optional and can only be performed if the lab report is available in an Excel format.

As in this case we do have an Excel file available we will import this into the PVTp file. This step serves to keep all the necessary data together in one
location and allows us to quality check the data in one location.

Double click on Add PVT Report in the Characterisation Wizard window.

This will place a PVT Report object onto the Characterisation Window.

Double click on the PVT Report object and select Browse to navigate to the required PVT report. In this can be found in the sample directory
under ...Samples\pvtp\Example 1 Oil\Lab Data.xlsx.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 5 of 54

Selecting Open will import the report into the object in the same format and tab structure as the original Excel file.

Data in this screen can be modified if necessary without affecting the original Excel sheet. In this instance some data formats cannot be automatically
interpreted so need to be manually modified.

Go to the CCE tab and modify the pressures to remove the * and ** signifying the reservoir and saturation pressure respectively.

Once this has been done, press OK to close the window.

Lab Report Quality Check

The assumption taken in most companies is that the individual Equation of State (EOS) descriptions are accurate and matched to good measurements from
lab reports. However the measurements themselves may carry experimental errors that engineers need to be aware of before forcing the EOS to reproduce
them.

There are various challenges associated with the fluid sampling and the lab measurements themselves. The fluid samples can be collected either:

1. Downhole
2. At Surface

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 6 of 54

Downhole samples run the risk of gas escaping through the seals of the sampling tool as the sample is brought to surface. When the sample is then taken
to the lab and restored to reservoir conditions, by elevating the sample to the reservoir temperature and pressure, the gas that escaped on the way to
surface will not be re-dissolved back into the oil, which will impact the measurements of the volumetric properties (i.e. GOR and Formation Volume Factor).
For this reason the measured saturation pressure should be checked with the reservoir pressure or an RFT measurement as this could lead to the
conclusion that there is a problem with the sample taken.

A surface sample is created by taking a separate sample of gas and oil at the separator during a well test. The separator pressure and temperature are
noted along with the in-situ GOR. To recreate the reservoir fluid in the lab, the oil and gas samples are re-combined in a common sample with the same
GOR in the field and then the temperature and pressure are elevated to reservoir conditions. If the GOR reading during the rest was wrong (e.g. unstable
flow from well, liquid carry over, etc.) then the re-combined fluid will not represent the reservoir fluid either. Therefore it is good practice to have the well test
report to hand to validate this

There are various checks that can be performed before progressing with the characterisation:

· If there are multiple samples in the report, identify which sample was used to perform the experiments
· Check the sample was given sufficient time to stabilise at reservoir conditions
· Perform a mass balance check
· Check for inconsistencies in the measurements (i.e. Oil Formation Volume Factor, Density, GORs, etc.)
· Check for contamination
· Check Hoffmann Plot for K values
· Check Saturation pressure vs Reservoir Pressure
· etc.

Review of Available Data

The laboratory experiments intend to describe different aspects of the fluid’s behaviour but only when combined they can give a full picture of the fluid’s
behaviour. For further details on how these experiments are performed and what they aim to simulate, please refer to Lab Experiments.

For example, the Constant Composition Expansion (CCE) experiment is analogous to the expansion of the fluid in the reservoir as the pressure is reduced
at a constant temperature. The SEP experiments captures the fluid path from the reservoir through the well, surface network and separator stages to
standard conditions.

The Differential Liberation (DiffLib) experiment describes how the composition in the reservoir can change over time as the pressure reduces and gas starts
evolving from the oil, whilst the temperature is kept constant. For a discussion on the differences between the results of the CCE and DiffLib experiments,
please refer to the Appendix.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 7 of 54

With the above in mind, in principle only two experiments, including the separator experiment are required to perform a characterisation. With additional
experiments, the fluid’s behaviour is better understood and therefore more confidence can be placed in a characterised EOS to be representative of the
fluid’s behaviour over the field’s life time.

The data available should then be quality checked to ensure that measurements of the following have been performed:

· Saturation Pressure at Reservoir Conditions

· Reservoir or Saturation Liquid density
· Surface Liquid density
· GOR

Compositional Profile Check

A compositional profile check tests whether whether the composition follows a natural trend or whether there are erratic spikes. Such spikes could be an
indicator of the sample having been contaminated, with for example drilling mud, or the sample having been improperly recombined. A plot of the fluid
composition as a function of carbon number is shown below for illustrative purposes.

As the trend is smooth and natural in appearance, it can be assumed that the sample was not contaminated and properly recombined.

Saturation Pressure
If the measured saturation pressure in the laboratory is greater than the measured reservoir pressure at the point of sampling (in the case of a separator
sample, this would be assumed to be the top of perforations), then this indicates an inconsistency between the sample and reservoir fluid. This is because
the saturation pressure at the point of sampling must me less than or equal to the reservoir pressure at the point of sampling.

The most extreme case, i.e. saturation pressure is equal to reservoir pressure, occurs at the Gas-Oil-Contact (GOC), where the gas and oil phase are in
equilibrium such that the gas and oil phase pressure must be the same. If we go beyond this limit, i.e. saturation pressure is greater than reservoir pressure
at the point of sampling, then this would suggest that the reservoir is not in equilibrium and that mass transfer occurs between the gas and oil phase. This
would typically not be the case, especially not at the start of the reservoir’s life time, as the reservoir would have equilibrated over the past millennia.

As such, if the saturation pressure is reported as being larger than the reservoir pressure, then this could indicate inaccuracies in the lab, e.g. recombination
to an incorrect GOR. Since the lab experiments have been conducted with this recombined fluid, the procedure in this case would be to progress with the
characterisation as usual and once complete perform a recombination to a known measurement of Psat or GOR.

Choice of EOS Model

Lab Report quality checked and no inconsistencies found, the characterisation process in PVTp can be started. The first step of the process is the decision
on which formulation of the EOS should be used, Peng-Robinson (PR) or Soave-Redlich-Kwong (SRK). The choice between either formulations has to be
made by the engineer, however neither formulation holds significant advantages over the other and a similar quality of characterised EOS can be achieved
with either. That being said, an important consideration in integrated modelling is that the same EOS model should be used for all underlying models (i.e.
you cannot mix and match EOS models of a different type).

Please refer to EOS Models for further information on the PR and SRK equation.

The PR EOS is used by default, but the chosen EOS model can be changed to SRK from the Main Menu Ribbon via Home | Calculation Options.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 8 of 54

Defining the Sample

With the EOS model selected the next step is to decide which components from the lab report are to be modelled explicitly as pure components and which
ones are to be lumped together as a pseudo component. The reasoning for lumping higher carbon number components (C7+) together is that due to the
exponentially growing number of isomers, it becomes exceedingly difficult to identify pure components to a high degree of accuracy. The cut-off point for
pure components is somewhat arbitrary and down to the engineer’s discretion, however as a general guide, components higher than C6 are typically
lumped into a pseudo component.

With this in mind, the laboratory composition is recalculated by lumping C7 and higher into a pseudo component (C7+) as below.

Component Mole % Molecular Weight

(g/mol)
N2 0
CO2 0.8
C1 36.22
C2 8.65
Default
C3 7.78
Database
iC4 1.94 Values
nC4 4.58
iC5 2.26
nC5 2.42
C6 3.25
C7+ 32.1 220

Please note, N2 will be omitted from the composition as it has a zero Mol % and therefore does not contribute to the phase behaviour.

This composition, along with other information relating to the sample will be entered into the PVT Sample data object. To add this data follow the steps
below.

Double click on Add PVT Sample in the Characterisation Wizard window.

This will place a PVT Sample object onto the Characterisation Window.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 9 of 54

Double click on the PVT Sample object to open the interface. Once opened we can begin populating some of fields with known data in the Sample
Information tab. In particular the Reservoir Conditions and Reference Conditions can be defined:

Moving down the tabs on the left hand side, next the Select Components tab can be filled in. This is where the pure and pseudo components can be
selected as per the composition. Select the required components, which will highlight them as shown below, and enter the number of pseudos required.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 10 of 54

Once this is done press Apply. This will transfer the selection to the following tabs.

In the Enter Composition tab, enter the component mole percentages and the molecular weight of the pseudo component. Molecular weights of the pure
components will automatically be entered best of database values.

Moving to the Enter Lab Data tab allows us to define all the lab experiments and enter data. To begin press Add Experiment and select the required
experiments from the pop-up menu. The following experiments need to be added:

1. Saturation Pressure
2. Constant Composition Expansion
3. Differential Liberation
4. Separator

This information can either be added manually or using the Importer. The following steps will illustrate both methods.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 11 of 54

Select the Saturation Pressure experiment and enter the data shown above manually. Ensure that care is taken with the units. This data can be found
within the CCE experiment, however entering it here identifies the saturation point explicitly allowing it to be specifically used in the file.

Select the Constant Composition Expansion experiment and press Import From PVT Report. Select the CCE data tab (highlighted below), highlight the
cells that contain the required data, press Set Selection as Data Table.

Once the data is selected, press Auto-Map Remaining to automatically populate the available data into the lab data table on the right hand side.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 12 of 54

This transfer will try and identify the variables based on the table name (e.g. pressure can be "pressure", "pres", "P" etc.) and identify the units. If data
cannot be picked up automatically, the it can be modified or entered manually, as is the case for the temperature.

Enter the temperature manually as below. Only one cell in the temperature column needs to be entered, and the program will assume that the rows below
are for the same temperature until otherwise specified.

Once all the data has been imported, press OK (next to Cancel Import) to complete the import. See screen-shot below for the clarification.

Repeat the import process for the Differential Liberation experiment.

Select the Separator experiment and press Import From PVT Report. Select the SEP data tab (highlighted below), highlight the cells that contain the
required data, press Set Selection as Data Table. In this case the first separator train will be used.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 13 of 54

Once the data is selected, press Auto-Map Remaining to automatically populate the available data into the lab data table on the right hand side. Note that
the pressure units have been automatically changed to psig.

Lastly we will want to add stock tank conditions manually, as due to the format of the stock tank data it will not be recognised in the import.

Add to following data for stock tank conditions:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 14 of 54

Press OK (next to Cancel Import) to accept the imported data.

Quality Check

A mass balance can be conducted to assess whether the measured fluid properties are consistent. The mass of a unit volume of oil and associated gas is
calculated at both surface and reservoir conditions. This can result in one of three scenarios:

· Surface and Reservoir mass are consistent

o Generally good as it means that the reported properties are consistent, however too small differences can suggest that the measurements
were adjusted to ensure mass balance.
o 0% difference is a clear sign of the lab report being modified to satisfy mass balance, and so should not be trusted.

· Small difference in the calculated Surface and Reservoir mass

o Differences are most likely due to inaccuracies in the laboratory procedures. As such the engineer must decide which measurements carry the
largest experimental error and take this into consideration when determining a regression strategy.

· Large difference in the calculated Surface and Reservoir mass

o This should be cause for concern as this suggests that the reported measurements are not consistent. As such, the reason for this
discrepancy should be investigated before continuing with the characterisation process.

To perform this quality check, select the Mass Balance Calculator tab. Enter the required data from the lab report, in this case data from the first separator
train.

Once all the required data has been entered press Calculate Mass Balance to perform the calculation.

As can be seen, the mass imbalance is quite small (0.11%), which suggests that the collected data is consistent. If the error is larger, then possible
inaccuracies in the measurements should be investigated further and also taken into consideration when developing a regression strategy.

Now that all required data has been entered in the the sample has been defined and some initial quality checks have been carried out. To close the PVT
Sample data object, press OK.

Define Pseudo Component

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 15 of 54

In the PVT Sample object, a single pseudo component was defined. This component is designed to represent all the components from C7 onwards, and as
such needs to be tuned in order to accurately do this.

Double click on Add Pseudo Properties in the Characterisation Wizard. Ensure that the "PVT Sample" is highlighted in the Navigation Tree before doing
this.

This will place a new Pseudo Properties and PVT Fluid object on the Characterisation Window, connected to the PVT Sample that has just been defined.

Double click on the Pseudo Properties object.

In the window that opens, the Pseudo components' critical properties are estimated from correlations using an input of the pseudo's molecular weight (MW),
specific gravity (SG) and boiling point temperature (BPT). The critical properties are required such that the EOS can perform any calculations.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 16 of 54

The MW entered in the PVT Sample object has already been carried over so that only a SG and BPT are required for the critical properties to be estimated
(as the boiling temperature mode if selected as default). If one of the three inputs is not reported in the lab data, then it can be estimated from a correlation.

In our case, the BPT is not reported and will have to be estimated from the provided MW and SG.

Enter the SG from the lab report and press Calculate to estimate the BPT and in turn the critical properties.

Note: the options for AutoMatch of Densities via BPts are selected as default, which was specifically developed by Petroleum Experts.. This means that
when the object is calculated, the AutoMatch process will be performed unless the user decides otherwise.

The Standing Katz method for calculating liquid phase densities is known to provide accurate results near standard conditions. As part of the AutoMatch,
PVTp performs two calculations of the liquid phase density in parallel. On the one hand, the EOS is used to calculate the K-values for a flash to standard
conditions, from which the individual phase compositions and in turn the density can be calculated. On the other hand, the fluid is flashed to standard
conditions using K values from the Hoffmann plot, which are in turn used to determine the phase compositions. With the composition of the liquid phase
known, the Standing-Katz method can be used to calculate the liquid phase density.

Please note that the use of K-values from the Hoffman plot, unlike the EOS, ensures that mass balance is met, however their use is somewhat limited to
ambient conditions, where the tie lines are straight.

When AutoMatch is used, PVTp regresses on the pseudo’s critical properties to match the liquid density predicted by the EOS to that predicted by Standing
-Katz by adjusting the pseudo's boiling point temperature. Changing the BPT, changes the value of the characteristic function (or b-function), such that a
different K-value is determined from the Hoffmann plot (example shown below), which alters the phase composition and therefore the calculated liquid
density from Standing Katz. However, changing the BPT also changes the calculated critical properties (via the chosen correlations) such that different K-
values and therefore phase compositions and liquid densities are calculated. This process is repeated until either the two methods converge, or the
maximum number of iterations is exceeded. For further information on the AutoMatch methodology, please refer to Automatch.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 17 of 54

With the pseudo properties defined, the EOS can technically be used to perform flash calculations, however since the correlations predict generic values,
the calculated fluid behaviour may be quite different to what is observed in the lab and thus the EOS CANNOT be used for predictive modelling. This is
therefore a good point to perform a quality check on the EOS.

Without needing to close any open windows, navigate to the Calculation tab in the Main Menu Ribbon.

Select the Separator object and click on the Calculation Window where the object is to be placed.

Once the Separator object is placed on the window, double click on it to open the screen. Note that object screens will open directly on top of the object
icon though can be moved around to the users preference.

In the window that opens enter the temperature and pressure of each separator stage. This can be done manually by either referring to the lab report itself,
or the PVT Sample object can be opened and the information copied and pasted as required.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 18 of 54

Before the separator train can be calculated, the fluid to be used must first be selected. I.e. each of the Characterisation Objects has an "output" fluid
description or EOS within it, any of which can be used to perform a flash calculation. Therefore PVTp needs to know which of the EOS descriptions is to be
used for this calculation.

Do one of the following:

1. Right click on the Pseudo Properties object and choose Select

2. Choose Select from the Panel Actions tab in the Main Menu Ribbon, then click on the Pseudo Properties object.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 19 of 54

3. As the Pseudo Properties and PVT Fluid objects contain the same EOS, one of the above option can also be done choosing the PVT Fluid
object.

Once the desired object has been selected, it will appear as below:

Now the Separator object can be calculated. To do this Press Calculate within the object.

Choose the Results Table tab and select the Stage Results view.

The calculated liquid density at standard conditions using the EOS of 835.8 kg/m3 compares well with the reported liquid density of 814.5 kg/m3 as this is a
difference of ≈ 2.5%. The calculated first stage GOR of 708 scf/STB and the reported GOR of 661 scf/STB in the lab report, although seemingly very
different are in fact within 10%, which is arguably an acceptable match. This good agreement is a result of AutoMatch feature being used, and if the user
were to repeat the above process again without AutoMatch enabled, the difference would be much larger.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 20 of 54

At this stage we will also want to access the saturation pressure. To do this a Phase Envelope calculation object is needed.

Without needing to close any open windows, navigate to the Calculation tab in the Main Menu Ribbon.

Select the Phase Envelope object and click on the Calculation Window where the object is to be placed.

Once the Phase Envelope object is placed on the window, double click on it to open the screen. Note that object screens will open directly on top of the
object icon though can be moved around to the users preference.

As the EOS within the Psuedo Properties object has already been selected, the Phase Envelope object simply needs to be calculated.

Press Calculate and go to the Results Plot tab to view the phase envelope.

Note: any relevant experimental data previously entered in the PVT Sample object will automatically appear on plots, as can be seen above for the
saturation pressure.

In this case we can see that there is reasonably good agreement with the saturation pressure, as the calculated line is within 2.7% of the lab data point.

That said, from these quality checks the following can be concluded:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 21 of 54

· Liquid density is realistic but requires further tuning to reproduce lab data
· GOR requires further tuning to reproduce lab data
· Psat requires further tuning to reproduce lab data

As the pseudo’s physical properties are typically determined with a large margin of error (± 10%) a match can be better achieved by changing the MW, SG
and BPT.

For example, increasing the MW increases the calculate liquid density, GOR and Psat.

For now, we will focus on reproducing the liquid density and the GOR first, bearing in mind that the GOR carries the greater margin of error, because Psat
can be adjusted at a later stage through splitting of the pseudo and introducing binary interaction coefficients.

To get a feel for how the MW, SG and BPT affect the separator and phase envelope calculation we will want to:

1. Change one parameter at a time

2. Re-calculate the Pseudo Properties data object
3. Re-calculate all of the calculation objects

Please note, this time the Automatch is not performed as this breaks the dependency of the calculated critical properties on the entered MW, SG and BPT
and therefore the impact of changing a parameter might be lost in the automatching process.

With the above analysis complete, try to achieve as close a match between the measured and calculated liquid phase density and GOR by changing the
MW and SG only. For this the following steps could be used.

1. Change a parameter in the Pseudo Properties object

2. Press Calculate to estimate the BPT and critical properties, ensuring that the AutoMatch options are selected to ensure the calculated liquid
phase density is realistic
3. Press Calculate Objects to re-calculate all the calculation objects at once for quality checking, i.e. Separator and Phase Envelope objects to
compare calculated to measured results

4. Repeat if necessary

One possible combination that fits the data well is shown below, though it should be noted that there are multiple solutions that could work equally well. To
ensure comparable results are achieved please enter a MW of 230 g/mol and a SG of 0.85, as otherwise the results of all subsequent steps will be different
from this guide.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 22 of 54

Pseudo Splitting
The EOS now reproduces the liquid density within about 0.7% and the first stage GOR within about 0.6%. This is sufficiently close that the remainder can
be achieved via the regression by adjusting for example the volume shift. The Psat however is only around 1987 psig compared to 2076 psig in the lab
report and requires further tuning. As such the aim for the next step will be to expand the phase envelope, whilst maintaining the match on the liquid density
and GOR.

Before introducing Binary Interaction Coefficients (BICs), we will first attempt to expand the phase envelope and increase Psat by splitting the pseudo into
smaller constituent pseudo components. For this, we will first use correlations to estimate a composition profile and then lump these constituent pseudo
components together into a number of pseudo components.

Introducing additional pseudo components, increases the granularity, i.e. increasing the phase envelope as there are now components with higher Tcs and
lower Pcs than before, and also the degrees of freedom giving the regression more flexibility to perform the matching.

Open the Pseudo Properties object by double clicking on it. Go to the Split Pseudo tab.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 23 of 54

Note that pseudo splitting is initially disabled, so in order to evaluate splitting options this will need to be enabled.

To get a feel for how the shape of the composition and the number of constituent pseudo components affect the phase envelope, and separator calculation

1. Set the split to Last Pseudo

2. Change one parameter at a time (Method/Alpha factor)
3. Choose the preferred profile
4. Move to the Split tab
5. Try different Split Numbers
6. Recalculate the Pseudo Properties object.
7. Recalculate all the calculation objects to quality check the split (Separator and Phase Envelope objects)

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 24 of 54

Initially, the aim will be to try and reproduce the compositional profile from the lab report, as this will ensure that the composition is consistent, which is
especially important when performing compositional tracking. However, it has to be remembered at the components do not represent any real-life counter
parts, as they rather represent a group of components, and as such it is possible to divert from the reported profile. That said, the original profile, is usually
a good starting point.

In our case however, the composition is only reported up until C6 and all subsequent components are lumped into the C7+ component. This gives us
relative freedom with the choice of profile that we can follow, but we should aim to have a C7 composition that is just below the reported composition of C6.

Through this investigation, it can be found that with the Whitson method it is possible to produce more complex profile shapes, however in this case the
most natural seeming trend is produced using the Petroleum Experts 2 method.

Therefore, select Petroleum Experts 2 as the profile method in the Profile tab.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 25 of 54

Next split the original Pseudo into 6 constituent pseudo components.

Go to the Split tab, enter a Split Number of 6. Check AutoMatch to adjust the critical properties such that the calculated liquid density is the same as that
calculated with the original pseudo component only. Press Calculate.

Press Calculate Objects.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 26 of 54

Review the Phase Envelope and Separator calculation object results.

Pseudo Split Quality Check

As an additional quality check, the degree of normalisation applied to the split profile should be verified before accepting the pseudo split. The reason for
this is that the calculated MW and SG of the constituent components may not add up to that of the original pseudo component, so that scaling is applied
when they do not. This is particularly problematic when using the Whitson method. Excessive normalisation is not desirable as it can result in the natural
MW and SG trends from the correlations not being maintained.

To check the degree of normalisation applied, the normalisation first needs to be disabled. In the Split Pseudo tab select the No Values Adjusted mode for
MW and SG normalisation respectively. In the table at the bottom of the screen, it can now be seen that the entered pseudo properties and the calculated
totals are not in perfect agreement.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 27 of 54

Since in this case the differences themselves are small, the degree of normalisation is acceptable. That said, if the differences were significant then the
user may wish to review the splitting method.

The Select Best function can be used to try and automatically reduce the differences by automatically selecting the best split settings.

For the example, switch the Properties Normalisation modes back to the default of All Values Adjusted.

Binary Interaction Coefficients

With the chosen Pseudo split, the EOS now reproduces the reported liquid density, GOR and Psat at reservoir conditions. In fact, the calculated Psat
actually slightly exceeds the reported, such that additional stretching of the phase envelope is not a requirement.

With this in mind, there is no direct need for the use of binary interaction coefficients (BICs), however they could still be included, albeit small values, to
provide the regression with additional flexibility.

Go to the BIC Input Data tab in the Pseudo Properties object.

Check Add Binary Interaction Coefficients and press Calculate. This will populate the table as per the chosen correlation as below:

Calculate the Phase Envelope object to check is the envelope is physical, and it should be noted that it is now too large, as expected.

In our case, it was decided to progress with BICs (though not exactly required) and limiting the maximum value of BICs to 0.01. To do this:

Check Set Max Value To and enter 0.005. Press Calculate.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 28 of 54

This results in the following Phase Envelope, and although the calculated saturation pressure is higher than desired, we know that this can be corrected
during the Regression stage.

Pseudo Output Fluid

This final tab of the Pseudo Properties object is the Output Fluid tab. This tab is continually being updated as pseudo component is defined. As can be
seen below, this tab will show all the EOS properties that have been set/defined. An important thing to note here is that this particular interface is read only.
While the user can view all the properties, they cannot be modified. If any variables need to be modified, this can be done within a PVT Fluid object.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 29 of 54

In this particular example, the fluid does not require any manual overriding.

Regression
The regression is the penultimate step of the characterisation process, in which the EOS is fine-tuned to the laboratory experiments. The fine-tuning is
performed by adjusting the critical properties, volume shifts and BICs, such that the difference between the reported and calculated parameters is
minimised.

The overall objective for the regression stage is to not only reproduce the liquid density and GOR but also to tune the EOS, such that the saturation
pressure, at reference conditions and the remaining data generated during the lab experiments, i.e. CCE, separator test, etc, can be reproduced. The
current EOS already gives a relatively close match on the stock tank liquid density, GOR and Psat, however the liquid density at reservoir conditions is
currently not being reproduced and requires further refining.

As such the objectives for the regression are as follows:

1. Match liquid density

2. Match GOR
3. Match Psat
4. Reproduce laboratory experiments

Strategy
To achieve the objectives above, the pseudo properties, i.e. Tc, Pc, BICs and Volume Shifts will be adjusted to minimise the error between the lab and
calculated data. However, since the regression is merely a mathematical tool it is important to first develop a regression strategy, and to quality check the
results after each regression step to prevent unnatural property trends being accepted. A regression strategy could be developed as follows:

1. Create a list of the current objectives and rank them in order of declining importance.
2. Consider what pseudo properties affect each of the objectives and by how much, i.e. how does objective X change by changing Y.
3. Consider how changing the selected properties for regression affect the other objectives.
4. Consider which parameters would be more difficult to readjust at a later stage and thus decide on a suitable anchor point.

However, before we start considering the above, we should take a moment to gain an overview of how well the current EOS reproduces the lab
experiments. As the lab data has already been entered in the PVT Sample object, direct comparisons can already be made. Taking the lab experiments in
turn we will go through the process of add the appropriate Calculation Data Object and running a calculation:

CCE:

Select the Constant Composition Expansion object from the Main Menu Ribbon and place it on the Calculation Window in the desired location.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 30 of 54

Double click on the Constant Composition Expansion object to open it.

In the window that opens, specify the temperature and pressure range over which the simulation is to be performed.

For the CCE we are only interested in the oil density and viscosity and thus it is not required to specify a separator train. If additional data, such as the GOR
had been provided in the lab report, then the same separator train as used in the experiment would need to be specified. This is because the GOR is path
dependent and thus using a different separator train would result in a different GOR to be calculated.

Press Calculate.

Go to the Results Plot tab and view the desired plots by selecting the variables from the Y-axis menus:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 31 of 54

It can be seen that the EOS (within the Pseudo Properties object) reproduces the general trend of the oil density in undersaturated conditions, however the
calculated density is consistently being overestimated.

Similarly the EOS reproduces the general trend of the oil viscosity in undersaturated conditions, although generally speaking the results are not very close
to the reported values.

DiffLib:

Select the Differential Liberation object from the Main Menu Ribbon and place it on the Calculation Window in the desired location, as was done for the
CCE.

Double click on the Differential Liberation object to open it.

Please note, by the nature of the DiffLib experiment, the composition of the fluid remaining in the container will begin to change as the pressure is reduced
below saturation since the gas evolved is removed from the cell. As such, the residual oil volume, which is used to calculate the GOR and Oil FVF for the
DiffLib experiment, is path dependent, i.e. the conditions the fluid is exposed to over the course of the experiment. If slightly different temperatures and
pressure are specified or whole stages are omitted, then the residual oil volume and thus GOR and Oil FVF will be different.

Therefore enter the exact pressure stages used during the DiffLib experiment:

Press Calculate.

Go to the Results Plot tab and view the desired plots by selecting the variables from the Y-axis menus:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 32 of 54

Above saturation pressure the CCE and DiffLib provide equivalent results as the composition is constant since no gas is evolved that could be removed
from the cells. As such, as we already observed in the CCE experiment, whilst the EOS reproduces the general trend, it overestimates the liquid density.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 33 of 54

As can be seen from the graphs above, the calculated GOR and Oil FVF trends are in relatively close agreement with lab report, though not good enough.

SEP:

The Separator object already exists in the Calculation Window as this was created earlier during an earlier step.

Double click on the Separator object to open it and press Calculate to recalculate the results.

Go to the Results Table tab and select the Stage Results view:

The calculated liquid density of 819.9 kg/m3 compares well with the reported value of 814.5 kg/m3 and the calculated GOR of 669 scf/stb also compares well
to the reported value of 661 scf/stb.

In summary, the EOS currently provides an adequate description of most of the fluid's properties, i.e. the liquid density and GOR from the SEP experiment,
the liquid density and viscosity above saturation from the CCE and also the GOR and Oil FVF from the DiffLib experiment. However, a current short coming
is the EOS's inability to reproduce the measured densities below saturation from the DiffLib Experiment.

Now take a moment to consider the these outcomes and the steps above to develop a possible regression strategy. An example of the development of a
possible regression strategy for this fluid can be seen below.

Example:
1. Create a list of the current objectives and rank them in order of declining importance.
From the analysis above it can be concluded that the liquid density below the bubble point in the DiffLib experiment requires the most improvement and
would thus rank highly on our list. Psat, is slightly too high and would thus rank below the saturated liquid density. The other parameters, liquid density
above the bubble point, the liquid density at standard conditions, GOR and Oil FVF, are already quite close to the reported lab data so that significant tuning
is not required and they would therefore rank low. In our case, this list could look as follows:

Importance Property
High Saturated Liquid Density
Saturated Oil Viscosity
Psat
GOR
Low Standard conditions density

2. Consider what pseudo properties affect each of the objectives and by how much, i.e. how does objective X change by changing Y.
We know that of the above, all properties but Psat are affected by Tc, Pc, BICs and VolShift, which is in fact independent of the VolShift. A thus a regression
could be performed only on the VolShift, however as the difference between the measured and calculated density from the DiffLib is quite large at low
pressures, this could result in over-fitting and can sometimes lead to negative densities being calculated under some conditions. As the DiffLib experiment is
path dependent and the liquid densities begin below the bubble point, this could be a sign of the phase behaviour not being fully captured. As such, though
brute force, a possible regression approach would be to regress on all parameters as once.The table above can then be expanded to include the regression
parameters.

Importance Property Parameter

High Saturated Liquid Density Tc, Pc, BICs, VolShift
Saturated Oil Viscosity Vc
Psat Tc, Pc, BICs
GOR Tc, Pc, BICs, VolShift
Low Standard conditions density Tc, Pc, BICs, VolShift

3. Consider how changing the selected properties for regression affect the other objectives.
As the calculated density is higher than the measured, the VolShift would need to be reduced to reduce the liquid density, which in turn would reduce the
calculated GOR and Oil FVF, which we know to be already close to the lab data. Similarly, reducing Tc and increasing Pc should result in lower liquid
density, as more of the lighter components will remain in the liquid phase, however this also has knock on effects on the GOR, Oil FVF and probably most
significantly the saturation pressure. The BICs could compensate for the reduction in Tc and increase in Pc, however the use of large BICs can lead to
significant distortions to the phase envelope and unphysical behaviour.

With the above in mind, it can often be helpful to introduce “anchor points”, to prevent the EOS from drifting too far away from certain parameters (e.g.
Psat). However the use of too many anchor points at any given time reduces the degrees of freedom and may thus prevent the regression to achieve its
primary objective. As such the number of anchor points should be kept to a minimum.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 34 of 54

4. Consider which parameters would be more difficult to readjust at a later stage and thus decide on a suitable anchor point.
Referring back to our example, regressing on the liquid density below saturation alone using Tc, Pc, BICs and VolShift, will affect the liquid density above
saturation, GOR, Psat etc. As we are already regressing using all parameters, it will be difficult to adjust the remaining properties at a later regression stage,
without undoing already made progress. As such Psat, the Separator density and GOR as well as the liquid density at and above saturation would be
suitable anchor points.

In summary, the objective of the initial, and ideally the final, regression step will be to adjust the critical properties, BICs and VolShift, to match the liquid
density below the bubble point, whilst maintaining our match on the the liquid density above the bubble point or at standard conditions, the separator GOR
and Psat.

Best Practice
Before the regression is started it is also worthwhile to discuss best practice during the regression. As such, please refer to the Regression Best Practices
section.

Primary Regression
We will now perform the initial regression on the liquid density, Psat and Separator density and GOR, which act as anachor point, using Tc, Pc, BICs and
VolShift of the pseudo components.

Double click on Add Regression in the Characterisation Wizard. Ensure that the "PVT Sample" is highlighted in the Navigation Tree before doing this.

This will place a new Regression and another PVT Fluid object on the Characterisation Window, connected to the PVT Fluid that has just been defined.

Double click on the Regression object

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 35 of 54

The first tab is the Lab Data Weightings tab. This is where any data points that we do not want to be included in the regression can be excluded and other
data points can be set with specific weightings. The idea being that the selecting fewer data points to match to will give the regression algorithm flexibility
and additional degrees of freedom to achieve a good match.

As the experimental data has already been added to the PVT Sample object, all of the data is automatically populated into the Regression data object.
Taking each experiment in turn:

Saturation Pressure

As the saturation pressure is definitely something that we will want to match to, this data will be left included.

It should be noted at obtaining accurate measurement at the bubble point, especially for near critical fluid is difficult, as the liquid and vapour phase become
hard to differentiate. To avoid these uncertainties, one may wish to instead regress on a fluid density near rather than at Psat.

Constant Composition Expansion

Select the Constant Composition Expansion from the Laboratory Experiment list. Here the liquid density and viscosity are of particular importance. For
the Relative Volume it is the slope of the curve rather than the absolute values that is of greater importance since it is directly linked to the fluid
compressibility.

Any data we are unlikely to use in the regression, such as the relative volume or erroneous data, can be excluded by selecting the relevant data points and
then pressing Exclude. Additionally, we may wish to exclude some data point in order to speed up the regression.

The process of deciding which data points to exclude, can be aided by plotting the entered data as a function of pressure and temperature, to identify
outlier/unphysical behaviour. This can be achieved by opening the PVT Sample and Regression objects side by side. The data can be plotted in the PVT
Sample object, and the respective data points included/excluded in the Regression object.

For example, when plotting the liquid density and zooming in on the density profile above bubble point, it can be seen that the curve appears to be slightly
kinked near the bubble point, see image below. This data point is best best excluded to avoid the EOS fine tuned to reproduce this density profile.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 36 of 54

For the CCE, exclude the following:

· All relative volume points

· The 2nd, 3rd, 4th, 6th, 8th, 10th, 12th, 14th entry of the liquid viscosity
· Every second entry of the oil density.

Differential Liberation

Care should be taken when matching directly to the differential liberation experiment as the results are more difficult to obtain, adjustments to the results to
ensure mass balance are not uncommon, last stage may be bled before equilibrium is established and thermal expansion of the vessel at ambient
conditions can also affect the results.

Similarly to the CCE experiment, the liquid density but also the gas oil ratio and oil formation volume factor are important.

Exclude the following points:

· The 2nd, 3rd, 5th, 8th, 10th and 12th entry of the Oil FVF
· The 2nd to 6th, 9th and 11th entry of the Solution GOR

Weightings can be assigned to data that is less trusted, for example due to inherent experimental inaccuracies, the original mass balance calculation was
off or it is desired for the regression to take smaller steps. To assign a weighting:

1. Select the relevant data points

2. Select the weighting from the drop down box (10, highest, is default).
3. Data points with weightings assigned will appear blue.

Now, assign a weighting of 6 to all entries in the oil density column.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 37 of 54

Separator

The GOR from the first separator stage and the liquid density at standard conditions are of particular importance. The reason for this being that on the first
separator stage most of the gas is evolved (i.e. the uncertainty in the measurement is lower) and because it is easiest to obtain an accurate oil sample at
standard conditions.

By the time we get to the regression stage, the EOS should already be somewhat reproducing the first stage GOR and liquid phase density at standard
conditions and thus do not require significant tuning. Nonetheless they are included in the regression as they act as an anchor point, which prevents the
EOS from drifting too far from the former, however we want them to have a small impact on the error function or otherwise the regression would be too
constrained and thus unable to match the other parameters.

Exclude the following:

1. the Oil FVF

2. The second stage GOR

set the following weightings:

1. The liquid density at standard conditions to a weighting of 7.

2. The first stage GOR to a weighting of 5.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 38 of 54

Once done go to the Regression Inputs tab.

In this tab the select the following match variables:

1. PSAT - Saturation Pressure

2. SEP - Oil Density
3. SEP - Gas Oil Ratio
4. DIFLIB - Oil Density

Set the regression algorithm to Levenberg-Marquardt.

Set the Match Using to Tcs Pcs Afs with Multiplier on Each Property.

Please note, it is generally recommended that the matching is perform on the basis of the CCE rather than the DIFLIB experiment as this provides a more
realistic evolution of the fluid in the reservoir and well. In fact, both experiments provide near identical results at medium to high pressures, especially above
the bubble point where they are in fact equivalent since the composition is constant. This is discussed in more detail in the Appendix.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 39 of 54

Typically, matching is performed against the CCE experiment with the DiffLib being used as a means of validating the fluid characterisation. In this case, as
there is only little data available from the CCE experiment, the DiffLib experiment will be used for matching.

Go to the Regression Variables tab.

Select the Tcs, Pcs, VolShifts and BICs of all pseudo components and in the case of the VolShifts also the pure components except CO2.

The easiest way to do this is to:

1. Set Allow Pure Tc Pc to No.

2. Select All On.
3. Clear the AF
4. Uncheck the CO2 Vol. Shift
5. Ensure the C1 and All Pseudos to Same Value for the BICs

Press Calculate at the bottom of the object.

Wait until the regression is complete. A progress bar will be displayed and in the top right corner the calculated error after each iteration is shown.

Primary Quality Check

Once the regression is complete, i.e. the error is within a tolerance, the iterations have been exceeded or no further improvements can be made, the
Regression Error Details and Output Fluid tabs will become active.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 40 of 54

The liquid density above and below the bubble point, the saturation pressure and GOR have all been matched to a high degree of accuracy as can be seen
by the errors above in the Regression Error Details tab. From this it can be concluded, that we appear to have met all of our primary regression objectives.

Next, it is important to verify that the calculated critical properties (Tc and Pc) follow smooth physical trends. If there are discontinuities such as step
changes this can be as sign of the regression over compensating for another variable or the variable not having a large effect on the regressed property.
This can be investigated by plotting the critical temperature and pressure as a function of the component number.

To do this go to the Output Fluid tab and select the Plot tab.

From here we can plot the critical pressure and temperature profiles, as shown below. This will by default plot the EOS feeding into the Regression object
and the EOS created within the Regression object for comparison.

From the plot above, it can be seen that the critical temperature has not changed significantly and appears to follow a physical trend. The critical pressures
have been slightly reduced and if anything now follow an even more natural profile.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 41 of 54

If the critical properties and/or phase envelope are found to be unphysical or the surface properties change significantly, then the regression strategy
requires review. As part of this review it should be established whether the pre-regression EOS provides a good enough description of the fluid behaviour as
the regression is only intended for fine tuning. If this is not the case then the pseudo properties, the pseudo profile and splitting as well as BICs could be
reviewed. Alternatively, the property causing the behaviour, e.g. unphysical phase envelope due to large BICs, could be omitted from the regression or the
regression could be repeated using only the heaviest components.

Validation
As the error results of the primary regression run are satisfactory, we now need to validate the EOS against all available lab data to ensure that the EOS is
representative of the fluid in a variety of scenarios. The procedure for this validation is to simulate the lab experiments the same as when we previously
quality checked the EOS prior to the primary regression.

The calculation objects for each of the experiments should be recalculated to allow comparison with the before- and after-regression EOS.

This can be done by "selecting" the Regression object, as shown below, by choosing Select from the Panel Actions tab in the Main Menu Ribbon:

Press Calculate Objects at the bottom of the screen. This will calculate all the Calculation Objects for both selected EOS's.

Below we will go through each of the experiments to validate the new EOS.

Phase Envelope

Opening the phase envelope will allow us to compare the phase envelope before and after the regression was performed, as shown below:

As we can see, the "Regression" stream is the new EOS and it reproduces the saturation pressure (at reservoir temperature) very well.

Separator

From the Separator experiment, it can be seen that the liquid density and first stage GOR match well against the lab data, both within a 0.1% difference.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 42 of 54

From the Separator results it is also possible to drill down further into the stage by stage results. Go to the Detailed Results tab.

One can then go through the stages and investigate how the composition of the vapour and liquid phase changes after each stage. Additionally it is possible
to plot the Hoffman quality plot by pressing Quality, to check whether a physically consistent flash is performed. If the lighter components all lie on a
straight line, then this means that a consistent flash is calculated. Heavier components, i.e. the last pseudo components can deviate from the line of best fit,
see Hoffman quality plot for the first separator stage below. Extreme curvature can be a sign of data inconsistencies, however as most components are on
or close to the straight line, this suggests the calculated K values are consistent.

Constant Composition Expansion

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 43 of 54

The liquid density above saturation is reproduced well however the liquid viscosity, especially below saturation is currently not being reproduced and
requires further matching.

Differential Liberation

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 44 of 54

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 45 of 54

The liquid density above and below saturation match closely against the lab data however there is a small deviation between the measured and calculated
Oil FVF and GOR.

In conclusion, the EOS now reproduces all the lab experiments satisfactorily apart from the liquid viscosity below the bubble point. That said, there is also
room to improve the match on GOR and Oil FVF in the differential liberation, however doing so will compromise the match already achieved for the other
properties.

Traditionally the decision to accept a match or try to regress further would have been a challenging one, as no-one wants to risk losing the match they have
worked hard to achieve already by trying to match too heavily and moving into an unphysical space. With the Data Object Approach, this decision can be
easily made, as no EOS is overwritten so the user can choose which EOS is the official final version.

With this in mind we will now perform a second regression to try and improve the overall match of the fluid by including the Oil FVF and GOR from the
Differential Liberation experiment in the regression.

Second Regression
We will now manually add an additional Regression object, however as we want to keep many of the same parameters as previously used, we will do this
by making a copy of the first Regression object.

Right click on the Regression object.

Select Copy from the menu. Right click in an empty space and select Paste.

Change the input connection to the new object to be from the PVT Fluid_1 object. Do this using the Connect function in the Panel Actions tab in the Main
Menu Ribbon.

Once created, right click on the object and select Change Label. Set the new name to "2nd Regression".

Open the 2nd Regression object and go to the Regression Inputs tab. Here we want to leave all the parameters already selected but include the DIFLIB
Oil FVF and Solution GOR. As we do not want to make too many changes, we will set the weighting to medium, as shown below:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 46 of 54

Press Calculate to begin the regression.

Second Validation QC
As before we will want to view the Regression Errors and component critical property distributions as detailed in the Primary Quality Check section.

Second Validation
As before, now is the time to check the state of the EOS against the lab data to hopefully see some improvement on the match we achieved after the first
Regression.

The calculation objects for each of the experiments should now be recalculated to allow comparison of the Regression and 2nd Regression EOS model.

This can be done by "selecting" the 2nd Regression object, as shown below, by choosing Select from the Panel Actions tab in the Main Menu Ribbon:

Press Calculate Objects at the bottom of the screen. This will calculate all the Calculation Objects for both selected EOS's.

Below we will go through each of the experiments to validate the 2nd Regression EOS.

Phase Envelope

Opening the phase envelope will allow us to compare the phase envelope before and after the regression was performed, as shown below:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 47 of 54

As we can see, the "Regression" stream is the new EOS and it reproduces the saturation pressure (at reservoir temperature) very well.

Separator

From the Separator experiment, it can be seen that the liquid density and first stage GOR match well against the lab data, with a 0.06% and 0.3% difference
respectively.

Constant Composition Expansion

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 48 of 54

The liquid density above saturation is reproduced well however the liquid viscosity, especially below saturation is currently not being reproduced and
requires further matching.

Differential Liberation

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 49 of 54

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 50 of 54

The liquid density above and below saturation and Oil FVF match closely against the lab data and there is a marked improvement to the GOR match.

In conclusion, the EOS now reproduces all the lab experiments satisfactorily apart from the liquid viscosity below the bubble point. So we will now perform
a separate Regression for the viscosity.

Viscosity Regression and Validation

As all parameters except for the CCE oil viscosity are now matching, this next section will focus on perfroming a specific regression for the viscosity.

The oil viscosity can be calculated using a number of models, which are explained in further detail under Engineering Theory and Calculations, and the
most commonly used being the Lohrenz Bray Clark model. As the model was originally developed for gases, its predictions for oils are often inaccurate,
meaning that matching against lab data needs to be performed. The matching against lab data can be performed twofold by regressing on the components
critical volumes or on the LBC coefficients used in its formulation. As such, the viscosity can be matched without undoing progress made as part of the
previous regressions on the insitu densities, since the critical volume is not used in any other property calculation.

Matching against different viscosity models such as Pedersen et al. is performed using a multiplier and a shift rather than regressing on the critical
properties. This is because these models are dependent on Tc and Pc instead of Vc and therefore progress made in a previous regression could be
undone.

We will now manually add an additional Regression object, again using the approach of making a copy of the first Regression object. This is simply to make
it a faster process as

Right click on the Regression object.

Select Copy from the menu. Right click in an empty space and select Paste.

Change the input connection to the new object to be from the PVT Fluid_1 object. Do this using the Connect function in the Panel Actions tab in the Main
Menu Ribbon.

Once created, right click on the object and select Change Label. Set the new name to "Viscosity Regression".

Open the Viscosity Regression object and go to the Regression Inputs tab.

Set the Regression On to Viscosity using the drop down menu and check the CCE - Oil Viscosity, as shown below:

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 51 of 54

Go to the Regression Variables tab. Select All On and press Calculate.

Once the regression is complete the Regression Error Details can be viewed, as done previously.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 52 of 54

From the results above, it can be seen that the liquid viscosity has been closely matched to the test data. This can be further verified by recalculating the
CCE experiment and comparing the calculated against the measured viscosity, as shown below.

The calculated oil viscosity provides a good match against the lab data from medium to low pressures, although at higher pressures some deviation from
the lab data is observed. However, as the reservoir pressure is at around 2340 psia a production simulation will not suffer from the deviation at higher
pressures as these pressures will not be encountered naturally in the reservoir.

Additional Validation
More extensive EOS validation in addition to simulating the Lab experiments is recommended because with each additional validation, the confidence the
predictive nature of the characterised EOS is increased. For example, one such additional validation step would be to use it as the fluid description in a
PROSPER model and to test whether well test data can reproduced through time. This is covered in more detail in Example 4.

Regression Summary

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 53 of 54

From the EOS validation procedure, it can be concluded that the EOS has been characterised, as it is now possible to reproduce all lab experiments,
meaning that for the conditions of the experiments a consistent fluid description is provided. Before progressing and pursuing a number of modelling
objectives with the now characterised EOS, let us perform a brief review of the characterisation process.

1. First we performed a review of the lab data available and also its quality, because the characterisation is fundamentally driven by the lab data. We
found that we had sufficient and more importantly consistent data available. This was then loaded into PVTp.
2. The EOS model was chosen and in this case the Peng-Robinson description was selected.
3. The pure components were selected and their composition entered
4. All lab data was imported.
5. Then the pseudo component properties were defined by entering the MW, SG and BPT from the Lab Report
6. A separator experiment was simulated to demonstrate that the EOS in its current form is unable to predict the liquid density or GOR at standard
conditions.
i. The MW, SG, and BPT were adjusted in order to match the reported liquid phase density and GOR using the Automatch procedure and
performing the quality checks after each iteration.
ii. As the EOS was unable to reproduce the Psat, we then split the pseudo by deciding on a composition profile and the number of components
to split this profile into. We retained the match on the liquid density and GOR, by performing the Automatch to the previously determined liquid
density from the original composition and performing the quality checks after each iteration.
iii. To provide the EOS with additional flexibility BICs were introduced, ensuring they are not too large to avoid distortions of the phase envelope.
The quality checks were performed after each iteration.
7. Then the penultimate phase of the characterisation, the regression was performed to fine-tune the EOS against the lab experiments
i. A Regression Strategy was developed to ensure each regression step is performed with a clear objective in mind, to prevent undoing already
made progress at a previous regression stage.
ii. The Lab data was then entered, weightings assigned to less reliable data points and excluding some data from the regression altogether.
iii. The primary regression step was then performed, by regressing on the insitu liquid density from the DiffLib experiment whilst using the
saturation pressure, GOR and Separator liquid density as an anchor point. After the regression, the results were quality checked by checking
the Tc and Pc profiles, the separator test, the Phase Envelope and determining whether the regression objective had been achieved.
iv. A second regression step was performed to fine tune the existing match, and achieved a better overall match.
8. The EOS was fully validated against the lab experiments, by simulating the different lab experiments and comparing the results to the lab report.
The EOS was found to reproduce all Lab Experiments, apart from the oil viscosity from the CCE experiment.
9. Additional regression was then performed to match the calculated against the measured viscosity. After the regression the results were quality
checked and validated by simulating the CCE experiment
10. As the EOS was able to reproduce all Lab Experiments and the critical properties follow a natural trend, it was concluded that the EOS was
successfully characterised.

EOS Export
The final thing to do is to export the EOS so that is can be used in an integrated model. To do this, follow the steps below.

Double click on Add Export in the Characterisation Wizard. Ensure that the "Viscosity Regression" is highlighted in the Navigation Tree before doing this.

This will place a new EOS Export object on the Characterisation Window, connected to the last regression object.

Double click on the EOS Export object. Select the IPM EoS Composition format.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023
Example 1 - Oil Page 54 of 54

Select Browse and map to the location where the EOS is to be saved, giving it an appropriate name.

Once ready, press Export at the bottom of the window to create the selected file format.

What to do next?
The EOS has now been characterised and has been shown to reproduce the Lab experiments, meaning that it can now be applied to achieve a variety of
modelling objectives, some of which are described below.

• Fluid Description
The EOS can now be used to describe the fluid behaviour, i.e. phase changes, and also to calculate its properties, i.e. density, viscosity etc. With this it is
now possible to for example use the EOS as the fluid description in a reservoir, well or process model. For examples on how to build compositional models
in MBAL, PROSPER, GAP and REVEAL, please refer to the Examples in the respective User Guides.

• Flow Assurance
Common Flow Assurance concerns such as Wax/Hydrate and Salt/Scale formation can also be addressed using the newly characterised EOS. For
example the formation envelopes can be calculated to gauge whether Wax, Hydrates and/or Salt formation is likely to be an issue over the field’s life time. If
this is the case, management strategies can be developed to counteract the adverse effects of solids appearance. For example in the case of Hydrate
formation, it is possible to calculate the required composition of an inhibitor, e.g. methanol, to prevent hydrate formation, this and further examples of solids
management are discussed in the Flow Assurance Manual. For examples on how these types of Flow Assurance calculations are performed in PVTp,
please refer to the Hydrates, Wax and Salt examples.

• Lumping Rule
As discussed previously, certain models have different requirements in terms of the granularity of the components modelled explicitly. For example, for a
reservoir simulator typically fewer components are modeled explicitly in order to reduce the models run time, whereas in a process model the full
composition is required to ensure more accurate results. As such, when building a full integrate model, one has two options, either to cater to each models
needs individually or to use a global fluid description and to compromise.

To achieve the former, PVTp can be used to develop a lumping rule, which translates the lumped composition used in the reservoir simulator to the full
composition used in the process model. For an example of this, please refer to Lumping/Delumping.

• Recombination
If the sample in the laboratory was recombined to an incorrect GOR, due to inaccurate measurements in the field, or the measured saturation pressure in
the lab is unphysical in the field, i.e. Psat is higher than measured reservoir pressure, then the fluid composition can be recombined to a Target GOR or
Psat. For an example of this, please refer to Recombination.

• Decontamination
If the sample is thought to have been contaminated during the sampling, i.e. the profile appears unnatural (e.g. peaks in the composition for higher
components), then a decontamination can be performed in PVTp to remove the contaminants from the composition. For an example of how a
decontamination can be performed in PVTp, please refer to Decontamination.

file:///C:/Users/hp/AppData/Local/Temp/~hh7ECD.htm 22-02-2023