New Features in CMG 2010 Software

New Features in CMG 2010 Software

By Computer Modelling Group Ltd.

New Features in CMG 2010 Software 2

Contents

Contents ......................................................................................................................................... 2

ew Features in IMEX 2010.10 ................................................................................................... 3

ew Features in GEM 2010.10 .................................................................................................... 6

ew Features in STARS 2010.10 ............................................................................................... 10

ew Features in CMOST 2010.10 ............................................................................................. 16

ew Features in Builder and Results 2010.10 .......................................................................... 23

ew Features in Launcher 2010.10 ........................................................................................... 26

ew Features in WinProp 2010.10 ............................................................................................ 27


New Features in IMEX 2010.10

Important Changes between IMEX 2010.10 and IMEX 2009.10

Completion Lumping Option

*LAYERCLUMP provides a way for the user to define a control lump - a set of layers

(perforations, completions) which possesses a name and a set of computed rate,

cumulative, and stream ratio values. To each control lump there is also an associated

setting value.

*LAYERRLUMP provides a way for the user to lump well layers in a group called a

report lump. For report lumps, stream rates, cumulatives and production rate ratios are

calculated and the user can design triggers based on these quantities. A given layer can be

assigned a partial membership to report lump and the same layer can belong to more than

one group.

*RM-LAYERCLUMP removes membership of a layer from the named control lump.

*RM-LAYERRLUMP removes membership of a layer from the named report lump.

*CLUMPSETTING assigns a control setting or flow restriction parameter to the named

control lumps. The flow restriction parameter is used as a multiplication factor in

adjusting the value of the well index for all the layers belonging to a given control lump.

There is also a set of computed rate, cumulative, and stream ratio values for each control

lump. Triggers may be defined using these quantities specific to named control lumps;

actions can be any valid well data lines, including lines which alter the control lump

setting values. The syntax of the control lump based trigger statement is:

*TRIGGER trigger_name *ON_CTRLLUMP control_lump_name quantity > value

The syntax of the report lump based trigger statement is:

*TRIGGER trigger_name *ON_RPTLUMP report_lump_name quantity > value


See *LAYERCLUMP keyword for a list of quantities which can be used in control and

report lump triggers.

*WHYSTAB *AUTOWELL Monitor

*MONITOR WHYSTAB *AUTOWELL frequency can be used to monitor production

well wellbore hydraulics stability using the *AUTOWELL option, which shuts the well

for a specified time.

*MXCNRPT Maximum Number of Continue Repeats

Every well having *CONT *REPEAT (continue-repeat) specified through *OPERATE is

entitled to one repeat of the current time step if one of its operating constraints is

violated. During convergence of a time step, it is possible that different wells violate and

then switch their operating constraints consecutively after each repeat. Although the

maximum number of possible repeat cannot exceed the number of wells, it may lead to a

significant increase of run time if there are many wells. Quantity nmxrpt limits the

number of such repeats per time step, and keyword *MXCNRPT lets the user override its

default value (of 1).

*KRPERF *SORMAX Entry for Trapped Oil Hysteresis

SORMAX on the *KRPERF keyword allow the user to change this saturation endpoint in

recurrent data. Sormax is the oil saturation in the two phase oil-water system at which oil

first becomes immobile during the boundary/primary imbibition process (the imbibition

starting from oil saturation equal to 1-Swcon). It only has an effect when the Krow

hysteresis or trapped oil hysteresis is modeled for the rock type of the completion.

Improved Parallel Performance

Improvements to the implementation of parallelization in IMEX have allowed the

Simulator to run more efficiently on larger number of threads on Shared Memory

Processors.

Trapped Oil Hysteresis Option

Trapped oil hysteresis is now available for Pcow and Krow curves. To invoke the trapped

oil hysteresis option, it is only necessary to define the imbibition Pcow in an SWT table

using constant Pcow for water saturation values equal to and great than 1-Sormax. Once

activated, trapped oil hysteresis couples Pcow and Krow hysteresis. Keyword *HYSKRO

is not necessary and will be ignored. The endpoint array *SORMAX has been added to

allow the user to enter values of Sormax on a block by block basis. The keyword

*DAMP-PCOW-TROIL has been added to reduce the capillary pressure difference

between imbibition and drainage curve which might result in a pressure oscillation.


Please see the Tutorial section: The Trapped Oil Hysteresis Option for Pcow and Krow

(Oil-Water System)

IMEX-GAP Interface Improvements

Handles Crossflow during Shutin (like *MODELSHUT)

When wells are shutin by GAP but experience crossflow during the shutin, the resulting

pressure profiles around the well may change drastically if the productivity of well layers

vary drastically and crossflow is occurring. This feature allows IMEX to shutin a GAP

/RESOLVE controlled well using a *MODELSHUT type option, which allows crossflow

in the shutin well. GAP/RESOLVE controls whether the well is *MODELSHUT or

simply shutin. The user does not need to specify this IMEX keyword.

Flux Sector Reporting Option

*FLUX_SECTORNAMES and *FLUX_ISECTOR keywords introduces special sector

definitions which are used to calculate reservoir flow between these flux sectors. Flow

into these sectors and between these sectors can be reported to the output print file and to

the simulation results file as time series data. Flow can be further broken down into I-

direction, J-direction and K-direction flow components.

*FLUX_SECTORNAMES defines a list of flux sector names and their corresponding

flux sector numbers. *FLUX_ISECTOR assigns these flux sector numbers to cells using

standard array input options. Both keywords can only appear once in the data set. Please

also see the *WPRN *SECTOR, *WSRF *SECTOR, *OUTPRN *FLUX_SECTOR, and

*OUTSRF *FLUX_SECTOR keywords for output options.

Entry of Permeability and Non Darcy Correction Factor in Recurrent data

This option allows the input/redefinition of the permeability arrays *PERMI, *PERMJ

and *PERMK and the non Darcy correction factor *NDARCYCOR in recurrent data.

Binary Data File Name Now Specified on the FILENAMES KEYWORD

The name of the binary file, where binary data produced from BUILDER is written to, is

now specified on the FILENAMES keyword. Previous versions of the simulator either

defaulted the filename of the cmgbin file to the data set name or had the file name

specified on the keyword BINARY_DATA. The new format is:

FILENAMES *BINDATA-IN 'anyfile.cmgbin'


New Features in GEM 2010.10

SIMULATOR CHANGES

Thermal Modelling in GEM

Beginning with version 2010.10, GEM supports use of a thermal option wherein the

temperature equation is solved along with the other equations.

The model includes an energy balance with convection, conduction and heat loss terms.

The convection terms include phase enthalpies. Aqueous phase enthalpy is determined

from a steam table look-up while oil and gas phase enthalpies are determined from the

equation of state. This allows modelling of non-ideal fluid effects and heating or cooling

due to fluid expansion or compression, including the Joule-Thomson effect. Conduction

is modelled with a term for thermal conductivity of the rock and fluids, and heat losses to

overburden, underburden and edge boundaries are calculated with an analytical

formulation.

Temperature of injected fluids may be specified with the tubing data format

(*IWELLBORE) or simply by specifying the keyword INJ-TEMP. See the Recurrent

Data section for details. The thermal model and thermal fluid properties are specified in

the Component Properties section (see keywords *THERMAL, ENTHCOEFF and

ENTHALPY-MIXING), and thermal rock properties are entered in the Reservoir

Description section (see keywords CP-ROCK, THCONR0 and HEAT-LOSS).

Information on setting up a thermal GEM run may be found in the Tutorial section of this

Users Guide. Also, the tpl\thr template folder includes a number of example data sets.

Details of the equations used in the thermal model are given in Appendix F.

Independent Geomechanics Grid

It is now possible to specify a geomechanics grid that is independent of the host

simulators grid.


The geomechanics grid can be specified with much larger grid blocks than the fluid flow

simulation grid, and can encompass as much of the surrounding rock as desired. This

allows for much faster computation compared to the case where the geomechanics

module uses the same grid as the fluid flow computation.

There are also options for translating and rotating the geomechanics grid to control the

position relative to the host grid. Deformation controls may be applied to force the host

grid to have no deformation on specified boundaries.

See the *GEOGRID keyword description in the Geomechanics section for more

information.

Improved Parallel Performance

Improvements to the implementation of parallelization in GEM have allowed the

Simulator to run more efficiently on larger number of threads on Shared Memory

Processors.

Array initialization, well term calculation in Jacobian building, large integer arithmetic

and memory allocation have all been enhanced to allow faster parallel runs of large

problems.

Full Well Solve

A method has been implemented to minimize solver-pointer re-generation that treats

shutin or wells on BHP constraint as solver wells. The work-saving is significant for

models with a large number of well layers. See the *INACTIVE-WELL-SOLVE

keyword in the Numerical Methods Control section for more details. The method is

turned ON by default.

Entry of Permeability and Non Darcy Correction Factor in Recurrent data

This option allows the input/redefinition of the permeability arrays *PERMI, *PERMJ

and *PERMK and the non Darcy correction factor *NDARCYCOR in recurrent data.

This feature can be used for modelling hydraulic fracturing events in the simulator.

GEM-Surface network models.

In previous versions of GEM, the IPR table passed to the surface network program was

generated with equally spaced intervals. Beginning with GEM 2009.13 a logarithmic

spacing is used. This is expected to be helpful in reducing oscillations for high-

productivity and backflowing wells.


Resistance factor calculations.

Resistance to flow due to solid deposition has now been generalized so that power-law or

Kozeny-Carman relations with adjustable exponents can be used for permeability

reduction calculations due to asphaltene or mineral deposition. See the *PERM-VS-POR

keyword in the Component Properties section for more details.

WELL MANAGEMENT CHANGES

Completion Lumping Option

*LAYERCLUMP provides a way for the user to define a control lump - a set of layers

(perforations, completions) which possesses a name and a set of computed rate,

cumulative, and stream ratio values. To each control lump there is also an associated

setting value.

*LAYERRLUMP provides a way for the user to lump well layers in a group called a

report lump. For report lumps, stream rates, cumulatives and production rate ratios are

calculated and the user can design triggers based on these quantities. A given layer can be

assigned a partial membership to report lump and the same layer can belong to more than

one group.

*RM-LAYERCLUMP removes membership of a layer from the named control lump.

*RM-LAYERRLUMP removes membership of a layer from the named report lump.

*CLUMPSETTING assigns a control setting or flow restriction parameter to the named

control lumps. The flow restriction parameter is used as a multiplication factor in

adjusting the value of the well index for all the layers belonging to a given control lump.

There is also a set of computed rate, cumulative, and stream ratio values for each control

lump. Triggers may be defined using these quantities specific to named control lumps;

actions can be any valid well data lines, including lines which alter the control lump

setting values. The syntax of the control lump based trigger statement is:

*TRIGGER trigger_name *ON_CTRLLUMP control_lump_name quantity > value

The syntax of the report lump based trigger statement is:

*TRIGGER trigger_name *ON_RPTLUMP report_lump_name quantity > value

See *LAYERCLUMP keyword for a list of quantities which can be used in control and

report lump triggers.

*WHYSTAB *AUTOWELL Monitor

*MONITOR WHYSTAB *AUTOWELL frequency can be used to monitor production

well wellbore hydraulics stability using the *AUTOWELL option, which shuts the well

for a specified time.


*MXCNRPT Maximum Number of Continue Repeats

Every well having *CONT *REPEAT (continue-repeat) specified through *OPERATE is

entitled to one repeat of the current time step if one of its operating constraints is

violated. During convergence of a time step, it is possible that different wells violate and

then switch their operating constraints consecutively after each repeat. Although the

maximum number of possible repeat cannot exceed the number of wells, it may lead to a

significant increase of run time if there are many wells. Quantity nmxrpt limits the

number of such repeats per time step, and keyword *MXCNRPT lets the user override its

default value (of 1).


New Features in STARS 2010.10

Important Changes between STARS 2010.10 and 2009.10

Flexible Wellbore Model

The Flexible Wellbore model allows you to simulate an advanced well geometry or

stream configuration with a method that is more accurate than a standard source-sink well

and more flexible and robust than a discretized wellbore. This release is the first

implementation stage of this feature, so some restrictions apply. This features keyword

manual is in file FlexWell_Keywords.doc in the release area under directory doc.

Independent Geomechanics Grid

You may specify a geomechanics grid that is independent of the host fluid-flow grid, for

example, to account for the influence of country rock on reservoir deformation.

Communication between the two grids is accomplished using a mapping algorithm. This

technique removes most restrictions on the host grid (e.g., local refined grids). New

keyword *GOUTSRF lets you plot geomechanics results on the independent

geomechanics grid.

SDDK Model

A variation of the subdomain natural fracture model, known as SDDK, allows you to

connect vertically the stacks of subdomain matrix blocks to account for capillary

continuity. You may also account for reinfiltration of fluids into the matrix at the

horizontal fracture as well as partial contact for heat conduction.


Water-Gas and Third-Phase Capillary Pressures

You may now specify water-gas capillary pressures in rock-fluid tables for situations

where the middle (oil) phase is absent, at initial conditions and during a run. This is more

accurate than using only water-oil and oil-gas capillary pressures for the water-gas case.

You may also specify explicitly a third-phase correction for capillary pressure.

Data Incompatibilities with Previous Versions of STARS

The following mandatory data changes must be done to an existing

STARS data set in order for it to work correctly with version 2010.

1. A case that specifies heat loss (*HLOSSPROP) in a grid direction blocked by null

cells will give a result different from v2008. In most cases this blockage was

undetected and unexpected, so the new result (heat loss through null cells) is

preferred. If the previous result is desired, disable heat loss through those null cells

via *HLOSSPROP, *ROCKTYPE and *THTYPE. This was in the 2009 release but

not the 2009 manual.

2. A 3D radial grid case using *GEOMECH with no prescribed condition on the inner

radial boundary will give a result different from v2008. Previously the inner

boundary was unconstrained radially, possibly giving unphysical results (e.g., well

hole shrinkage from injection). Now the default constraint on the inner radial

boundary is the adjacent fluids pressure. This was in the 2009 release but not the

2009 manual.

3. Option *SR2PREC *SINGLE (write reals to SR2 binary files in 4-byte mode) is no

longer supported and the alternative *SR2PREC *DOUBLE (8-byte mode) is used

exclusively. A restart written with *SINGLE by a version before 2009.10 can be

read by versions 2009 and later, but the child run will use *DOUBLE. This was in

the 2009 release but not the 2009 manual.

4. When grid array data is specified, qualifier *RG is not allowed together with data

option *EQUALSI or *MOD.

5. The optional pathname for *BINARY_DATA should no longer follow that keyword.

Use *FILENAME *BINDATA-IN instead.

6. Command-line argument -nowait is no longer supported. See command-line

argument -wait.

7. Option *DYNSR2MODE *DYNAMIC is no longer supported. The action

corresponding to *DYNSR2MODE *STATIC is now mandatory.

8. Keyword *FLUIDHEAT is no longer supported. Use *GEOGRID instead to model

geomechanics in a region larger than the fluid/heat reservoir.

9. The documented default of *PINCHOUT-TOL is now honoured. Data sets without

*PINCHOUT-TOL but with cell thicknesses less than the documented default (e.g.,


micro models) may have run in previous versions. Now, such a data set requires

*PINCHOUT-TOL to override the default (e.g., with value 0).

New Keywords Added to STARS 2010.10 1. New command-line argument -wait allows a submitted job to wait for up to 72

hours for a license to become available.

2. New *FILENAME subkeyword *BINDATA-IN allows you to override the default

filename or pathname of the file used by *BINARY_DATA.

3. New special history *STRESSDIFF (stress differential for a UBA) is available with

*GEOMECH.

4. New special history *PRODSTEAMR allows you to plot well steam production as

cold-water equivalent liquid. See template STTST07.

5. New keyword *SCONNECT allows you to specify a connection between two cells

that are not normally considered neighbours. Some restrictions apply.

6. New *SUBDOMAIN subkeyword *SGMETHOD allows you to specify the method

for calculating matrix-fracture interblock geometry parameters.

7. New keyword *TRANSD allows you to connect vertically the stacks of matrix cells

of a subdomain model, commonly called an SDDK model. See template STFRR006.

8. New keyword *SD_HCMULT allows you to modify conductive heat flow across the

connection created by *TRANSD. See template STFRR006.

9. New keyword *GASD-MIN allows you to override the default value of minimum

allowed gas compressibility factor.

10. New *VISCTABLE subkeyword *ATPRES allows you to specify viscosity that

varies with pressure as well as temperature. See template STFLU038.

11. Keyword *KRINTRP now accepts wettability subkeywords *WATWET, *OILWET,

etc., so that different interpolation sets can have different wettability.

12. New *SWT subkeyword *PCGW allows you to specify water-gas capillary pressure

directly. New keyword *PCGWEND allows you to modify and scale the water-gas

capillary pressures. See template STSMO036.

13. New keyword *PC_3RD_PHASE allows you to specify a third-phase correction for

capillary pressure.

14. New keyword *SD_REINF allows you to specify reinfiltration between matrix and

fracture at a connection created by *TRANSD. See template STFRR006.

15. New keyword *GEOGRID allows you to specify a geomechanics grid that is

different from the host fluid-flow grid. This method can account for the influence of

country rock on reservoir deformation. Section Host Grid versus Independent Grid

is added to the introduction of the Geomechanics chapter. Section Modelling


Geomechanics Outside Host Fluid Reservoir now refers to *GEOGRID instead of

*FLUIDHEAT. See templates STGEO050-55.

16. New keyword *GOUTSRF allows you to use Results 3D to plot geomechanics

quantities for the grid specified by *GEOGRID. This information is dumped to an

additional set of SR2 files.

17. New keyword *GRTEMTAB allows you to specify geomechanical properties that

vary with temperature. See template STGEO049.

18. New subkeyword *SOR is available with *GCONI *TARGET.

19. New keyword *MXCNRPT lets you control the number of allowed timestep repeats

caused by well constraint changes.

20. Option *MONITOR *AUTOWELL is available for constraint *WHYSTAB. See

template STWWM051.

21. Subkeywords *TEMP and *O2CONC are available as well quantities with

*TRIGGER.

22. The new Flexible Wellbore feature is accessed via keywords *PERF_FLX,

*FLX_WELLBORE, *ACTIVATE and *DE-ACTIVATE; output is controlled by

*WPRN *FLEXWELL and *OUTSRF *FLEXLAYER. These keywords are

described in file FlexWell_Keywords.doc in the release area under directory doc.

These keywords are not found in the STARS 2010 manual.

Enhancements to Existing Keywords 1. Geomechanics option *UNLOADSTR is now available with *GEOM3D. See

template STGEO047. This was in the 2009 release but not the 2009 manual.

2. The effect of keyword *RANGECHECK *OFF is cancelled at the end of each data

section and the end of each segment of recurrent data. This limits the effect of

*RANGECHECK *OFF whose corresponding *ON is forgotten or damaged. It is

recommended that *RANGECHECK *OFF be used as little as possible.

3. Relative pathnames entered via *FILENAME and *INCLUDE are relative to the

directory containing the main data file. Previously such a relative pathname was

relative to the current working directory, which may not be well defined when

running across a network.

4. The effective limit of number of cells available with *GRID *CORNER has risen

from about 22 million to about 178 million. This limit is due to the 4-byte indices

used in the SR2 files.

5. Improve inheritance of keywords *FRFRAC and *FORMINFRAC. See template

STGRO051.

6. The default for keyword *PRPOR was improved.

7. Keyword *INTCOMP is no longer limited to *RPT wettability option *WATWET.


8. Restrictions on keywords *PCWEND and *PCGEND are enforced for consistency.

9. Add section *DEPTH_AVE for Gas/Water Reservoir to description of

*VERTICAL keyword

10. Improve *TRANZONE description with reference to new *SWT subkeyword

*PCGW.

11. Improve *UPSTREAM description.

12. Improve table in *PPATTERN description.

13. The *PPATTERN algorithms ensure that all the members of a discretized wellbore

fall within the same Parasol class, thereby enhancing stability of multi-threaded runs.

14. Add section Geomechanics Porosities to the introduction of chapter Geomechanics,

to describe the various porosities available in Results.

15. Add section Closed Perforations to *HEAD-METHOD description.

16. Improve DEFAULT and EXPLANATION for *QUAL.

17. The documented default of *PINCHOUT-TOL is now honoured. Data sets without

*PINCHOUT-TOL but with cell thicknesses less than the documented default (e.g.,

micro models) may have run in previous versions. Now, such a data set requires

*PINCHOUT-TOL to override the default (e.g., with value 0).

New and Changed Template Data Sets These files can be found in CMG release area /cmg/stars/2010.vv/tpl

where vv is the particular version number.

Fluid Types (directory /flu)

stflu038.dat Test/Illustrate *VISCTABLE *ATPRES with *VISCTYPE, *WATPHASE,

*OILPHASE

Fractured Reservoirs (directory /frr)

stfrr006.dat Illustrate/Verify SUBDOMAIN-DK Option: *TRANSD, *SD_REINF,

*SD_HCMULT

Geomechanics (directory /geo)

stgeo047.dat Unload Stress for 3D Cylindrical Grid with *UNLOADSTR

stgeo048.dat Generalized Plastic Model *GENPLAST, plus *DLOADBC3D

stgeo049.dat Test/Illustrate *GRTEMTAB - T-Dependent Geomaterial Properties

stgeo050.dat Test/Illustrate *GEOGRID *GCORNER with *GRID *CORNER

stgeo051.dat Test/Illustrate *GEOGRID *CART with *GRID *RADIAL

stgeo052.dat Test/Illustrate *GEOGRID *RADIAL with *GRID *CART


stgeo053.dat Dean Problem #4, Water Flood - Single *CART Grid

stgeo054.dat Dean Problem #4, Water Flood - Dual *CART Grids

stgeo055.dat Dean Problem #1, Reservoir Embedded in Geomechanics Region

Grid Options (directory /gro)

stgro051.dat Test/Illustrate Partial Inheritance for *FRFRAC and *FORMINFRAC

Simulator Options (directory /smo)

stsmo036.dat Change: add *SWT subkeyword *PCGW

stsmo056.dat Change: *LINEAR to *EXP

Wells and Well Management (directory /wwm)

stwwm051.dat Illustrate/Verify *MONITOR *WHP *AUTOWELL

stwwm059.dat FlexWell with T/A, SAGD Production- Annulus Only

stwwm060.dat FlexWell SAGD Production in Tubing & Annulus

stwwm061.dat FlexWell with T/A, Variable Tubulars, SAGD Production in T/A

stwwm062.dat FlexWell with T/A, SAGD Production, SAM - Annulus Only

stwwm063.dat Replace S/S *GRAV-FRIC-HLOS Well with FlexWell - Cyclic steam

injection

stwwm064.dat FlexWell, Multi-Phase Co-Injection, SAGD Annulus Production Only

stwwm065.dat FlexWell, Multi-Phase Co-Inj, SAGD T/A Production, Operating Constraint

stwwm066.dat FlexWell, Multi-Phase Co-Injection, SAGD T/A Production


New Features in CMOST 2010.10

Important Changes between 2009.12 and 2010.10

NPV Present Date

By default, CMOST chooses the simulation start date of the base SR2 file as the Present

Date in net present value calculation. In 2009.12 and prior, it is not possible to change the

Present Date. Starting from 2010.10, the user can choose any date as the Present Date.

The only requirement is that all cash flow must occur on or after the Present Date. For

example, if the base SR2 file consists of 10 years of history matching phase and 5 years

of prediction phase, the user may want to choose the start date of the prediction phase as

the Present Date for optimizing the prediction phase.

Objective Functions for Different Task Types

In 2009.12 and prior, History Matching Error is only available for History Matching

tasks. Starting from 2010.10, History Matching Error is also available for Optimization

and Sensitivity Analysis tasks. This feature is useful for DynaGrid optimization because

the optimization often needs to consider two objectives: reduce run time and maintain

accuracy of the simulation results. See Types of Objective Functions for the list of

objective functions supported in different types of task.

History Matching Well Log Data

A new objective term: Property vs. Distance Objective Term is added for History

Matching Error objective functions. In the Property vs. Distance objective term

calculation, the property vs. distance data retrieved from the SR2 file is compared to the

data available in the well log file. Similar to history matching error terms, the relative

error between the simulated data (SR2 file) and historical data (well log file) is

calculated. Therefore, this feature allows history matching well/production log data in

CMOST. See Property vs. Distance Objective Term for details.


Fluid Contact Depth Property

Based on the fluid saturation data along a well path, fluid contact depth calculations can

be performed in CMOST. If the saturation data are available at many time steps, the same

contact depth can be calculated at many time steps to obtain a fluid contact depth vs. time

series. The fluid contact depth vs. time series can then be used to define History

Matching Error Objective Terms and Raw Simulation Result Objective Terms. See

Create Fluid Contact Properties for details.

Latin Hypercube Design for History Matching and Optimization

In 2009.12, Latin hypercube design is only available for Sensitivity Analysis and

Uncertainty Assessment tasks. Starting from 2010.10, Latin hypercube design is also

available for History Matching and Optimization tasks. If Latin hypercube design is used

for a History Matching or Optimization task, sensitivity analysis results (Effect Estimates

and Response Surfaces) will be calculated by CMOST.

Time-Series Result Observers

Starting from 2010.10, the user can specify the start and end date times for each time-

series observer. If the number of display points specified by the user is more than the

number of data points available in the SR2 file for the given time period (from start time

to end time), all data points available in the SR2 file will be recorded and displayed by

CMOST. If the number of display points specified by the user is less than the number of

data points available in the SR2 file, CMOST will spread points evenly from the start

time to the end time.

Pre-Simulation Dataset Processing Command

This feature replaces the User Defined Application for Processing Dataset feature in

the 2009.12 version. For task files created using 2009.12, the fields will be automatically

converted using the new format. This feature allows the use of a user-specified

application to modify the datasets created by CMOST before they are submitted. The

new implementation in 2010.10 allows the user to configure the command line to launch

the user-specified application. See Pre-Simulation Dataset Processing Command for

details.

Run Builder Silently

2010.10 Builder allows the keyword RESULTS SILENTEXEC to define calculations to

be performed in a dataset. For example, the user may want to let Builder re-calculate the

relative permeability table using new endpoints and exponents of the Corey equation. The

user could also let Builder use a formula to change a property array (for instance, modify

the permeability array according to sectors). CMOST 2010.10 supports the integration of


Builder in the dataset generating process: (1) CMOST creates a dataset. (2) Builder

modifies the dataset using the calculations defined in RESULTS SILENTEXEC. (3)

CMOST submits the job. See Pre-Simulation Dataset Processing Command for how to

configure Run Builder Silently in CMOST.

Disk I/O Monitor

If CMOST is configured to run multiple concurrent simulation jobs simultaneously, one

important factor that has the biggest influence on the stability of CMOST and the

computer operating system is the total data transfer rate to the file server (including

simulation output files and the CMOST result file). Problems caused by this factorsuch

as unresponsive application, failed jobs, or intermittent disk I/O errorare often difficult

to diagnose. To provide troubleshooting information for these kinds of problems, a new

feature has been implemented to monitor the total data transfer rate during a CMOST run.

See Disk File I/O Monitor for details.

Job Record and File Management

The More Engine Options dialog can be used to configure how CMOST manages job

records in Launcher and simulation output files on the disk. For example, the user can let

CMOST keep or delete simulation output files (.irf, .mrf, .out) for abnormal termination

jobs.

Stop or Pause Engine during Restart

In 2009.12 and prior, it is not possible to stop or pause the engine during the restart

process. Improvements have been made in 2010.10 so that the user is able to stop or

pause the engine during the restart process.

Tornado Plot Improvement

Several improvements have been made in Tornado plots: (1) The Maximum and

Minimum objective function values of all experimental design jobs will be displayed in

the plot to indicate the total variation of the objective function. When Maximum and

Minimum are displayed together with Target (field history value), it visually shows

whether it is possible to get a match. (2) Maximum, Minimum, and Target will always be

displayed at the top of the plot for easy comparison. (3) In addition to Tornado plots for

absolute effects, Tornado plots for t ratios are included. (4) Effect estimates for four types

of polynomial models (linear, quadratic, reduced linear, and reduced quadratic) are

calculated. In reduced linear or reduced quadratic model, statistically insignificant terms

are removed automatically.


Response Surface Improvement

Several improvements have been made in the Sensitivity Analysis and Uncertainty

Assessment tasks using a classical experimental design or a Latin hypercube design: (1)

Response surfaces are available in both Sensitivity Analysis and Uncertainty Assessment

tasks. (2) Depending on the settings of parameters and the sampling method selected,

four types of polynomial models (linear, quadratic, reduced linear, and reduced

quadratic) can be created. (3) Many new response surface statistics are calculated. For

example, R-Square Adjusted, R-Square Prediction, Analysis of Variance, t ratio, variance

inflation factor, etc. See Appendix G: Post Processing for details.

CMOST Host Service2

Starting from 2010.10, CMOST Host Service name is changed to CMOST Host

Service2. The new port number is 9088. There are two important new features in the

new service. First, each CMOST task file is run by a cmost.engine.exe process. If there

are two tasks running concurrently, there will be two cmost.engine.exe processes shown

in Windows Task Manager. There will be no interference between these two processes.

Second, when the new service launches the cmost.engine.exe process, it always finds the

same version of cmost.engine.exe as CMOST Studio. This guarantees the version

consistency between CMOST Engine and CMOST Studio.

Important Changes between 2009.10 and 2009.12

Latin Hypercube Design

The implementation of Latin Hypercube design allows Sensitivity Analysis and

Uncertainty Assessment for any number of parameters and any number of samples

values. See Latin Hypercube Design and Options for Latin Hypercube Sampling for

details.

Parameter Interaction Effects and Quadratic Effects

In addition to main (linear) effects iax , both interaction effects ji xbx and quadratic

effects 2

icx can be determined if three or more sample values are entered for a parameter.

See Quadratic Model Effect Estimates for details.

Tornado Plots in Uncertainty Assessment

Tornado plots of effect estimates are available for both Sensitivity Analysis and

Uncertainty Assessment.


One-Parameter-at-a-Time Method for Sensitivity Analysis

The traditional one-parameter-at-a-time method for sensitivity analysis has been

included. See One-Parameter-at-a-Time Sampling and Localized Effect Estimates for

details.

Discrete Probability Distribution

Discrete probability distribution has been added. See Discrete Probability Distribution

for details.

Truncated Normal and Log Normal Distribution

Normal and log normal distribution can be truncated by a range defined by the user. See

Truncated ormal Distribution and Truncated Log ormal Distribution for details.

Suggest Sample Values Based on Prior Probability Distribution

For Uncertainty Assessment using response surface and Monte Carlo simulation, it is

important to ensure that the sample values used to build the response surface are

consistent with the prior probability distribution used in Monte Carlo simulation. This

new feature provides a convenient way to select sample values based on the prior

probability distribution. See Suggest Sample Values Based on Prior Probability

Distribution for details.

Simulation Auto Restart

This feature is useful if there are CMOST parameters in the Well and Recurrent data

section and restart records are written by using WRST TNEXT. With this feature, when

CMOST creates a new job it will check whether the new job can restart from an old job

that has already finished and still has SR2 files. If the new job can restart from an old job,

CMOST will automatically write restart keywords in the dataset so that the simulator will

make a restart run for the new job and save run time. See Simulation Auto Restart for

details.

Kill Jobs in CMOST

Running simulation jobs can be killed in CMOST directly. There are two options

regarding how CMOST will treat the job after it is killed (see Kill Simulation Jobs for

details).

Kill the job and treat it as an abnormal termination job: This option should be

chosen if its suspected that the job is experiencing convergence problems and you dont

want CMOST to re-run the same job pattern (combination of parameter values).


Kill the job and treat it as a killed job: This option should be chosen if you want to

kill the job and let CMOST re-create and re-schedule a new job to a different scheduler

for the same job pattern.

Option to Re-run Abnormal Termination Jobs

This option has been added to allow the user to choose whether abnormal termination

jobs (jobs that didnt run to complete due to numerical problems) need to be re-run

during a CMOST restart run. See CMOST Restart Configuration for details.

User-Configurable Criterion of Optimal Solutions

This feature allows the user to configure the criterion used by CMOST to determine

whether a job is an optimal solution. If a job meets the selected criterion, the job will be

marked as an optimal job and its related symbols/lines will be displayed using the color

for optimal jobs (the default is green) in CMOST Studio. The user can also control the

engine stop criterion by selecting User threshold and specifying objective function

threshold. See Criterion of Optimal Solutions for details.

User-Specified Executable for Processing Dataset

This feature allows the use of a third-party application to modify the datasets created by

CMOST before they are submitted. For example, the user may want to adjust variogram

parameters in history matching. In this case, an external geological modelling package

must be used to generate porosity and/or permeability arrays for each dataset created by

CMOST. After that, CMOST will submit a simulation job using the modified dataset. See

Pre-Simulation Dataset Processing Command for details.

Full Support of INCLUDE, BINARY_DATA, and FILENAMES INDEX-IN Keywords

Improvements have been made on the handling of INCLUDE, BINARY_DATA, and

FILENAMES INDEX-IN keywords in the CMOST master dataset. In addition to

handling file names without directory information, both relative paths and absolute paths

are supported. See Handling Files Referenced by Master Dataset for details.

Write Simulation Log File in the Same Directory as the Dataset

To allow users to easily monitor the progress of simulation jobs running on remote

computers, simulation log files will always be written in the same directory as the dataset

even when the feature write output files to a temporary folder on the execution computer

and copy all files back to the simulator working directory when the job is done is

enabled.


Licensing Discount

For new simulator versions, a licensing discount will be applied to simulation jobs

submitted by CMOST.

Sampling Method Section Used Instead of Experimental Design

As part of the updates for Latin Hypercube, the Experimental Design section that used

to exist under the CMOST Task tree view has been replaced by the Sampling Method

option. The Sampling Method section still contains the design job patterns that existed

under Experimental Design. However, some updates to the Sampling Method section

include the ability to set different Latin Hypercube configurations, and the ability to add

verification runs for Uncertainty Assessment tasks. See Sampling Method for more

details.

Settings for Optimization Method are in Separate Section

In CMOST 2009.10 and prior, the settings for the optimization method were outlined in

the Run Configurations section of a task file. They are now located under the

Optimization Method section. See Optimization Method for details.


New Features in Builder and Results 2010.10

New Features in Builder 2010 1. An interface has been added to STARS datasets for supporting the new

FLX_WELLBORE keywords. Please see the FlexWells (STARS) section for more

information.

2. An interface has been added to IMEX datasets for supporting the new Flux Sector

keywords. See the Flux Sector (IMEX Only) section for more information.

3. The interface for Saving Builder datasets has been updated to improve the

customization of how datasets are saved. The user can now save most keywords in

their own include file / binary file, or group together multiple adjacent keywords (as

opposed to being limited to saving entire sections of the dataset). The updated

version of the Save As dialog will provide better tailoring for CMOST template files.

See Saving Your Work to a Simulator Input File (Dataset).

4. The liquid phase viscosity tables in STARS datasets have been updated to allow the

user to create multiple pressure based viscosity tables. See Entering Liquid Phase

Viscosity Data.

5. The Hydraulically Fractured Wells dialog has been updated to support the creation of

Stimulated Reservoir Volume (SRV) stages. These stages can be created with the

mouse or by incorporating microseismic data into Builder. See the Hydraulically

Fractured Wells (IMEX & GEM) and Microseismic Events sections for more

information.

6. Builder now supports microseismic data. Using the Microseismic Data Manager,

you can import and export files, change how microseismic events are displayed, as

well as animate events in both 2D and 3D. The data manager also enables you to use

microseismic events as a guide to creating a new Cartesian grid. Microseismic data

can also be used in modeling a Stimulated Reservoir Volume (SRV) stage via the


Hydraulically Fractured Wells dialog. See the Hydraulically Fractured Wells

(IMEX & GEM) and Microseismic Events sections for more information.

7. In Builder, it is possible to create or modify Parameters in CMOST Master Datasets

(CMM files). The ability to do this for formula related parameters and relative

permeability end point parameters has been added. See the Parameterizing Relative

Permeability Correlations and Parameterizing Formulas sections for more

information.

8. Builder now supports the ability to have multiple Numerical sets for STARS datasets.

These numerical sets are time dependent, so different dates can use different

numerical settings. See the Setting and Editing umerical Controls (STARS)

section for more information.

9. The Compaction/Dilation section for GEM datasets has been updated to support

some new thermal based keywords. See the Setting Thermal Properties for the

Rock Type (GEM only) section for more information.

10. In Builder and Results 3D, new functionality has been added in the formulas dialog

which allows spatial properties to be calculated based on the values in neighboring

blocks. The list of functions available is discussed in the List of Operators section.

New Features in Results Graph 2010 1. Multi-selection support has been added to Results Graph. You can multi-select

(using the or key) when adding and deleting curves, deleting plots,

and opening files. When opening multiple files by dragging and dropping them from

Launcher, a separate session file will open for each of the results files selected. If

you want multiple files to opening in the same session file, this can be done by using

multi-selection and the following options under the File menu:

Open CMG Simulation Results

Open Field History

Open PA Load Format

Open CHEARS Simulation Results

2. The Property vs. Distance dialog has been updated to allow users to use a Flex Well

for the data path. This feature is only available for STARS based simulation results.

3. If sectors exist in the results file you have open in Graph, a Property vs. Distance plot

can be created which uses the values in a particular sector for the data path. Based on

grid depth, the average property value in the sector chosen will be plotted.

4. Curves which plot the fluid contact depth in a reservoir over time can be added.

Refer to the section Creating a "Fluid Contact Depth" Parameter for more

information.

5. In Results 3D and Results Graph, the user will have the option of creating a *.tsdata

file when opening simulation results (*.irf) files. The purpose of the *.tsdata file is to


re-index the *.irf file and save the information in the *.tsdata file, so that any further

actions which reference this*.irf will be quicker (e.g. re-opening the file, creating

plots, etc.). Refer to the sections that reference the Preferences for more information,

i.e. Results 3D & Builder User Preferences and Graph User Preferences.

New Features in Results 3D 2010 1. The option to update the results while the simulation is running has been added to

Results 3D. It is possible to update the results on a periodic basis, or on demand.

Refer to the section Periodically Re-read Simulator Output for more information.

2. Results 3D now supports multi-selection when opening files.

3. In Builder and Results 3D, new functionality has been added in the formulas dialog

which allows spatial properties to be calculated based on the values in neighboring

blocks. The list of functions available is discussed in the List of Operators section.

4. In Results 3D and Results Graph, the user will have the option of creating a *.tsdata

file when opening simulation results (*.irf) files. The *.tsdata file caches time series

data in a form that makes subsequent accesses much faster. Refer to the section

Results 3D & Builder User Preferences for more information.

5. The Property vs. Distance dialog has been updated to allow users to use a Flex Well

for the data path. This feature is only available for STARS-based simulation results.

6. If sectors exist in the results file you have open, a Property vs. Distance plot can be

created which uses the values in a particular sector for the data path. Based on grid

depth, the average property value in the sector chosen will be plotted.

New Features in DataImporter 2010 A number of defects have been corrected including the following:

1. DataImporter converts PORV (absolute) to VOLMOD (relative)

2. DataImporter places SWINIT keyword before VERTICAL keyword

3. DataImporter converts SWCR keyword as SWCON when it should be SWCRIT

4. DataImporter Converts PERMZ Array Incorrectly if No Whitespace Before Trailing

Slash

5. DataImporter converts RSCONSTT PVT Incorrectly

6. Rock Compressibility keywords created for rocktype (CCPOR, CPRPOR) instead of

for whole reservoir (CPOR, PRPOR)

7. MODEL GASWATER not converted properly

8. PVT Section not produced for dataset


New Features in Launcher 2010.10

Whats New in Launcher 2010 1. All CMG documentation (including sidecar simulator manual

pages) may now be accessed from the CMG Documentation

Browser. This may be accessed through the Help menu in Launcher.

2. Support for the PBS scheduler has been added.

3. Job-specific environment variables may be added per scheduler.

4. The remote job bootstrapper (cmgsimrun) now uses a response file

for command arguments.


New Features in WinProp 2010.10

New Approach for the Multiple Contact Miscibility Calculation

A Tie Line calculation method has been added to WinProp in the Multiple

Contact Miscibility Calculation section to calculate the minimum

miscibility pressure (MMP) or minimum miscibility enrichment (MME).

This method takes the combined condensing and vaporizing displacement

mechanisms into consideration, as well as the existing pure vaporizing or

pure condensing mechanisms. With the pressure or enrichment increasing,

all key tie lines, including the initial tie line, injection tie line and the

crossover tie lines, can be found simultaneously based on the method of

characteristics theory. The MMP or MME can be determined once any of

these key tie lines length becomes zero. Please see the Multiple Contact

Miscibility Calculation section of the Users Guide for more details. Use

of the feature is illustrated in the mcm-combined-U2002rich-MMP.dat

template data set.

Liquid Viscosity-Temperature Table for Multiple Pressures

The STARS PVT generation option can now calculate and output multiple

liquid viscosity-temperature tables over a defined pressure range. This is

compatible with a new STARS option to allow pressure dependence of

liquid viscosities available in STARS 2010.10.

Documents

New Features in CMG 2010 Software