Special topics
This integrated well modeling tutorial combines existing data with consistent
engineering assumptions to develop a solid well bore model using PROSPER. The
developed well bore model is then used to simulate the performance of the well under
possible future operating conditions with different GOR, water cut and reservoir
pressure. For this, sensitivity runs are performed with different GOR, water cut and
reservoir pressure. The approach can be easily extended to any other possible
sensitivity variable: tubing size, skin, permeability, thickness and so on.
3.1.1.1 Objectives
This tutorial demonstrates how to analyse the performance of an offshore oil producing
well in an integrated fashion. In the process, the following questions are addressed:
- What are the basic data required to build a PROSPER model for a naturally flowing oil
well,
- How to match a black oil PVT against black oil correlations,
- How to tune the overall heat transfer coefficient in the surroundings of a well based
upon well test data
- How to quality-check raw well test data,
- Why and how to perform a correlation comparison,
- How to perform a Verrtical Lift Perfoamnce (VLP) matching,
- How to select a VLP correlation,
- How to model a highly deviated well,
- How to calibrate the well inflow with the well test data,
- How to validate a PROSPER well model against actual well test data,
- How to match multiple well tests,
- How to use the calibrated PROSPER model to predict the future performance of a
given well bore.
3.1.1.2 Statement of the Problem
A naturally flowing highly deviated oil producing well has been drilled and tested.
Extensive fluid and well test data are available.
It is required to develop a PROSPER model that is calibrated against PVT lab data and
well test data.
Once the model is built and calibrated, it will be used for sensitivity run.
3.1.1.3 General Approach
In this tutorial, the following road map will be followed:
a. Construct a PROSPER well model with a matched and validated PVT model
b. Analyse and match the well test data against the PROSPER model to obtain a
calibrated well bore model
c. Use the calibrated well bore model to simulate the performance of the well under
changing conditions: water cut, GOR and reservoir pressure.
3.1.1.4 Available Data & Information
The data required for a well test analysis are:
- PVT data
- Reservoir Data
- Well Data: deviation survey, tubing, casing, and temperature data
- Actual production test data: stabilized phase rates, flowing temperatures and
pressures.
When building a PROSPER well model, there are in general two possible situations
to consider with respect to the PVT data. There can be a situation where enough
PVT data are available. The second contrasting situation arise when there is
If comprehensive black oil PVT data are available, then it is recommended to match
the PVT data to the black oil correlations.
The advantages of matching the black oil data to correlations are:
- The PVT matching process offers an opportunity to check the quality of the PVT
data themselves by analyzing the magnitude of the correction required to
reproduce the various PVT data: GOR, Oil FVF, Bubble point pressure and oil
viscosity.
- At the end of the PVT matching process, the most suitable black oil correlation for
the particular crude is selected and used for further calculations.
- Unlike PVT look-up tables, carefully matched black oil correlations may generally
be used for operating conditions beyond the values used for the PVT matching.
In the event no extensive PVT data or no reliable PVT data are available, the black
oil correlation may generally be used directly without any matching to start with.
The selection of the appropriate black oil correlation is a matter of engineering
judgment that is at the discretion of the project engineer. For details about the
published validity range of each black oil correlation in PROSPER, the relevant
references listed in the appendix A of this manual may be consulted.
It is strongly recommended to match the black oil correlations against lab
measurements as soon as a PVT report becomes available.
The purpose of each specific study determines the type of reservoir model to be
selected and consequently the parameters required.
In the absence of typical reservoir properties like permeability, net pay, skin and so
on, the simple Productivity Index (PI entry ) or its equivalent for gas (C & n) may
be used.
If extensive reservoir and skin parameters are available, then an alternative inflow
model that better fits the purpose of the study is to be considered.
The well equipment data like deviation survey, surface equipment, down hole
equipment and geothermal gradient are generally available from well bore
schematic and temperature logs.
The overall heat transfer coefficient in the surroundings of the well bore can be either
computed with the enthalpy balance model or derived from well test data.
When the rough approximation method is being used, the following overall heat
transfer coefficients values may be considered as starting values:
It is recommended to fine-tune the overall heat transfer coefficient with well test data
as soon as well test are available.
It is recommended to keep the default average heat capacities displayed in the
program under | System | Equipment | Average Heat Capacities unaltered.
Note that it is perfectly possible to develop a PROSPER model in the absence of well
test data. In this case, reasonable assumptions are to be made based upon the
specific operating conditions of the well and based upon existing relevant
experiences.
Generally, the use of the Petroleum Experts 2 multiphase correlation may be
considered when no well test data are available for the simple reason that the
Petroleum Experts 2 multiphase correlation has so far consistently given
reasonable results in very diverse operating conditions.
However, since there is no universal multiphase correlation, it is recommended to
verify the suitability of the selected multiphase correlation (Petroleum Experts 2)
as soon as well test data become available.
PROSPER offers a coherent and consistent VLP matching procedure that helps the
user to select the most suitable correlation for any particular well. This consistent
VLP matching procedure will be described and applied later in this tutorial in the
sub-section "Well Test Analysis: Step by Step Procedure".
Surface Data
Note that the fluid properties may originate from a thermodynamic simulation (Constant
Mass Expansion with separator correction) using a compositional fluid calibration
package like PVTp.
3.1.1.4.2 Well Equipment Data (Tubing etc)
Deviation survey
The deviation survey is the reference for all subsequent depths inputs in the equipment
section.
Surface Equipment
This well model will not include pipelines downstream of the well head.
Geothermal Gradient
*Using the Wong-Clifford skin model, the deviation angle of the well well bore across
the reservoir is calculated based upon the user-entered perforation intervals in
Launch PROSPER, select | Options | Options and make the following choices:
Select | PVT | Input Data and populate the PVT entry screen as follow:
Surface Data
In order to enter the lab data, select | Match Data and populate the screen as shown
below:
PVT matching
For the PVT matching, select | Regression | Match All | OK | Parameters and this is what
can be seen:
This screen shows the values of the Match Parameters (Parameter 1= multiplier and
Parameter 2 = shift) associated with each fluid property for all the correlations.
The Standard Deviation is also displayed which represents the overall goodness of fit.
Select an appropriate correlation bearing in mind that the black oil correlations whose
parameters are adjusted the least (i.e. Parameter 1 close to 1.0 and Parameter 2 close
to 0.0) are the most suitable candidates as they represent possibly similar fluids to the
one being modelled.
The value of the Standard Deviation is also displayed. Generally, the lower the Standard
Deviation, the better the fit.
In the case at hand, the combination Glaso / Beal et al will be selected because they
require the least adjustment. For this, select | Done | Done and make sure that the black
oil correlations Glaso / Beal have been selected on the PVT- Input Data screen:
Select | Done and save the file to secure the data inputted into the PROSPER well bore
model.
Select | System | Equipment (Tubing etc) | All | Edit to start the well bore description with
the deviation survey:
Deviation survey
4112 4000
5845 5500
7800 7000
10135 8500
11135 9000 Top of perforation
The deviation survey is the reference for all subsequent depths inputs in the equipment
section.
Surface Equipment
This well model will not include pipelines downstream of the well head.
Therefore select | Cancel to skip this screen:
Simply point the cursor on the relevant cell within the column labeled "Type" underneath
the "Xmas Tree" to select the equipment type from a drop-down menu: tubing, SSSV,
Restriction or casing.
Geothermal Gradient
To visualize the well bore sketch, select | Summary | Draw Down Hole and the next
sketch is generated:
For the selection of the inflow performance model, select | System | Inflow Performance
and make the following choices:
Select the Input Data Tab in the top right corner to continue with the reservoir data
inputs:
Reservoir Permeability: 100 mD
Reservoir Thickness: 100 ft
Drainage area: 350 acres
Dietz shape factor: 31.6
Well bore Radius: 0.354 ft
Select the | Mech / Geom Skin tab at the bottom of the screen to proceed with the skin
data entry:
Mechanical skin: 0
Enable Wong-Clifford: ON
Select the next tab "Dev / PP Skin" at the bottom of the screen to enter the parameters
required to computed the deviation and partial skin using the Wong-Clifford method:
In order to generate an Inflow curve, simply select | Calculate and the following IPR curve
is generated:
PROSPER well bore model) and actual measurements. The reconciliation process can
only be successful if the mathematical model and the actual measurements are both
reliable.
The various sequences followed so far to construct the PROSPER model have been
specifically designed to lead to a reliable, sound PROSPER well bore model.
The main purpose of the critical review of the raw well test data is to assess the validity
of each well test data before entering them into PROSPER.
A few of the questions to be addressed in the critical data review process are:
- How reliable is each reported measurement?
- How do the test data compare with historical trends?
- How does the produced GOR compares with the PVT model?
Let us critically review the well test data to be analysed in this tutorial:
Dates Comm WHF WHF Water Liquid Gauge Gas Oil Ratio
ent P T Cut Rate Pressure at
gauge depth
dd/mm/ psig degF % stb/d psig scf/stb
yyyy
01/01/20 Low 1000 150 25 6100 3655 1500
00 Rate
02/01/20 Medium 800 180 25 9800 3505 500
00 Rate
03/01/20 High 500 200 25 13450 3365 475
00 Rate
In order to enter the well test data, select | Matching | Matching | VLP/IPR (Quality
Check).
When analyzing multiple well tests, it is good practice to estimate the U-value for each
For, this select the task button "Estimate U Value" and the following sub-screen is
shown:
For this, simply select the test by clicking on the corresponding row number.
Afterwards, select | Correlation Comparison | OK and then select the correlations Duns
and Ros Modified, Fancher-Brown, Petroleum Experts 2 and Petroleum Experts 5 for
instance:
Proceed with | Calculate | Calculate | OK | Plot and the following plot is shown:
One can see that the test point is completely outside established physical boundaries.
The physical boundaries are defined by the Fancher-Brown correlation that neglects
slippage effects between the phases and consequently under-estimate the pressure
loss and the Duns and Ros Modified correlation that generally over-estimates pressure
drops in oil wells.
During the critical review of the well test data, it was suspected that the reported GOR
values were inconsistent.
Indeed, with a bubble point pressure of 3400 psig and a reservoir pressure of 4000
psig, the produced GOR must be equal to the solution GOR of 700 scf/stb.
Therefore we will go back and correct all GOR values to 700 scf/stb as show below:
After correcting the GOR, one can now re-estimate the U-value for each test. For this
simply select | Estimate U-value and the algorithm will estimate and display the U value
for each test. In the end, the average U-value is calculate and displayed:
We will select | Yes and | OK to transfer the averaged U-value for all 3 tests to the
geothermal gradient screen.
Based upon engineering judgment, one may very well not accept the averaged U-value
and use an alternative one.
Here are the correlation comparison plots obtained for each test:
Because the Petroleum Experts 2 correlation consistently reproduces each well test
with reasonable accuracy, this particular correlation will be selected for further analysis.
It is possible to match multiple well tests simultaneously in PROSPER. For this, one
simply need to keep all well tests enabled and the VLP matching algorithm will attempt
to match all enabled well tests simultaneously.
In the following, a better approach is applied: match one test and verify how the match
performs against non-matched well tests.
For this, return to the VLP/IPR matching screen, select the most reliable well test. The
choice of the most reliable well test is a matter of judgment that depends upon all the
circumstances surrounding the test. In this example, it will be assumed that the well test
with the highest flow rate is the more stable for the simple reason that it is more likely to
be stable, to be in the friction dominated region of the tubing performance curve.
To perform the match, select the "High Rate" well test (here on row number 3) and then
select | Match VLP as highlighted in the screen-shot below:
If no specific well test is selected, the VLP matching algorithm in PROSPER will match
all well tests simultaneously. A possible drawback of this approach is that possibly
invalid well tests may not be easily detected.
Note that the VLP correlation "Petroleum Experts 2" appears now appended with two
numbers: gravity multiplier and friction multiplier. Both numbers are equal to unity
suggesting that virtually no correction was required to match the well test.
Now select | Done to return to the previous screen.
To benchmark the VLP matching against the two other well tests, simply enable them
and then perform correlation comparisons for each well test in turn.
For this, enable the well tests, select each well test and then select | Correlation
Comparison | Ok | Calculate | Calculate | Plot.
This is how the correlation comparison looks like for the "Low Rate Test":
And this is how the correlation comparison looks like for the "Medium Rate Test":
In both cases, one can see that the matched correlation reproduces very well the other
well tests.
This fact validates the suitability of the correlation for use in predictive mode.
In order to superimpose the IPR, return back to the VLP/IPR matching screen and then
select the task button VLP/IPR.
Please confirm the completion of the Calculation with | OK and the the screen is
populated with calculation results as shown below:
For each well test, actual test rate and simulated (solution) test rate are calculated and
displayed in th bottom right corner of the screen.
A graphical plot is generated when the button Plot is selected:
The plot above shows for each well test, the IPR, the VLP the test point and the
simulated solution.
In the right corner of the screen, one can see the statistical comparison between
measured test rate and test bottom hole flowing pressure versus simulated rates and
pressures.
The largest error is less than 2% and can be considered as acceptable.
Now one can select | Main | File | Save to update the PROSPER model.
Parameters: Values
Well head flowing pressure (psig): 300
In order to perform the sensitivity runs, select | Calculation | System (Ipr + Vlp) | 3
Variables and make the following entries:
Select | Continue and choose the relevant sensitivity variables water cut and reservoir
pressure using the drop down menu:
Select | Continue | Calculate | OK | Plot | System Plot and the following graphical
representation is generated:
This plot shows the liquid rate as function of water cut for different reservoir pressures.
This completes this tutorial.