Tutorial 4: History matching using ECLIPSE Office and

SimOpt
Introduction
This tutorial examines the integration of an ECLIPSE Office project with the SIS history
matching aid, SimOpt. In this example an existing simulation, along with associated historical
well water-cut data, is imported into ECLIPSE Office to provide a basic project and examine
the match between the simulation model and the well history.
The project is then exported into SimOpt, where the match can be analyzed using the ECLIPSE
gradients option and a regression problem can be set up that is used to improve the history
match.
The matched simulation model is then exported from SimOpt and associated with the ECLIPSE
Office project as a new case.
Prediction simulations are made using the multiple runs options in ECLIPSE Office.
This tutorial has been broken down into stages. Each stage is self-contained and can be used
independently of the full tutorial.
Hint

If you are new to ECLIPSE Office or have not completed this tutorial before you are
advised to complete all stages in the order in which they appear.

This tutorial takes about forty-five minutes to complete.

Stages

"Setting up an ECLIPSE Office project from an existing simulation model" on page 86

"Exporting an ECLIPSE Office project for use in SimOpt" on page 88

"Setting up a history matching project in SimOpt" on page 89

"Analyzing the current history match and designing a regression problem" on page 92

"Exporting a SimOpt simulation model for import into ECLIPSE Office" on page 94

"Multiple prediction runs in ECLIPSE Office" on page 94

Setting up an ECLIPSE Office project from an

existing simulation model

86

Start ECLIPSE Office.

ECLIPSE Office: File | New Project

Call the new project SNARKSIM.

ECLIPSE Office: Case | Import

Import the data-set SNARKSIM.DATA

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt

ECLIPSE Office: Module | Run Manager

Run Manager: Submit | Runs

ECLIPSE Office: Module | Results Viewer

Result Viewer: File | Open Current Case | SUMMARY

10 Select Read All Summary Vectors in the Extract/Load Summary Vectors panel.
11 Select Read All Reports in the Extract/Load Summary Vectors panel.
12 Click on Load in the Extract/Load Summary Vectors panel.
13 Results Viewer: LinePlot | User
14 Select TIME in the X-Axis Vectors list of the User Templates panel.
15 Select WWCT:PROD1 in the Y-Axis list of the User Templates panel.
16 Click on Add to List.
17 Click OK in the User Templates panel.

Comparing simulation results and observed data

Now that the ECLIPSE Office project has been set up and the simulation run, the results can be
compared with the observed well water-cut values.
1

Results Viewer: File | Open Observed | Column Format

Open the file prod1.wwct

Set the line numbers for the column format in the Column Format User Data panel from
their original settings to the required settings as shown in the following table, and then click
on OK .

Table 4.1

Line Numbers when importing Column Format User Data

Line Descriptions

Original Line Number

Required Line Number

Mnemonics

Units

Scale Factors

Well or Group Names

Lgr Name or Number

Local Cell Numbers

First Line of Data

Results Viewer: LinePlot | Observed

Select TIME in the X-Axis Vectors list of the Select Observed Vectors for plotting
panel.

Select WWCT:PROD1 in the Y-Axis Vectors list of the Select Observed Vectors for
plotting panel.

Click on Add to List.

Select Add to Graph.

Click on OK.

ECLIPSE Office 2007.1 User Guide

Tutorials
Tutorial 4: History matching using ECLIPSE Office and SimOpt

87

Discussion
The match between the simulated and historical water-cut data for well PROD1 shows that the
simulation model currently does a reasonable job over the period of the simulation, but has far
too much water produced far to soon in the early history period and insufficient water produced
in the late history period. This could be due to transmissibility across the faults being too high,
or perhaps Z-transmissibility being too high. Stage 3 of this tutorial, "Exporting a SimOpt
simulation model for import into ECLIPSE Office" on page 94, will use SimOpt to examine
these parameters and attempt a better history match.
Before this can be done, however, the simulation model must be made ready for
SimOpt."Exporting an ECLIPSE Office project for use in SimOpt" on page 88 of this tutorial
examines the transition from an ECLIPSE Office project to a SimOpt project.

Exporting an ECLIPSE Office project for use in

SimOpt
If you did not complete Stage 1 before starting this section of the tutorial, you will need to set
up an ECLIPSE Office project with the data file SNARKSIM.DATA.
Hint

If you are not sure how to set up an ECLIPSE Office project, see "Setting up an
ECLIPSE Office project from an existing simulation model" on page 86

The first step in preparing a simulation model for SimOpt is to create a new case in the ECLIPSE
Office project.
1

ECLIPSE Office: Case | Add Case | Clone

This creates a new ECLIPSE Office case that contains the same information as the default
case. The New Case Name panel is displayed.

Note

SimOpt appends the imported data set name with _1 in the file-root name (that is
BASE.DATA becomes BASE_1.DATA). This is so that SimOpt versions of a
simulation model are not confused with the original data set. It is important that the
original dataset file-root name does not end with an underscore and a number; SimOpt
rejects such datasets as they can be confused with the SimOpt versions of the
simulation model.

Change the Case Name to ORIGINAL.

Click on OK.

Click on the ORIGINAL case to make it the active case.

ECLIPSE Office: File | Save Project.

Creating the INIT file

SimOpt requires GRID and INIT files as well as the dataset. By default, the GRID file is
generated by this case. Before exporting the simulation model ORIGINAL from ECLIPSE
Office, we need to create the INIT file.

88

ECLIPSE Office: Module | Data Manager

Data Manager: Section | Grid

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt

Grid Section: Subsection | GRID keywords

GRID Keyword Section: View | Keywords

Select Operational Keywords in the Keyword Type list.

GRID Keyword Section: Edit | Insert Keyword...

Select Output INIT file in the Keyword Selection panel.

Grid Section: File | Save

Click on Yes then Save to save the current contents of the Grid Section.

10 Grid Section: File | Close

11 Data Manager: File | Close
12 ECLIPSE Office: Module | Run Manager
Hint

This run is only to generate an INIT file (there were no changes to the simulation).
The simulation can be turned off to save time.

13 Click on Turn Off Simulation (at the bottom right of the Run Manager).
14 Run Manager: Submit | Runs
15 After the run has finished, the file must be written without the NOSIM keyword, ready for
SimOpt to read it.
16 Click on NOSIM (Turn Off Simulation) so that it is no longer checked.
17 Run Manager: File | Write Data.
18 Close the Run Manager and save the settings.

Discussion
SimOpt requires a working simulation model and a GRID and INIT file to set up a project.
ECLIPSE Office can provide the data-checking and keyword editing facilities to create the files
for a SimOpt history matching project.

Setting up a history matching project in SimOpt

1

Start SimOpt and call the project SNARK.

The Getting Started panel appears. Click on the Specify DATA file button.

Import the file ORIGINAL_E100.DATA.

In the Getting Started panel, click on the Specify GRID file button.

Import the file ORIGINAL_E100.EGRID.

Note

The file ORIGINAL_E100.INIT has also been imported during this step.

SimOpt: Project | Import | Observed Data | Production | Graf User Vector Data
Format

Import the file GRAF.WWCT.

ECLIPSE Office 2007.1 User Guide

Tutorials
Tutorial 4: History matching using ECLIPSE Office and SimOpt

89

Click on the OK button on the message box that warns you that you must set measurement
errors.

Now that the basic data for a SimOpt project has been entered, save the project.
9

SimOpt: Project | Save

Note

The Well Water Cut data (WWCT) that has been imported has no estimate of the error
in the measurement. This information is necessary for SimOpt to calculate the RMS
(root-mean-square) fit between the observed and simulated data, and so must now be
added manually.

10 SimOpt: Observed Data | Control...

11 Select Set Meas. Error on the Block Action (Selected Set) radio button.
12 Enter a value of 0.01 in the Meas. Error text entry field.
13 Click on Apply Block Action.
14 Click on Close.

Sensitivity model
The simulation model and observed data are now ready for sensitivity studies. In this case the
three faults in the model and the Z-transmissibility of the entire grid will be used in the
sensitivity model.
1

SimOpt: Parameter | Add

Select ZTrans in the Parameter Type list and click on Apply.

Select Fault on the Domain radio button.

Select FaultTr in the Parameter Type list.

Click on FAULTS1 in the Domain Names list and then Shift+Click on FAULTS3 to
select all three faults.

Click on Apply.

Click on Close

Click in the + sign to the left of the Version 1 node in the Parameter Versions hierarchy
tree (that is, the central panel on the main screen).

Defining the transmissibilty multipliers

The fault transmissibility multipliers act along the entire length of each fault. The Ztransmissibility multiplier, however, acts on a number of grid cells and so its region of activity
must be defined.
1

SimOpt: Parameter | IJK Region Editor...

IJK Region Editor: Region | Define | Using Box(es)...

Now we define the region to cover the entire grid:

90

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt

Enter the following data in row 1 of the table and then click on OK:

I1(nI:14)

I2(nI:14)

J1(nJ:14)

J2(nJ:14)

K1(nK:5)

K2(nK:5)

14

14

Click on OK.

IJK Region Editor: File | Close

Defining starting values

Now that the parameters of interest have been defined, their starting values and upper and lower
bounds must be set.
1

SimOpt: Parameter | Control...

Hint

The fault transmissibility multiplier varies between 0 (a sealing fault) and 1 (an
invisible fault). If there is no initial estimate of the multipliers value, the usual
situation, the following rule of thumb can be applied:
A change in the transmissibility of a near-sealing fault generally has a much greater
effect on reservoir flow (and thus greater sensitivity) than a change in the
transmissibility of a near-invisible fault. Consequently, it is usually useful to set the
initial value of the fault modifiers to, say, 0.1 and allow the value to vary between 0
and 1.

Hint

Vertical transmissibility estimates can be in error by a factor of 100 or more.

Consequently it is usually wise to allow the Z transmissibility multiplier to vary from
the original simulation model value (that is a multiplier of 1) down from and up to 100
hundred times the original value, that is from 0.01 to 100.

Enter the following data in the table in the Parameter Control panel:

Table 4.2

Parameter control starting values

Type

Domain

Modifier

Modifier Min.

Modifier Max.

ZTrans

Region1

0.01

100

FaultTr

FAULTS1

0.01

1E-5

FaultTr

FAULTS2

0.01

1E-5

FaultTr

FAULTS3

0.01

1E-5

Hint

The most efficient way to change the modifier value, minimum value and maximum
value for the fault transmissibility parameters is to use the Type filter to make the table
display only the parameters of type FaultTr, then use the block actions Set Modifier
Value, Set Minimum Value and Set Maximum Value to make the required changes
to all the fault transmissibility parameters at the same time.

Click on Apply in the Parameter Control panel.

Click on Close.

ECLIPSE Office 2007.1 User Guide

Tutorials
Tutorial 4: History matching using ECLIPSE Office and SimOpt

91

Discussion
A project in SimOpt contains a simulation model, GRID and INIT files, some observed well
histories (with standard errors for the observations), and parameters that are varied to try to
improve the history match.
The project is now complete, and so sensitivities can be calculated. By taking advantage of the
Gradient option in ECLIPSE, the sensitivity analysis can be performed in a single simulation
run. The next section examines the analysis of sensitivities to provide a basis for regression to
improve the history match.

Analyzing the current history match and

designing a regression problem
1

SimOpt: Analysis | Simulate

Select Gradient from the Run Mode radio buttons and then click on Start (and click on
Yes when prompted to save the project).

Hint

To view the simulation run while it is in progress, click, with the right mouse button,
on the node for well PROD1 in the Observed Data menu and select Plot. This plots
the observed data for well PROD1 with the simulation superimposed.

When the simulation has finished, look in the log window at the bottom of the main
window.
The RMS fit between all the observed and simulated data is around 24.

Note

The multipliers for FAULTS2 and FAULTS3 are marked as redundant. This means that
they are not important to the fitting of the data and so can be turned off.

SimOpt: Parameter | Control

Set the activity of FAULTS2 and FAULTS3 to Semi-active using the drop-down menus in
the Activity column of the table.

Note

The semi-active activity means that these parameters are no longer used in the
analysis and their gradients are no longer calculated during a simulation, but their
current modifier values are still used on the model. The only difference when using the
alternative inactive activity is that the current modifier values are not used.

Click on Apply

Click on Close.

Reviewing the observed data and remaining parameters

Now that the redundant parameters have been removed from the analysis, the remaining
parameters can be viewed.

92

SimOpt: Analysis | Match

Click on Update.

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt

The RMS folder shows that the RMS and sensitivities of well PROD2 are zero. This is because
well PROD2 shuts at time zero so the simulation will always perfectly match the (zero water
cut) observed values. Including this observed data in the history match is pointless; it only
serves to make the overall match (as reported by the RMS) look better than it really is. For this
reason, this observed data should be taken out of the history match problem.
Note

In general, all observed data in the history matching project should be examined
carefully to make sure that it really contributes to the history matching problem.

SimOpt: Observed Data | Control...

In the table, set the Status of the WWCT data for PROD2 to Inactive. Click on Apply.

Note

This does not delete the data from the project. It merely removes it from the RMS
calculation. The data can be added back into the RMS calculation at a later stage by
setting its status to active.

Click on Close.

Match analysis panel

1

Click on Update.
This recalculates the analysis without including the PROD2 data.

Click on the Correlation folder.

The off-diagonal elements of the correlation matrix are around 0.33, and so the two
remaining parameters are not strongly correlated. If parameters are strongly correlated, the
problem may be mathematically ill-posed (which means that regression may not improve
the match). In this case the problem seems well-posed and so it is possible to use non-linear
regression to iteratively update the modifier values to improve the history match.

Note

It is important to remember that the analysis is linear and that the response of the
simulation model to changes in the multipliers will be non-linear. Consequently, it is
unwise to regress for more than 3 or 4 iterations, as each iteration takes the simulation
model further away from the point at which the analysis was valid.

SimOpt: Analysis | Simulate

Select Regression in the Run Mode radio button.

Set the Max Runs Simulator to 3

Click on Update parameters (and click on Yes in the warning dialog that appears). This
updates the parameter values using the gradients already calculated for the parameters.

Click on Start (and click on Yes when prompted to save the project).

Hint

If you want to view the progress of the RMS during the regression, click on the Plot
RMS button. (This updates after each simulation run). As before, runtime monitoring
of well data is available.

The final RMS fit is 11.7, a big improvement over the ORIGINAL data-set imported into
SimOpt.

ECLIPSE Office 2007.1 User Guide

Tutorials
Tutorial 4: History matching using ECLIPSE Office and SimOpt

93

To see the match between observed and simulation data, select the PROD1 WWCT node
on the Observed Data hierarchy tree.

Select Observed Data | Plot to see the improved history match

Discussion
Following a sensitivity run, the gradients were analyzed using tools available in SimOpt. This
analysis facilitated the design of a well-conditioned regression problem. SimOpt was then used
to run the ECLIPSE simulator, iteratively updating the selected parameters to provide an
improved history match.

Exporting a SimOpt simulation model for import

into ECLIPSE Office
The new simulation model that resulted from history matching can now be exported from
SimOpt and used to make prediction runs using the multiple runs feature in ECLIPSE Office.
1

SimOpt: Project | Export | Simulation Model

Click on OK in the Project Export Comments panel.

Save the data-set with the name MATCHED.DATA

Close the file viewer displaying the exported DATA file.

SimOpt: Project | Exit

Click on Yes to save the project.

Discussion
The matched simulation (that is with the multipliers applied) can be exported from SimOpt. This
is the original simulation with multiplier keywords added so that, at a later stage, the original
simulation model can, if required, be retrieved.

Multiple prediction runs in ECLIPSE Office

1

ECLIPSE Office: File | New Project

ECLIPSE Office: Case | Add Case | New

Change the Case Name to PREDICTION and select OK.

Click on the PREDICTION case to make it the active case.

ECLIPSE Office: Case | Import...

Import MATCHED.DATA.

Adding a prediction period

Now that the history matched version of the simulation model is in ECLIPSE Office, the next
step is to increase the length of the simulation (that is add a prediction period).

94

ECLIPSE Office: Module | Data Manager

Data Manager: Section | Schedule

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt

Select the last date in the data-set (1 January 2030) in the Time - Dates list.

Schedule Section: Time | Insert

Select Time Step on the New Time Entry Choice radio button.

Enter 365 days as the Time Step.

Enter 5 in the Num entry box of the New Time panel.

Click on OK.
This adds five simulation timesteps, each a year in length, to the existing data-set.

Defining the rate controls

The next step is to define rate controls for the prediction period.
1

Schedule Section: Event | New

Select Well Controls and Limits in the Event Types list

Select Production Well Control in the Events list

Click on OK.

Enter the following information in the Schedule Section fields for the new WCONPROD
keyword:

Well

PROD1

Open/Shut flag

Open

Control

ORAT

Oil Rate

1000 stb/day

Click on the Multiple Runs button.

In the Oil Rate column, enter a value of 3000 stb/day in the second row of the
Production Well Control table.

Click on OK in the Production Well Control panel.

Click on Apply in the Schedule Section panel.

10 Schedule Section: File | Save

11 Save the Schedule and Multiple Sensitivities file.
12 Schedule Section: File | Close.

Prediction runs
The multiple prediction runs are defined, so the prediction runs can now be made.
1

Data Manager: File | Close

ECLIPSE Office: Module | Run Manager

Ensure that the NOSIM (Turn off Simulation) checkbox is not checked because the
simulation run is required for generating Summary Vector Reports.

Run Manager: Submit | Runs

Run Manager: File | Close

ECLIPSE Office 2007.1 User Guide

Tutorials
Tutorial 4: History matching using ECLIPSE Office and SimOpt

95

Viewing results
Once the simulations have finished, the results can be seen using the Result Viewer.
1

ECLIPSE Office: Module | Results Viewer

Results Viewer: File | Open Current Case | Summary

Select both cases (to select a case, click on it and then click on the >> button).

Click on OK.

Select Read All Summary Vectors and Read All Reports in the Extract/Load
Summary Vectors panel

Click on Load

Results Viewer: Tools | Calculate Totals | Well Production

Select All in the Names list

Select PROD1 in the WGNames list

10 Click on Add to List

11 Click on OK.
The graph shows the oil and water production totals for the two prediction runs. It can be
seen that improved total production is achieved in scenario 2 (3000 stb/day).

Discussion
In this final stage, multiple prediction runs were made in ECLIPSE Office so that the effects of
different well controls on future production could be determined.

96

Tutorials
ECLIPSE Office 2007.1 User Guide
Tutorial 4: History matching using ECLIPSE Office and SimOpt