Unground Hydrogen Storage

Simulation Using CMG

Tutorial
CMG 2021 General Release (GEM 2021)

1
Contents
Contents.................................................................................................................................................. 2
Scenario: H2 Storage in Aquifer ................................................................................................................ 4
Exercise 1: Aquifer Base Case .................................................................................................................. 5
Reservoir Section ................................................................................................................................. 6
Component Section ........................................................................................................................... 13
Rock-Fluid Section ............................................................................................................................. 18
Initialization Section .......................................................................................................................... 18
Numerical Section ............................................................................................................................. 19
Wells & Recurrent Section ................................................................................................................. 20
I/O Control section ............................................................................................................................ 27
Results Interpretation ........................................................................................................................ 29
Exercise 2: Hysteresis Trapping .............................................................................................................. 32
Results Interpretation ........................................................................................................................ 32
Exercise 3: Hydrogen Solubility in Water................................................................................................ 33
Adding solubility parameters to the WinProp model .......................................................................... 34
Results Interpretation ........................................................................................................................ 35
Exercise 4: Hydrogen Diffusivity in Water .............................................................................................. 38
Results Interpretation ........................................................................................................................ 38
Exercise 5: Extending Simulation Time for Base, Hysteresis, Solubility, and Diffusion Models................. 39
Extending Production (H2 Trapping) ................................................................................................... 39
Results Interpretation ........................................................................................................................ 40
Extending Simulation for 100 Years (Diffusion) .................................................................................. 41
Results Interpretation ........................................................................................................................ 42
Exercise 6: Multiple Storage Cycles ........................................................................................................ 43
Adding Cycling Group to the model ................................................................................................... 44
Results Interpretation ........................................................................................................................ 49
Adding Monitor Constraint to PRODUCER .......................................................................................... 50
Results Interpretation ........................................................................................................................ 51
Exercise 7: Cushion Gas Injection ........................................................................................................... 52
Results Interpretation ........................................................................................................................ 54
Exercise 8: Methanation Reaction ......................................................................................................... 57
Results Interpretation ........................................................................................................................ 61

2
Exercise 9: Cap-Rock Integrity (GEOMECHANICS) ................................................................................... 63
Results Interpretation ........................................................................................................................ 72
Exercise 10: GEM – CoFlow Coupling to Model Surface Facilities ........................................................... 74
Importing the WinProp generated fluid model into CoFlow ............................................................... 74
Importing GEM .dat file into CoFlow .................................................................................................. 76
Creating the Wells ............................................................................................................................. 78
Coupling GEM Model Wells to CoFlow Model Wells........................................................................... 90
Creating Injection and Production Facilities ....................................................................................... 92
Set-up Integrated Network Simulations ............................................................................................. 97

3
Scenario: H2 Storage in Aquifer

We have a wind turbine farm that can produce 4.5e7 m3 of hydrogen with its excess electricity during the
peak production periods in spring and summer (March to August). The nearby city requires 500,000
m3/day of hydrogen to cover the demand peak for electricity in winter (December, January and February).
In this course, we will build an infinitely acting anticline aquifer model and design multiple cycles of
hydrogen injection and production based on the hydrogen availability (in spring & summer) and demand
(in winter) to see if this aquifer can store and deliver the required hydrogen to the city during the peak
electricity consumption season.

 The base model is constructed in the first step. We will not introduce any cushion gas at the
beginning while setting up the injection / production periods. Later, we will see if we need to
inject any cushion gas into this model.
 In the second step, hysteresis trapping is introduced into the base case to see how it will affect
the production.
 In the next steps, hydrogen solubility and diffusivity in water are defined for the base case, and
their effects on the hydrogen storage are investigated.
 Later, after adding hysteresis, solubility and diffusivity to the model, we will add four more cycles
to operation to see how the process will go forward while having multiple cycles.
 After that, one period of cushion gas injection will be defined for the storage process. N2 will be
injected into the model as cushion gas.
 Then, we will change the cushion gas to CO2, and also introduce the Methanation reaction into
the model.
 Geomechanics will be introduced to the base case with some slight modifications to monitor the
shear failure of caprock in the next step.
 At last, we will build the injection and production surface facilities in CoFlow, and will couple them
to the GEM reservoir model using CoFlow-X.

Software used: Builder 2021.10, WinProp 2021.10, GEM 2021.10, Results 2021.10 (all from CMG 2020
General Release), CoFlow 2021.30

4
Exercise 1: Aquifer Base Case

1. Open CMG Launcher, and then, open Builder 2021.10.

2. In Builder, click on File  New.
3. Here we need to select the simulator (GEM), working unit (SI), porosity system (Single Porosity),
and the starting date (2021-01-01) for the simulation. Click OK after setting these items. A new
window will pop-up after clicking OK, which will only show the reservoir simulation setting. Click
OK on the window as well.

5
Now, we have a blank GEM model. We will set up the model in the following order:

 Import the reservoir grid from RESCUE file.

 Build the fluid model using WinProp and import it in Builder.
 Add the relative permeability tables to the model.
 Define the initial state of the model (initial reservoir pressure, fluid contacts, saturations, and the
composition of the fluids).
 Adjust the numerical settings for the model.
 Construct the injection and production wells and their operating periods.
 Set up the input / output section by selecting the output parameters and their writing frequencies.

Reservoir Section

4. In Builder, go to File  Import from another file  RESCUE/RESQML Model. Navigate to the
Required Data/RESCUE folder, and select Aquifer-Grid.bin. A window will pop up that shows the
description of the rescue model. Click OK on this window.
5. The next pop-up window defines how we are treating the inactive cells while importing the
reservoir grids from the rescue file into Builder. Select the following option, then, click OK.

6. The next pop-up window allows us to select the reservoir properties that we want to import from
the rescue file into Builder. Builder will automatically detect the properties if their names in the
rescue file are identical to their Builder names. Make sure the following six properties are selected
to import into Builder (NULL blocks, Pinchout Array, Porosity, Permeability I, Permeability J, and
Permeability K). Then, click OK.

6
 7. Save the model as Exercise-1_H2-storage_Aquifer_base-model.dat in the student solutions
directory.

There is a drop-down menu at the top left corner of Builder for selecting different reservoir views (IJ-2D,
IK-2D, JK-2D, and 3D). Use that to select different reservoir views and get familiar with the reservoir shape.

7
This is a homogenous anticline reservoir with 20% constant porosity, and 100 mD constant permeability
in all three dimensions (I, J, and K). The following two photos show a 3D view and a 2D-IK view of the
model.

8. The Reservoir section in the Model Tree View shows a warning regarding rock compressibility. We
need to set this property to remove the warning, otherwise, the CPOR will default to zero. Go to
Reservoir  Rock Compressibility. Set CPOR to 5.8e-7 1/kPa and PRPOR to 6400 kPa in the Rock
Compressibility window.
Note: Rock compressibility will be constant in our model, and the reservoir rock does not expand
by temperature. However, the user is able to define pressure dependent rock compressibility,
thermal expansion for the rock, and also pressure-temperature cross-term coefficient of the

8
 formation effective porosity. The Rock Compressibility and Thermal Expansion Coefficient pages
in the GEM User Guide have detailed information about these options.

9. We want to artificially enlarge the grid blocks at the bottom of our model to get an infinitely acting
behavior in it. For this purpose, we will multiple the volume of the grid blocks below the depth of
900 m by 10000. This will be done using the Volume Modifiers option (VOLMOD keyword). Firstly,
we need to write a formula for defining VOLMOD based on depth. Then, we can assign that
formula to Volume Modifiers. Go to Tools  Formula Manager.

9
10. In the Formula Manager window:
a. Click on New to create a new formula.
b. Change the Formula Name to Aquifer VOLMOD.
c. Click on Add Independent Variables, and select Grid Top at 2021-01-01, then click OK on
that window to go back to the Formula Manager main window.
d. Write following formula in space that is in the middle of the middle of window:
IF ( X0 > 900 ) THEN ( 10000 ) ELSE ( 1 )
e. Click Apply on the Formula Manager window. Then, Click OK on the message that pops
up. Then, click OK on the Formula Manager window to close it.

11. Click on Specify Property to open the General Property Specification window.
12. Find the Volume Modifiers property from the drop-down menu.
13. Right click on the Whole Grid cell for Volume Modifiers, then select Formula.
14. In the Property Specification window that pops up, click on Formula, then select Aquifer VOLMOD
and click OK.
15. Click OK on the Property Specifications window. Then, click OK on the General Property
Specification window.

10
16. Make sure the Volume Modifiers property is selected on the Block/Corner Value Calculation
window. Then, click OK on this window.

Note: Steps 17 to 20 are only for the users that have Builder 2021.10, since the Block Selector wizard
is introduced to Builder from this version. If you do not have this version, you can simply skip these
steps. They won’t affect your simulation.

17. We want to make a sector for the part of aquifer that has VOLMOD = 1 to monitor its saturation
changes throughout the storage process (These are the grid blocks with grid top smaller than 900
m).
Go to Tools  Blocks Selector, and select the Property filter option. Choose the Volume
Modifiers from the drop-down menu. Then, select = and write 1 in front of it. After that, click on
the + button in front of 1. Then, click on the => button to add the blocks with this specification to
the list.

11
18. Click OK on the Blocks Selector window. The Create/Assign Block Group window will pop up. Select
New Sector on this window, and rename the Block Group Name to Aquifer-top. Then, click OK.

19. The Add/Edit/Delete a Sector window will open automatically. Name the sector as Aquifer-top,
and click OK.

20. Click on Add the blocks in selected regions to the sector button. Then, click Apply and OK.

12
Now, the reservoir grid, porosity, permeability, compressibility and volume modifiers are set-up. We will
continue constructing the remaining parts of the model.

Component Section

21. Save the model in Builder, and go to Launcher.

In this course WinProp is used to create a fluid model with four components: H2, N2, CO2, and CH4. H2 is
the target component for storage and production. N2 will be used as cushion gas in one of the exercises,
and CO2 will be injected in another exercise again as cushion while introducing the Methanation reaction
into the model. CH4 is the product of the Methanation reaction. The WinProp correlations are used to
obtain the aqueous phase solubility parameters for these components. Please note that we will not be
matching the WinProp PVT model with any experimental measurements in this course.

22. Open WinProp 2021.10 from Launcher.

23. In the Comment Line, write: H2 Storage in Aquifer – Basic PVT. In the Title Line 1, write: H2 – N2
- CO2 – CH4 Then, select PR (1978) for Equation of state, kPa & deg C for Unit, and mole for Feed.

24. Go to Component Selection/Properties, and set the Default Reservoir Temperature to 50 degree
Celsius.
Click on Ins Lib to open the WinProp component library. Then, select CO2, N2, CH4 and H2 from
library. Click OK to close the library.

13
25. In the Composition section, set the following values for the primary composition of each
component, and leave the secondary composition zero.

26. Save the WinProp model as WinProp_H2 Storage in Aquifer – Basic PVT.dat in the student
solutions directory.
27. Right click on Composition, and Add After  Simulator PVT  CMG GEM EOS Model.

14
28. In the CMG GEM EOS Model, select the following option for writing the data, and set the reservoir
temperature to 50 degree Celsius. Then, save and run the model.

29. WinProp will generate WinProp_H2 Storage in Aquifer – Basic PVT.gem file in the student
solutions directory after running the model. We will import the fluid model in Builder

15
30. Go back to the reservoir model in Builder. In Builder go to Component  Import WinProp-
generated Model. Navigate to the student solutions folder and open WinProp_H2 Storage in
Aquifer – Basic PVT.gem.
31. Click OK on the following message.
IMPORTANT Note: The acentric factor for H2 is negative. WinProp and GEM accept negative
acentric factors for components but Builder complains about it with a message whenever you open
a model with negative acentric factor. Builder also shows an error message in the component
section regarding the negative acentric factor. This is due to an old set-up in Builder, which will be
modified in the future releases. You can simply ignore this message and error.
Builder also shows a red sign for the interaction coefficients. This is a warning regarding zero
interaction coefficients between some components. You can ignore this message as well.

32. Select No to the following question. This is asking if we want to use the composition of the
WinProp model to set the fluid composition in our reservoir model. We will set the reservoir fluid
composition in the initialization section.

16
The Reservoir section will show an error after adding the fluid model to the Component section. This is
happening since we need to define the composition of our reservoir fluid.

33. Go to Component  Component Properties EOSSet: 1, then double click on Properties of

Water. Change the reference pressure for Water Compressibility to 6400 kPa and Water
Compressibility to 4.6E-7 1/kPa, then click Apply and OK.

17
Rock-Fluid Section

34. Go to Rock-Fluid  Create/Edit Rock Types. The Rock Types window will open. In this window,
click on the triangle next to Rock Type and choose New Rock Type. This will create a blank rock
type.

35. The relative permeability curves can be generated using built-in correlations (based on end-
points) or can be copied to Builder if the user has the tables. The correlations option can be found
under the Tools button. In this exercise, we will copy the existing relative permeability tables to
Builder.

Go to the Required Data folder, and open the relative permeability curves_aquifer.xls file. Copy
and paste the water-oil and liquid-gas tables to the Builder wizard. Make sure that the Gas
Saturation is selected for the Liquid-Gas Kr Table dependency before pasting the data in the
liquid-gas table. Make sure that the Include capillary pressure box is unchecked for both tables.

Initialization Section

36. Open the Initial Conditions wizard from the Initial Conditions section in the Model Tree View.

37. In the first tab, select VERTICAL DEPTH_AVE  WATER_GAS  NOTRANZONE.

18
 38. In the second tab (Init. Region Parameters) input the values as shown in the following figure
(REFPRES: 6400 kPa, REFDEPTH: 650m, DWGC: 600m), then click OK.
The highest point of the reservoir is at the 650 m depth. We are setting the Water-Gas Contact
(DWGC) to 600 m to avoid having a gas cap in the model. Consequently, setting the Gas Cap
Composition (ZGAS) to 100% N2 will not have any physical effects on the model at this stage.
However, we have to set this value since it is a requirement for the model set-up.

Numerical Section

39. In the Numerical section (in the Model Tree View), double click on Numerical Controls.

19
 40. In the Numerical window, scroll down to NORMs, and set the followings, then click Apply and OK.
Save the model.

41. Close Builder, and open the model in CEDIT 2021.10, and navigate to the Numerical Controls
section. Then, add the following keyword to this section, and save the model.

*COMBINATIVE ON

Wells & Recurrent Section

Our wind farm can produce 4.5e7 m3 of hydrogen over 6 months (March to August). This will provide us
with 244,565.22 m3/day of hydrogen to inject into the aquifer. The initial average reservoir pressure is
about 8800 kPa, and the maximum pressure we can provide at the injection facilities is 11,000 kPa. We
need to produce hydrogen at 500,000 m3/day of hydrogen in December 2021, January 2022 and February
2022 (over 90 days).

20
At this stage, we will construct the injection and production wells. Then, we will set up the operating
periods based on the hydrogen availability and the anticipated demand. Our primary plan for the
operating constraints is to inject at MAX STG of 244,565.22 m3/day with MAX BHP of 11,000 kPa (as the
second constraint), and produce with the MAX STG of 500,000 m3/day and MIN BHP of 101.325 kPa
(atmospheric pressure) without any monitor constraints. If we are able to inject and produce with these
rates over the whole injection and production periods, we will be injecting 45,000,000.4 m3 (184 day *
244,565.22 m3/day) of hydrogen into the reservoir and producing 45,000,000 m3 (90 day * 500,000
m3/day). Although operating the production well at atmospheric pressure and without monitor
constraints might not be practical we are taking this approach to see:

 how this aquifer would deliver hydrogen based on the daily need over three months of the
consumption peak in case we allow the producer BHP to drop to 101.325 kPa, and
 whether or not this approach results in considerable water production from the aquifer.

Please note that maximizing the cumulative gas production using the operating constraints or modifying
the perforated intervals is not a purpose of this exercise.

42. Open the model again in Builder. Go to Wells & Recurrent  Dates.

43. Click on Add a range of dates. Choose From: 2021-01-01; To: 2023-03-01, by Month. Click OK and
click OK again to the message that appears. Then, CLOSE the Simulation Dates wizard.

21
44. We are going to create two vertical wells (INJECTOR and PRODUCR) at the top of the aquifer.
These two wells will be perforated in the same interval. In reality, injection and production will be
done from one well for this scenario. However, we need to define separate wells for each well
type (i.e. INEJCTOR or PRODUCER) for the simulation purposes. Right click on Wells, and click on
New.

45. Name the well INJECTOR, and select INJECTOR as the well type. We are defining this well at 2021-
01-01.

46. Click on the Constraints tab and check the box for Constraint Definition.

47. Using select new (in the Constraint column of the table), select OPERATE. Then select STG surface
gas rate, MAX, 244565.22 m3/d, CONT REPEAT. Then, add another operating constraint for BHP
bottom hole pressure, MAX, 11000 kPa, CONT REPEAT.

48. Under the Injected Fluid tab, select SOLVENT as the injected fluid, then specify the mole fraction
of H2 as 1.0. Then, click OK.

22
 49. Now, the location and perforation interval for the injection well has to be defined. In the Model
Tree View, expand Wells and INJECTOR, and double click on 2021-01-01 PERF to open the Well
Completion Data (PERF) wizard.

50. Go to the Perforations tab. Add perforations for blocks 48 23 1:4. Then, click Apply and OK.

All the sections in the Model Tree View (except the component section, which was explained earlier)
should have a green check mark now. However, we still need to add another well (PRODUCER) and modify
some other parts of the model.

23
51. Since PRODUCER and INJECTOR are at the same location in our design, we will the Copy Well
option to define PRODUCER. Go to Wells & Recurrent  Copy Well.

52. Select INJECTOR and click on Next to go the next step.

53. In Step 2, select Copy all perforation dates.

54. In Step 3, select Copy Geometry  Use the geometry that is specified for the copied perforation.

55. Click next on Step 4, and select the following options on Step 5.

24
56. In Step 6, write PRODUCR in the New Well cell. Then, click on Finish.

57. Go to Wells & Recurrent  Wells  PRODUCER. In the ID & Type tab, select PRODUCER for
Type. Then, click Apply.

58. Click on the Constraints tab and check the box for Constraint Definition. Using select new (in the
Constraint column of the table), select OPERATE. Select STG surface gas rate, MAX, 500000 m3/d,
CONT REPEAT for it. Then, add another operating constraint for BHP bottom hole pressure, MIN,
101.325 kPa, CONT REPEAT. Then, click Apply and OK.

59. Change the reservoir view to IK-2D, and go to Plane 23. Here, you can see where we have placed
the injection and production wells in the aquifer.

25
The following table shows the injection / production plan that we have for one storage cycle in this aquifer:

Starting Duration Injection rate Production

Stage Name End Date
Date (month) (m3/day) rate (m3/day)

Hydrogen
1 2022-03-01 2022-09-01 6 244,565.22 -
injection

2 Shut-in 2022-09-01 2022-12-01 3 - -

3 Production 2022-12-01 2023-03-01 3 - 500,000

The injection and production wells are shut-in from 2021-01-01 to 2022-03-01, and the project starts
at 2022-03-01. Later, we may use this shut-in period to inject cushion gas into the aquifer.

60. Go to Wells & Recurrent  Wells  INJECTOR to open the Well Events wizard. Then, go to the
Options tab for INJECTOR at 2021-01-01. Activate the Status, and change its Type/Value to
SHUTIN. Then, click Apply.

26
 61. Change the date to 2022-03-01 for INJECTOR (from the dates drop-down menu), and go to
Options tab. Activate the Status option at this date and change its Type/Value to OPEN. Then,
Click Apply.

62. Change the date to 2022-09-01 for INJECTOR, and go to the Options tab. Activate the Status
option at this date and change its Type/Value to SHUTIN. Then, Click Apply.

63. Click on PRODUCER on the left side of the window. Then, go to its Options tab (at 2021-01-01).
Activate the Status option, and change its Type/Value to SHUTIN. Then, click Apply.

64. Change the date to 2022-12-01 for PRODUCER, and click on the Options tab. Activate the Status
option at this date and change its Type/Value to OPEN. Then, Click Apply and OK.

65. Go to Wells & Recurrent  Open Time-Line View. This option shows the operating periods of the
wells on a timeline, which is very useful to check the operating periods of the wells when the
model has several wells with multiple operating periods.

I/O Control section

66. Go to I/O Control  Simulation Results Output to open the Simulation Results File Writing
window. Then, do the following steps in this window.
67. Set the Output Type to SR3.
68. Leave the options in the WSRF table as they are. This will write both the Well information at each
timestep and the Grid information at each TIME or DATE (for the current model: each month) to
the sr3 file.

27
 Note: You can change the Grid Writing Frequency to Specified frequency and write any frequency
values for it if you want to get more grid information in the sr3 file. However, writing the grid
information with a timestep frequency can create a massive sr3 file (several GBs). Therefore, this
option must be chosen with extra caution.
69. In the OUTSRF table, change the Reservoir Variable to ALL variables.
70. Click on the small green plus on the left side of the OUTSRF table to open the Select New Date
and Type window. Select Well at 2021-01-01, then click OK.

71. Select Well values for all layers (LAYER ALL DOWNHOLE) for the Well Information. This will output
the fluid flow information of the perforated layers at the surface and reservoir conditions.

28
 72. Click on Select in front of Grid Information. Here, we will select the grid level parameters that we
want to print to the sr3 file. Select the following parameters from the list. Then, click OK.

Keyword Variable Description

PERM Permeability in each direction

POROS Current porosity

PRES Pressure

SG Gas Saturation

SO Oil Saturation

SW Water Saturation
TEMP Temperature

WALL Mole fraction of all components in the aqueous phase

YALL Mole fraction of all components in gas phase

ZALL Global model fraction of hydrocarbon components

73. Save the model in Builder, and run it with GEM 2021.10 through Launcher.

Results Interpretation

74. In Launcher, drag and drop the Exercise-1_H2-storage_Aquifer_base-model.sr3 file on Results

2021.10.

75. In time series, select

a. Wells  Exercise-1…  INJECTOR  Gas Rate SC. Then, click on Add To New Plot.

b. Wells  Exercise-1…  PRODUCER  Gas Rate SC. Then, click on Add Curve.

c. Wells  Exercise-1… PRODUCER  Water Rate SC. Then, click on Add Curve.

29
76. You can use the Line Color and Line Style features to modify the color and style of the curves.

77. In time series, select

a. Wells  Exercise-1…  H2-INJECTOR  Cumulative Gas SC. Then, click on Add To New
Plot.

b. Wells  Exercise-1…  PRODUCER  Cumulative Gas SC. Then, click on Add Curve.

c. Wells  Exercise-1…  PRODUCER  Cumulative Water SC. Then, click on Add Curve.

30
The above plots show that our basic aquifer model can store 4.5e7 Sm3 of hydrogen and deliver it at a
rate of 500,000 m3/day for 82 days during the peak consumption period. The city will have a hydrogen
shortage for the last 8 days of the peak consumption season. Please note that we are not injecting any
cushion gas into the aquifer in this scenario. The water also breakthroughs into the production well, and
its rate increases to 4000 m3/day in the last week of the production period, which is very high. This is
basically due to operating the production well on STG MAX 500,000 m3/day, and BHP MIN 101.325 kPa.
Now, we need to introduce the other concepts to the model and see their effects on the storage.

78. You can see different properties at the Grid level in the 2D or 3D views from the Reservoir section.
The following Ternary profile at the end of injection period (2022-09-01) shows that the injected
gas is accumulating at the top of aquifer.

79. Save the Results project as H2 storage in Aquifer.results.

31
Exercise 2: Hysteresis Trapping

The purpose of this exercise is to add the structural trapping to the base case by incorporating hysteresis.
Different hysteresis methods can be used to model relative permeability hysteresis for the non-wetting
phase (gas) in GEM. In this exercise we will use the LAND trapping function to model gas hysteresis.
Detailed information about the gas phase hysteresis can be found in these two pages in the GEM user
guide: Gas Phase Relative Permeability Hysteresis Modelling (Optional) and Gas Relative Permeability
Hysteresis Parameter (Optional).

1. Open Exercise-1_H2-storage_Aquifer_base-model.dat in BUILDER 2021.10.

2. Navigate to the end of the Rock-Fluid section, and add HYSKRG 0.3 there.

Hysteresis modelling methods can be defined through Rock-Fluid  Rock Fluid Types  Hysteresis
Modelling tab in Builder:

3. Save the dataset as Exercise-2_H2-storage_Aquifer_hysteresis.dat and run the model with GEM
2021.10.

Results Interpretation

4. Add the Exercise-2_H2-storage_Aquifer_hysteresis.sr3 file to the Results project that you saved
(H2 storage in Aquifer.results) to compare the results of this model with those from Exercise-
1_H2-storage_Aquifer_base-model.dat. This can be done in the Input  Data Sources  Add
Files.

32
The production rate plots of the hysteresis model show that by including hysteresis in the model, the
aquifer cannot deliver hydrogen at 500,000 m3/day for the last 5 weeks of the production period. The
cumulative production plots show that the water production has tripled after adding gas hysteresis to the
model.

Exercise 3: Hydrogen Solubility in Water

The purpose of this exercise is to add hydrogen solubility in water to the base model (Exercise-1_H2-
storage_Aquifer_base-model.dat), and see its effect on the storage process.

33
Adding solubility parameters to the WinProp model

1. Open WinProp_H2 Storage in Aquifer – Basic PVT.dat in WinProp.

2. WinProp can be used to calculate the solubility parameters (Henry’s Coefficients).

3. Open the Titles/EOS/Units form and change the title to H2 Storage in Aquifer – Solubility.

4. Go to Component Selection/Properties  Aqueous Phase tab, make sure the Li-Nghiem's

Method (1986) is selected for Henry's Law Constant Correlation.

5. Go to the CMG GEM EOS Model form. Select the option to Print the aqueous phase component
solubility parameters, and select the option to Recalculate the parameter at P & T. Then, enter
the reservoir temperature (50 degree Celsius) for T and initial average reservoir pressure of 8800
kPa for P.
Uncheck the option to Write component heating values for GEM if it is activated. We do not need
it in this exercise.
6. Save the file as WinProp_H2 Storage in Aquifer – Solubility.dat and Run it. This will generate a
fluid model file named WinProp_H2 Storage in Aquifer – Solubility.gem. This file will include
solubility parameters.
7. Exit WinProp.

8. Compare the WinProp_H2 Storage in Aquifer – Basic PVT.gem and WinProp_H2 Storage in Aquifer
– Solubility.gem to see the solubility parameters were calculated for the four components. The
following screenshot shows the Henry constant values that the Li-Nghiem’s method has
calculated for the four components in our model.

Note: Checking literature Hydrogen solubility coefficients calculated can be replaced with slightly more
accurate values. We will change HENRYC H2 to 9.0417e6 and its VINFINITY to 0.037. Also, to keep the initial

34
condition of our model identical to the base model, we will change the HENRCYC value of N2 to zero. This
will make nitrogen insoluble in water, which also is not far from reality due to the small solubility of
nitrogen in water.

9. Open the Exercise-1_H2-storage_Aquifer_base-model.dat model in CEDIT 2021.10, and navigate

to the end of the Component Properties section.
10. Copy the following keyword to the end of the Component Properties section (right before the
Rock-Fluid Data section).

*SOLUBILITY
** HENRYC calculated at 50.00 deg C
*HENRYC 2.9260124E+05 0 5.9536754E+06 9.0417e6
*REFPH 8.8000000E+03 8.8000000E+03 8.8000000E+03 8.8000000E+03
*VINFINITY 3.5014918E-02 3.1807722E-02 3.5168231E-02 3.7E-02

11. Save the dataset as Exercise-3_H2-storage_Aquifer_solubility.dat. Then, run it with GEM 2021.10
through Launcher.

Results Interpretation

12. Add the Exercise-3_H2-storage_Aquifer_solubility.sr3 file to the Results project that you saved
(H2 storage in Aquifer.results) to compare the results of this model with those from Exercise-
1_H2-storage_Aquifer_base-model.dat.

35
The above plot shows that by including hydrogen solubility in water the production behavior slightly
changes from the base model. The cumulative produced gas during the three-month period goes down a
little bit, which is a result of hydrogen dissolution in water. The water production goes up a bit. This is due
to the earlier switch of the production constraint from STG to BHP in the solubility model. The effect of
hydrogen solubility does not seem to be very significant in the current model. This is due to the small
solubility of hydrogen in water.

13. In time series, select

a. Sectors  Exercise-1…  Aquifer-top  Gas Ave Sat SCTR. Then, click on Add To New
Plot.
b. Sectors  Exercise-3…  Aquifer-top  Gas Ave Sat SCTR. Then, click on Add Curve.

36
 14. Drag and drop IK 2D views of Exercise-1… and Exercise-3… on a Dashboard. Then, go to plane 23,
select Water Mole Fraction (H2) for property, and move the date to 2023-Mar-01.

The above plots show that in the Aquifer-top sector, the average gas saturation of the solubility model is
a bit lower than the base model throughout the cycle. This is due to the dissolution of hydrogen in water.
Also, the Water Mole Fraction (H2) profiles at the end of the production shows that approximately 600-
700 ppm of H2 is still dissolved in the aqueous phase. This hydrogen could get trapped in aquifer.

Later, in Exercise-5, we will see if including solubility has any effects on hydrogen trapping in the reservoir.

37
Exercise 4: Hydrogen Diffusivity in Water

1. Open the Exercise-3_H2-storage_Aquifer_solubility.dat model in CEDIT 2021.10, and navigate to

the end of the Component Properties section.
2. Copy the following keyword to the end of the Component Properties section (right before the
Rock-Fluid Data section).

*DIFFC-AQU 0 0 0 8.5E-05

Note: the above diffusion coefficient MUST have the unit cm2/sec no matter what unit system
is used for the simulation.

3. Save the model as Exercise-4_H2-storage_Aquifer_solubility&diffusivity.dat, and run it with

GEM 2021.10.

Results Interpretation

4. Open the sr3 file in Results project that you have and add the gas injection and production rates
to the gas rate plot that we made for the previous models. The plot shows that hydrogen diffusion
does not have any effects on the injection / production rates for the one cycle that we have
compared to the only solubility model. That is due to the small diffusion rate of hydrogen in water.

38
Exercise 5: Extending Simulation Time for Base, Hysteresis, Solubility,
and Diffusion Models

Extending Production (H2 Trapping)

The purpose of this exercise is to see how much hydrogen will remain in the reservoir if we drain the
aquifer to the producer’s original pressure in four years. We will do the following changes to the base,
hysteresis and solubility models for this purpose:

 Change the PRODUCER’s BHP MIN from 101.325 to 6500 kPa, which is the well block pressure of
the PRODUCER’s reference layer before the start of hydrogen injection.
 Extend the production period for 4 more years.

1. Open the Exercise-1_H2-storage_Aquifer_base-model.dat model in Builder 2021.10.

2. Go to Wells & Recurrent  Dates.
3. Click on Add a range of dates. Choose From: 2021-01-01; To: 2027-03-01, by Month. Click OK and
click OK again to the message that appears. Then, close the Simulation Dates window.
4. Go to Wells & Recurrent  Wells. Then, go to the Constraints tab for PRODUCER at 2020-01-01.
Change the BHP MIN to 6500 kPa. Then, click Apply and OK.
5. Save the model as Exercise-5_H2-storage_Aquifer_base-model_BHP6500_EXTEND.dat.
6. Repeat steps 1-5 for the Exercise-2_H2-storage_Aquifer_hysteresis.dat model, and save it as
Exercise-5_H2-storage_Aquifer_hysteresis_BHP6500_EXTEND.dat.

39
 7. Repeat steps 1-5 for the Exercise-3_H2-storage_Aquifer_solubility.dat model, and save it as
Exercise-5_H2-storage_Aquifer_solubility_BHP6500_EXTEND.dat.
8. Run these three models with GEM 2021.10.

Results Interpretation

9. Open the following three sr3 files in Results:

 Exercise-5_H2-storage_Aquifer_base-model_BHP6500_EXTEND.sr3
 Exercise-5_H2-storage_Aquifer_hysteresis_BHP6500_EXTEND.sr3
 Exercise-5_H2-storage_Aquifer_solubility_BHP6500_EXTEND.sr3
10. Plot the Cumulative Gas SC of PRODUCER for these three models on separate plots.
11. Use Formulas  Time Series to create a formula for Trapped Gas (%) that remains in the aquifer
for each of these models. Then, add these curves to the Cumulative Gas SC plot.

12. Save this Results project at Exercise-5_BONUS.results in the Student Solutions directory.

40
The above plot shows that under the current production scenario, the amount of the trapped injected gas
for each model is:

 ~ 9% for the base model

 ~ 11% for the solubility model
 ~ 41% for the hysteresis model

Extending Simulation for 100 Years (Diffusion)

Diffusion is a very slow process. Therefore, capturing its effect on saturation and composition distributions
requires a long period of time. The purpose of this part of the course is show the effect of hydrogen
diffusivity on the saturation and composition distribution in the aquifer. To show this, we will do the
following changes on the solubility and solubility&diffusivity models.

41
  Shut in the PRODUCER.
 Extend the simulation time for 100 year.

13. Open the Exercise-3_H2-storage_Aquifer_solubility.dat model in Builder 2021.10.

14. Go to Wells & Recurrent  Wells. Right click on the 2022-12-01 OPEN event for PRODUCER, and
select Delete event selected in the List. Then, click OK.
15. Go to Wells & Recurrent  Dates.
16. Click on Add a range of dates. Choose From: 2021-01-01; To: 2123-03-01, by Year. Click OK and
click OK again to the message that appears. Then, close the Simulation Dates window.
17. Save the model as Exercise-5_H2-storage_Aquifer_solubility_shutin-PROD_extend100Y.dat.
18. Repeat steps 13-16 for the Exercise-4_H2-storage_Aquifer_solubility&diffusivity model, and
save it as Exercise-5_H2-storage_Aquifer_solubility&diffusivity_shutin-PROD_extend100Y.dat.
19. Run these two models with GEM 2021.10 through Launcher.

Results Interpretation

20. Open the following two sr3 files in the Results project that you saved for exercise 5
(Exercise5_BONUS.results):
 Exercise-5_H2-storage_Aquifer_solubility_shutin-PROD_extend100Y.sr3
 Exercise-5_H2-storage_Aquifer_solubility&diffusivity_shutin-PROD_extend100Y.sr3
21. Use Formulas  Spatial Property to create a formula for Water Saturation * Water Mole
Fraction (H2) for each of these models. These formulas can show the amount of H2 in Water Phase
in the grid blocks.

42
 22. Drag and drop IK 2D views of the above two models a Dashboard. Then, go to plane 23. Go to
DASHBOARD tab and deactivate Reservoir Properties in the Synchronization section. Select the
formulas that you made for each model, and move the date to 2123-Jan-01.

The above profiles for the Sw * mole fraction of H2 in the water phase show that the transition zone light-
blue to dark-blue is thicker in the right profile. This means that in the solubility&diffusivity model (left
profile), more hydrogen exists in water phase (dissolved&diffused) at the deeper layers compared to the
solubility model. This is due to the diffusivity of hydrogen in water.

Exercise 6: Multiple Storage Cycles

In this exercise we will include hysteresis, solubility and diffusivity in our model, and we will add four more
storage cycles to the one that we have, to see how our aquifer behaves under five storage cycles.

1. Open Exercise-4_H2-storage_Aquifer_solubility&diffusivity.dat in CEDIT 2021.10.

2. Navigate to the end of the Rock-Fluid section, and add HYSKRG 0.3 there.

43
 3. Save the dataset as Exercise-6_H2-storage_Aquifer_five-cycles.dat. Then, open it in Builder
2021.10.
4. Got to Wells & Recurrent  Dates to open the Simulation Dates window.
5. Click on Add a range of dates on the top right corner of the window
6. Click on the Calendar in front of To, and select 2027-03-01.
7. Use the 1 Month step, and click OK to create the dates.

Adding Cycling Group to the model

We are going to operate the INJECTOR and PRODUCER wells with the following schedule using a
Cycling Group. The cyclic operation starts at 2022-03-01 with six month (184 day) of injection. Then,
it continues with 3 month (91 day) of injector and producer shut-in, and the last part of cycle is 3
month (90 day) of production. This cycle will repeat five times from 2022-03-01 to 2027-03-01.

H2 Injection Shut-in period Production

• Six month (184 • Three month • Three month

day) (91 day) (90 day)

8. Go to Wells & Recurrent  Groups. Name the first group as Field at 2021-01-01, then click on
Add new group. We cannot attach wells to the Field group since it is the top-level group.

44
9. Name the second group as Cyclic INJ-PROD at 2021-01-01, then click on OK.

10. Go to Wells & Recurrent  Groups  Field  Cyclic INJ-PROD, and click on Attach Wells.

45
11. Select both INJECTOR and PRODUCER in the Well-Group Attachment page. Then, click Apply and
OK.

12. Change the date to 2022-03-01 from the drop-down menu. then go to the Cycling Group tab.
13. Change the Cycle Part Number (NPARTS) to 3, and deactivate the Water Injection option.

14. In the Stream Target Types part, change all of the Production and Gas Injection cells to STG –
Stock Tank Gas.

46
 15. Activate the Stream Maximum Rates option, and set the following values for them.

16. Activate the Other Options part, and set the following Max Cycle Part Duration and Starting Time
Step Size for each cycle part. Then, click Apply.

The Cycling Group tab at 2022-03-01 should look like the following photo after these modifications.

47
17. From the Event drop-down menu, select Turn ON/OFF Cycling Group Control, then click Apply.

18. From the Event drop-down menu, select Set Starting Cycle Part. Make sure the value is 1, then
click Apply.

19. From the Event drop-down menu, select Set Ending Cycle Part. Make sure the value is 3, then
click Apply.

20. Now, the cycling group is set up. Click OK to close the Group Events window.
21. Save the model in Builder, and run it with GEM 2021.10 through Launcher.

48
Results Interpretation

22. In Launcher, drag and drop the Exercise-6_H2-storage_Aquifer_five-cycles.sr3 file on Results

2021.10.
23. Save the Results project as Exercise-6_mutiple cycles.results at the beginning.

24. In time series, select

a. Wells  Exercise-6…  INJECTOR  Gas Rate SC. Then, click on Add To New Plot.

b. Wells  Exercise-6…  PRODUCER  Gas Rate SC. Then, click on Add Curve.

c. Wells  Exercise-6…  PRODUCER  Water Rate SC. Then, click on Add Curve.

25. In time series, select

a. Sectors  Exercise-6…  Aquifer-top  Gas Ave Sat SCTR. Then, click on Add To New Plot.

b. Wells  Exercise-6…  INJECTOR  Gas Rate SC. Then, click on Add Curve.

49
The rates plot shows that the injection plan that we have for the current aquifer cannot deliver 500,000
m3/day of hydrogen for winter in the first few cycles. However, by adding more cycles to the storage
process, the aquifer reaches a point where it can deliver 500,000 m3/day of hydrogen for the whole three
months. Looking at the Average Gas Saturation plot, you can see that the average gas saturation in the
Aquifer-top sector increases with each gas injection cycle. That is due to the hysteresis, solubility and
diffusivity parameters that we have defined for the model. Some H2 will remain in the aquifer at the end
of each cycle, and this will help to build up the average gas saturation in the aquifer, and consequently, it
results in more H2 production in the last cycles. You should note that this production behavior could be
specific to the homogeneous anticline aquifer that we have in this exercise.

Adding Monitor Constraint to PRODUCER

The five-cycle exercise shows that operating the production well on STG MAX 500,000 m3/d and BHP MIN
101.325 kPa results in large water production rates (~ 3000 m3/day) at the end of each production cycle,
especially in the first three cycles. Our production facility is just capable of handling 200 m3/day of water.
Therefore, we need to stop production when the water rate hits this limit. To simulate this, we can add
monitor constraint to the production well.

26. Open Exercise-6_H2-storage_Aquifer_five-cycles.dat in Builder 2021.10.

27. Go to Wells & Recurrent  Wells  PRODUCER  2021-01-01  Constraints tab.

28. Using select new (in the Constraint column of the table), select MONITOR. Then select STW
surface water rate, MAX, 200 m3/d, SHUTIN. Then, click Apply and OK.

50
 This will shut in PRODUCR whenever its surface water rate reaches 200 m3/day.

29. Save the model as Exercise-6_H2-storage_Aquifer_five-cycles-monitor.dat, and run it with GEM

2021.10.

Results Interpretation

30. Add the Exercise-6_H2-storage_Aquifer_five-cycles-monitor.sr3 file to the Results project that

you saved (Exercise-6_mutiple cycles.results) to compare the results of this model with those
from Exercise-6_H2-storage_Aquifer_five-cycles.dat.
The following first plot is the injection gas rate and production gas and water rates for all five
cycles of the models. The second plot only shows the production rates for the first cycle.

51
 31. In time series, select

a. Wells  Exercise-6_H2-storage_Aquifer_five-cycles.dat  PRODUCER  Cumulative

Water SC. Then, click on Add To New Plot.

b. Wells  Exercise-6_H2-storage_Aquifer_five-cycles-monitor.dat  PRODUCER 

Cumulative Water SC. Then, click on Add Curve.

These above plots show that using the monitor constraint, the model can produce the same amount of
gas without having to produce 113,814 m3 of water. The cumulative water produced has reduced to
1,006.14 m3.

Exercise 7: Cushion Gas Injection

The results of Exercise-6_H2-storage_Aquifer_five-cycles-monitor showed that at the end of five storage

cycles 27,433,472 Sm3 of hydrogen remains in the aquifer. In the current exercise, we want to inject the
same surface volume of cushion gas into the aquifer, and see if it affects the hydrogen production. Please
note that in the current scenario we just want to add one period of cushion gas injection prior to the start
of the hydrogen injection / production cycles.

We will do this cushion gas injection in two scenarios. In the first one, N2 will be used as cushion gas,
assuming that it is not soluble in water. In the second one, CO2 will be used as cushion gas, and the CO2
solubility in water and the methanation reaction (CO2 + 4 H2 = CH4 + 2 H2O) will be included in model.

52
1. Open Exercise-6_H2-storage_Aquifer_five-cycles-monitor.dat in Builder, and save is as
Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat.
2. Go to Wells & Recurrent  Wells  INJECTOR  2022-03-01  Constraints tab.
3. Activate the Constraint definition option, and click Apply. This will write the constraints and
injected fluid set-up that we had at 2021-01-01 to 2022-03-01.
4. We will inject 27,433,472 Sm3 of N2 over six months (2021-01-01 to 2021-07-01) into the aquifer.
To do this, we should inject it at STG of 134,992 m3/day.
5. Go to the Constraints tab at 2021-01-01 for INJECROR. Change STG surface gas rate, MAX, to
134992 m3/day, CONT REPEAT. Click Apply.
6. Go to Injected Fluid tab. Write 1 for N2 mole fraction, and change the H2 mole fraction to 0.

7. For INJECTOR, right click on SHUTIN at 2021-01-01, and select Delete events selected in the list.

8. Go to the Options tab for INJECTOR at 2021-07-01. Activate the Status, and change its Type/Value
to SHUTIN. Then, click Apply and OK.

53
 9. Go to I/O Control  Simulation Results Output to open the Simulation Results File Writing
window. Click on the small green plus on the left side of the OUTSRF table to open the Select
New Date and Type window. Select Well at 2021-01-01, then click OK.

10. Select Production split information with compositions (PSPLIT) for the Well Information. This
will output the injection / production rate of each component into the sr3 file. Then, click OK to
close the window.

11. Save the model and run it with GEM through Launcher.

Results Interpretation

12. Add the Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.sr3 file to the Results

project Exercise-6_mutiple cycles.results. We want to compare the results of this model with
those from Exercise-1-6_H2-storage_Aquifer_five-cycles-monitor.dat.

54
The above plot shows that by adding a period of N2 injection, the aquifer can deliver 500,000 m3/day of
gas during the production periods. We need to see how much of this produced gas is H2 and whether or
not the hydrogen production has increased by injecting cushion gas into the aquifer.

13. In time series, select

a. Wells  Exercise-6_H2-storage_Aquifer_five-cycles-monitor.dat  PRODUCER  Gas

Mole Rate($C) SC  H2. Then, click on Add To New Plot.

b. Wells  Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat 
PRODUCER  Gas Mole Rate($C) SC  H2. Then, click on Add Curve.
c. Wells  Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat 
PRODUCER  Gas Mole Rate($C) SC  N2. Then, click on Add Curve.
14. Use Formulas  Time Series to create a formula for % difference of the cumulative produced gas
between the base model and the hysteresis model, and the other one for % difference of the
cumulative produced gas between the Exercise-1-6_H2-storage_Aquifer_five-cycles-monitor
model and the cushion gas model. We will plot this curve along with the cumulative mole
produced of H2 in each model.

55
15. In time series, select

d. Wells  Exercise-6_H2-storage_Aquifer_five-cycles-monitor.dat  PRODUCER 

Cumulative Gas Moles($C) SC  H2. Then, click on Add To New Plot.

e. Wells  Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat 
PRODUCER  Cumulative Gas Moles($C) SC  H2. Then, click on Add Curve.
f. Wells  Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat 
PRODUCER  Cumulative Gas Moles($C) SC  N2. Then, click on Add Curve.
g. Wells  Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-N2.dat 
PRODUCER  % difference in H2 CUM MOLE  H2. Then, click on Add Curve.

56
The above two plots show that the cumulative Hydrogen produced from aquifer is increased when we
inject cushion gas into the model. However, we need to consider that Nitrogen is also produced with
Hydrogen. The increase in the cumulative Hydrogen production is more in the first cycle (~28%), and it
decreases in the next cycles. This could be due to having only one Nitrogen injection period in this
scenario. Adding more Nitrogen injection cycles could increase the hydrogen production as the storage
cycles go forward.

Note: GEM outputs the production mole and mass rate of each component for separately. You can use
these outputs in the formula manager to calculate the volumetric rate of the components.

Exercise 8: Methanation Reaction

As noted in Exercise 6, cyclic hydrogen injection/production entails the loss of some of the injected
hydrogen due to the fact that hydrogen gets dissolved in the native reservoir fluids, or gets traps in the
pore space due to hysteresis.

In Exercise 7, it was tried to reclaim some of the lost hydrogen by injecting some inert cushion gas such as
nitrogen. However, nitrogen is not very cheap. Therefore, some field operators have come up with the
idea to use CO2 or flare gases as cushion to not only reduce the amount of lost hydrogen in the reservoir
but also sequester the CO2 that would be otherwise acting as a GHG and polluting the environment. While
this idea seems to be promising, one needs to know that the cushion CO2 can react with the chasing
hydrogen gas and produce methane and water through the following reaction:

CO2 + 4 H2 = CH4 + 2 H2O

57
This reaction is called methanation reaction and can be modelled in CMG GEM and STARS. In the following
steps it will be shown how a basic methanation reaction can be added to our H2 Storage model. The values
used for the reactions are generic and subject to tuning based on the user’s experimental data and field
observations. We will take the dataset of Exercise 7, and replace its cushion gas (N2) with CO2.

1. In Launcher copy the model (Exercise-7_H2-storage_Aquifer_five-cycles-monitor_cushion-

N2.dat), and save it as Exercise-8_Methanation.dat.
2. Open the file Exercise-8_Methanation.dat using CEDIT 2021.10 and navigate to the end of the
Component Properties section (right before the start of Rock-Fluid Data section) and add the
following lines there:

*GEOCHEM_V2
*REACTION-RATE-ARN 'CO2' + 4 'H2' = 'CH4' + 2 'H2O'
*REACTION-ORDER 'CO2' 1 'H2' 1
*FREQUENCY-FACTOR 1.0e-3
*ACTIVATION-ENERGY 0

In the above:
 The geochemical reaction between CO2 and H2 is flagged by the keyword GEOCHEM_V2.
 The line starting with the keyword REACTION-RATE-ARN specifies the stoichiometry of Arrhenius
based reaction rate expression. It is seen that one CO2 reacts with four H2 to produce one CH4 and
two H2O molecules.
 REACTION-ORDER defines the reaction order with respect to different reactants. In this case,
reaction order is one for both reactants which means there is a linear relationship between the
concentration of each reactant and the rate at which the reaction occurs.

58
  FREQUENCY-FACTOR, specified on the fourth line, is the frequency factor or pre-exponential
factor that is used in Arrhenius rate reaction. The value specified by this keyword defines the
number of successful collisions between the reactants per day. The word successful has been
underlines in order to underline the fact that not all the collisions between the reactants trigger
the reaction to proceed. A successful collision is the collision which leads to reactants eventually
convert to products. Please note the unit of the frequency factor is {1/day} in our case. The speed
of the reaction is directly proportional to the value of the frequency factor.
 ACTIVATION-ENERGY defines the level of dependency of the reaction to the current state of the
internal energy of the reactants. The value of zero makes the reaction independent of
temperature of the system.
3. While in the text editor, navigate to the initialization section of the dataset marked by the
keyword INITIAL. Since we are injecting CO2, it is numerically more stable if the model is initialized
with a little CO2 from the beginning. Under the ZGAS keyword which introduces the input of gas
cap composition enter the mole fraction of CO2, N2, H2O and H2 components in the gas cap fluid
respectively as shown below:

*ZGAS 0.1 0.9 0.0 0.0

At the first look, the composition of gas cap (which is above the top of the aquifer model) does
not seem to have any effect on the model. However, since an equilibrium state is assumed in the
model, CO2 and N2 will get dissolved in the aquifer according to their solubility in the aqueous
phase.

Now we need to define the CO2 cushion injection prior to hydrogen injection. We will do this part of the
tutorial in Builder.

59
4. Save the model. Close it, and then open it in Builder.
5. Go to Well & Recurrent  Wells  INJECTOR. Go to the Injected Fluid tab at 2021-01-01, and
change the injected fluid to 100% CO2 and the rate to half the nitrogen i.e. STG MAX 70,000
m3/day as shown below:

As calculated in Exercise 6 and noted in Exercise 7, at the end of five storage cycles 27,433,472
Sm3 of hydrogen remains in the aquifer. The adopted rate of CO2 (70,000 Sm3/day) means the
total CO2 volume of 27,433,472 Sm3 which should be ideally injected into the aquifer must occur
over a course of about 13 months. As such we will adjust the SHUTIN date of the INJECTOR to
2022-02-01 as shown below:

6. Save the model and run it with GEM.

60
Results Interpretation

Note: The only purpose of this exercise is to show how we can introduce the Methanation reaction into a
GEM model. The input reaction parameters are not accurate or reliable. Therefore, the results that we
present here do not apply to a real underground Methanation process.

7. Add the Exercise-8_Methanation.sr3 file to the Results project Exercise-6_mutiple cycles.results.

We want to compare the results of this model with those from Exercise-7_H2-
storage_Aquifer_five-cycles-monitor_cushion-N2.sr3.
8. Plot Gas Rate SC for INJECTOR and PRODUCER, and Water Rate SC for PRODUCER. The following
Rates plot shows that while having the Methantion reaction in the model, the production well has
stopped operating pretty fast in the production cycles. That is due the fact that Water is a product
of this reaction, and the production well in out model has a surface water rate monitor constraint.
Water production goes up by the reaction, and reaches the limit for shutting in the producer.

9. In time series, plot the followings on one plot:

 Wells  Exercise-6…  PRODUCER  Cumulative Gas Moles($C) SC  H2.
 Wells  Exercise-7…  PRODUCER  Cumulative Gas Moles($C) SC  H2, N2.
 Wells  Exercise-8…  PRODUCER  Cumulative Gas Moles($C) SC  H2, CO2, CH4.

61
The above plot shows that in the Methanation exercise, we are producing some CH4, which is only the
results of the Methanation reaction in the aquifer. H2 production has also halved compared to the
Cushion_N2 model. This decrease in the hydrogen production has two reasons. Firstly, H2 gets consumed
in the Methanation reaction, and secondly, operating the producer using a surface water rate constraint
stops the production cycles pretty fast. Therefore, a part of gas that could be produced remains in the
aquifer.

10. You can also go to the 2D / 3D views to see the mode fraction of each component in different
phases at the grid level. For example, the following profile is the mole fraction of CH4 in the gas
phase at the end of the last cycle.

62
Exercise 9: Cap-Rock Integrity (GEOMECHANICS)

In this part of the exercise we will add a simple geomechanics calculation to our base model (Exercise-
1_H2-storage_Aquifer_base-model.dat) to see if the cap rock reaches its failure criteria and crosses from
elastic into plastic behavior. For this purpose, we are going to do the followings to the base model:

 Changing the top K Layer (K = 1) to cap rock by assigning extremely low permeability to it.
 Removing VOLMOD 10000 from the bottom part of aquifer.
 Defining two separate geomechanics rock types, one for the aquifer and the other one for the
caprock (K Layer 1).
 Changing the injection/production scenario by injecting H2 at high pressure into the aquifer.

Note: the geomechanics inputs that are used in this part are just two samples to demonstrate how we can
model cap rock failure using GEM. The practical geomechanical properties need to either be measured in
the laboratory or found in literature. Also, more advanced capabilities and applications of geomechanics
in CMG can be found in our Geomechanics course.

11. Open the base model (Exercise-1_H2-storage_Aquifer _base-model.dat) in Builder 2021.10, and
save it as Exercise-9_H2-storage_Aquifer_caprock-integrity_originalBHP.dat at the beginning.
12. Click on Specify Property to open the General Property Specification window.
13. Enter the following values for Permeability I, Permeability J, and Permeability K, and Volume
modifiers. You will first need to use the Delete key on your keyboard to delete the values are in
the Whole Grid cell for these properties. Then, click OK.
A message may show up regarding the rescue file; click OK on it. Make sure Permeability I,
Permeability J, and Permeability K are selected on the Block/Corner Value Calculation window.
Then, click OK on that window.

63
14. Go to Geomechanics  Calculation and Numerical Options in order to enable geomechanics.
The window which appears consists of three sections: General Parameters, Plastic Deformation
Solution Control Parameters, and Matrix Solver Control Parameters.

15. For this model, we will use a simple 3D Elasto-plastic Mohr-Coulomb model. Check the "3D"
checkbox for Finite Element Method Dimension 2D or 3D (GEOM3D).
16. Keep the Coupling Option (GCOUPLING) as default (2). This will apply two way coupling in the
simulation in which the fluid-flow porosity is a function of pressure, temperature and total
mean stress.
17. Max Iterations for solving the Force Balance Equations (INTERGEO) should be changed to 500.
In some situations, more iterations may be required for convergence to be achieved.

64
18. After selecting this option, click OK on the window. A message will appear stating that “Data is
required for at least one rock type”. Press Ok on this message.
19. You will see that the Geomechanics Section in the tree view has a Red X, meaning that it is
incomplete and requires more input. Go to Geomechanics  Geomechanical Rock Types.
20. In this window, click on the small triangle that is in front of Rock Type and Add New Rock Type. A
message will appear informing to be careful about the definition of Geomechanical and Reservoir
Rock Types.

21. For this model, we will use Elasto-Plastic Mohr-Coulomb model, which is the default model.

65
22. For a Simple Mohr-Coulomb calculation, we can only define 3 values for the model to be able to
calculate the stresses and strains. These values are the Young’s Modulus, Poisson’s Ratio, and
Cohesion. Input the values as shown below, then click Apply and OK. The Geomechanics section
will turn green.
A huge value is chosen for the Cohesion of this rock type, so that it does not reach the failure
point. This rock type will be assigned to the aquifer.

23. Repeat step 22 to add the second rock type using the following information. Rock Type 2 will later
be assigned to the K Layer 1 as cap rock. Then, Click Apply and OK to close the Geomechanical
Rock Types window.

66
24. Go to Geomechanics  Array Properties to open Specify Geomechanics Properties window.

67
25. Select the Deformation Rock Type property from the drop-down menu. Change its data type to
KVAR and fill the values for different K Layers based on the following picture. Then, click OK.

26. Go to Geomechanics  Geomechanics Stress  Initial Stress Distribution. Then, enter the
following values for the six components of stress in the Stress section of the STRESS Input
window. Then, press OK.

27. With the addition of Geomechanics to the dataset, additional outputs need to be defined in order
to visualize the required information in Results. Go to I/O Control  Simulation Results Output.

68
28. Change the frequencies to the following values. Also, define the Grid Deform under the WSRF.

29. Click on the Select button for the Grid Output Selection. In the window which appears enable
outputs for the following, then click OK.

BIOT Biot’s constant

BULKVOL Bulk Volume
GCOHESION Cohesion Value
GEORTYPE Rock type number for geomechanical material
PERM Permeability in each direction
POISSON Poisson’s ratio
PORDIFF Difference in geomechanics porosity and reservoir porosity
POROS Current porosity
PRES Pressure
SAFACTOR Safety Factor
SG Gas Saturation
SO Oil Saturation
STRAINVOL Volumetric Strain
STRESEFF Mean Effective Stress
STRESMNP Mean Principle Stress
STRESMXP Maximum Principle Stress
STRESSM Mean Total Stress
STRNEPL Effective Plastic Strain
STRNMNP Minimum Principle Strain
STRNMXP Maximum Principle Strain
SUBSIDGEO Vertical Displacement based on geomechanics
SW Water Saturation
TEMP Temperature
TSAFACTOR Tensile safety factor

69
 TSUBSIDGEO Vertical Displacement Down based on geomechanics, at top of cell
TVERDPLGEO Vertical Displacement Up based on geomechanics, at top of cell
VDISPL Displacement vector on each grid block
VERDSPLGEO Vertical Displacement Up based on geomechanics, block center
VPOROSGEO Porosity calculated form geomechanics module
VPOROSTGEO True porosity calculated from geomechanics module
WALL Mole fraction of all components in the aqueous phase
YALL Mole fraction of all components in gas phase
YLDSTATE Yield State
YOUNG Young’s Elastic Modulus
ZALL Global model fraction of hydrocarbon components

30. Add Monitor Constraint to shut the injector when STG drops below 1000 m3/day

31. Go to Wells & Recurrent  Wells. Select the 2022-09-01 SHUTIN event for INJECTOR and the
2022-12-01 OPEN event for PRODUCER. Right click and select Delete events selected in the list.
Then, click Apply and OK.

70
71
 32. Go to Wells & Recurrent  Wells (2)  INJECTOR  2021-01-01 PERF to open Well Completion
Data (PERF) window.

33. In this window, go to the Perforation tab and delete perforation 48 23 1 for both H2-INJECTOR
and PRODUCER. With the change that we made to K Layer 1, this perforation is in cap rock and
needs to be deleted.
34. Save the model as Exercise-9_H2-storage_Aquifer_caprock-integrity_originalBHP.dat.
35. Go to Wells & Recurrent  Wells. Go to the Constraints tab for INJECTOR (at 2021-01-01) and
change BHP to MAX 17000 kPa. Then, click Apply and OK.
36. Save this model as Exercise-9_H2-storage_Aquifer_caprock-integrity_increaseBHP.dat.
37. Run these two models with GEM 2021.10.

Results Interpretation

38. Open the following two sr3 files in Results 2021.10:

 Exercise-9_H2-storage_Aquifer_caprock-integrity_originalBHP.sr3
 Exercise-9_H2-storage_Aquifer_caprock-integrity_increaseBHP.sr3
39. Since we are just injecting gas into the model to increase its pressure, we will not be comparing
the fluid rates and reservoir property changes due to geomechanics. We will only check if the
Yield State flag of the cap rock grid blocks (K Layer 1) has changed from 0 to 1. If that happens, it
means that the grid block has reached its failure criteria and crossed from elastic behavior into
plastic.

72
 Drag and drop Aerial views of the above two models a Dashboard. Change the property to Yield
State, and move the Date to the end of the run (2023-Mar-01). All the cap rock grid blocks of the
orginalBHP model have Yield State of 0, but some of the cap rock grid blocks of the increseBHP
model have Yield Sate of 1 at the end of the run. These parts of the cap rock have reached their
failure criteria and crossed from elastic into plastic behavior.

In this example, we have modeled geomechanics in a simplified way; we are just showing a workflow on
how a shear failure of the caprock can be modeled. However, the escape of H2 into the overburden is not
modeled. You can also use Barton-Bandis model to model both the tensile and shear failures of caprock,
and also model leakage of H2 into overburden. This workflow is available in our Geomechanics course.

73
Exercise 10: GEM – CoFlow Coupling to Model Surface Facilities

In this exercise, we will show how GEM reservoir model is coupled with Well/Facility network in CoFlow
to model an example of integrated underground H2 storage including all elements of reservoir, well and
surface facility.

To achieve such an integrated modelling, sections such as Fluids, Reservoirs, Wells, Facilities and
Simulations are required to set up appropriately in CoFlow.

Importing the WinProp generated fluid model into CoFlow

1. Open CoFlow from your desktop or its installation directory. Create a new project: Training and
a new study: H2 FACILITY, and click Open Study.
2. There should be a blank Base Case in the H2 FACILITY study once CoFlow gets initiated. Rename
the new case to H2 FACILITY.
3. The unit system drop-down menu is at the bottom right corner of the window. Make sure it is SI.

4. Go to Case Building  Fluids  Create guided task . Select EOS From EOS File, and enter
H2_EOS for the Model Name and Task Name. Click OK.

74
5. Go to Fluids: H2_EOS  Import EOS File  Edit .

6. In the Import EOS File window, click on the green arrow button (Select and EOS file in either GEM
or Eclipse keyword format) and browse for the gem file provided (WinProp_H2 Storage in Aquifer
– CoFlow.gem), which can be found under the Required Data folder. Open this file. Then, click OK

7. Click on the play button to validate the imported file. The stage will turn green, which indicates
the imported file has been validated.

75
 8. Go to Fluids: H2_EOS  EOS Model Creation  Edit to validate the ranges of pressure and
default correlations and tuning factor. Keep all the default values as shown below.

9. Click OK and then click the play button to validate the default ranges and correlations for the
previous guided task. The section of fluids should show all the validated stages in green.

Importing GEM .dat file into CoFlow

10. Go to Case Building  Reservoirs  Create guided task . Select Using IMEX/GEM Dataset
from the Guided Task list and enter GEM Model for both Subsurface Element Name and Task
Name sections. Then Click Ok.

76
11. Go to Reservoirs: GEM Model  IMEX/GEM Dataset  Edit . Set the directory path of the
GEM .dat file which is Exercise-10_H2-storage-surface-facilities_Aquifer-model.dat and can be
found from the Required Data folder.

12. Select Exercise-10_H2-storage-surface-facilities_Aquifer-model.dat and click Open. The Import

IMEX/GEM datasets window will show the name of all the well names exist in the GEM dataset.
User is required to check the Include box of those only wells that will be coupling with the defined
wells in CoFlow in next stages. In this exercise, we include both INJECTOR and PRODUCER wells.
Click OK.

77
 13. Click the play button to validate the Reservoirs guided task. Green check mark indicates
GEM model has been successfully imported into CoFlow.

Creating the Wells

In this part of the exercise, we will define wells, completions and constraints of wells for the CoFlow well
model. Then we couple wells from GEM .dat file into the wells have been created in CoFlow Wells stage.

14. Go to Case Building  Wells  Create guided task . Keep Integrated Multi-Well
Construction option and enter INJnPROD for the Task Name. Click OK.

15. Go to Wells: INJnPROD  Define Wells  Edit .

16. Click on the Import new wells icon. We will import the well trajectories from a RESCUE files.
Browse to the Required Data / RESCUE_for_CoFlow folder and open the
well-trajectory_RESCUE.bin RESCUE file. Select all the wells to import, and click OK.

78
17. The well trajectory information will be loaded in the Define Wells dialog. By default, all the wells
are defined as producers.

18. Use the Well Definition section to define the following details:

Table 1: Well Definition

Well Name Type Status
INJECTOR Injector Open
PRODUCER Producer Is Shut-in

79
19. To changes the well type, select Injector or Producer from the drop-down menu under Well
Definition section. Click OK after the selection, then click on the play button to validate this
guided task.

20. Now we need to define perforations and other completion equipment for each well. Go to Wells:
INJnPROD  Perforations and Equipment  Edit .

21. Select INJECTOR, click on Add a new section , set the Section Type to Tubing and End MD of
the tubing to be 650 m. Set the Fidelity of the tubing to High Fidelity, the Absolute Roughness to
0.001524 cm and the Pressure Drop Model to Homogeneous. Leave the default values for the
thermal option, overall heat transfer coefficient, and inside diameter of the tubing.

80
 22. Click on to add a new section right after the tubing (in the Unperforated section). Set the
Section Type to Perforation and enter the End MD as 706 m. Enter the Radius value as 0.0762 m.
Uncheck Attach to Reservoir.

23. Also uncheck the option Infer From Reservoir in the Fluid Model section. Then, select Fluid Name
H2_EOS from the drop-down menu.

24. Since we will be creating a facility for this case, the composition and temperature of the injected
fluid will be taken from the facility. Check the Infer From Facility box, and click OK

For explicit coupling, IMEX/GEM will run the reservoir simulation and provide CoFlow with an IPR curve
at the end of each time step. Therefore, all the wells that are defined in CoFlow need to be attached to
an IPR curve. This IPR curve will change during the run. However, we need to supply a fake IPR curve to
validate the current stage of CoFlow study.

25. Select INJECTOR. Click on the perforation, then on the button for IPR Model. Select Multiphase
Table as the IPR Model, and enter the numbers from the following photo for INJECTOR. Then,
click OK.

81
26. We can copy the equipment from INJECTOR to PRODUCER. On INJECTOR, select the two pieces
of equipment and click on the copy button.

82
27. Go to PRODUCER and click on the tubing hanger and then click on the paste button to add the
equipment copied from INJECTOR.

28. Add the End MDs of the equipment, 650 m for the tubing, and 706 m for the perforation. For the
fluids model, uncheck Infer from Reservoir, and select H2_EOS from the dropdown menu. Also,
input the reservoir temperature as 50 degC.

29. We have to make changes on the IPR table to be compatible with producer wells. Click on the
perforation equipment, then on the IPR model section and replace the values in table with the
following numbers provided. Then, click OK.

83
30. In the same dialogue, go to the section of Pressure Dependent Composition  Table. Add two
rows to using the green plus button, and write the following dummy values in the table. Click OK.

31. Click OK on the Perforations and Equipment window, and click play to validate the stage.

84
Note: Remember, these are just proxy perforations. Any numbers entered are only for data completion
purposes, and real perforation details, IPRs, and well indices, will be provided by the reservoir simulator.

32. Go to Wells: INJnPROD  Specify Well Constraints  Edit . Here, we will enter the well

constraints. Select INJECTOR first. Add two constraints as shown below:

 Tophole Gas Rate SC Inst <= 134992 m3/day

 Bottomhole Pressure <= 11000 kPa

33. While you are selecting the Tophole Gas Rate SC Inst constraint of INJECTOR, add a row to the
time profile table. Set its date to March 1, 2022, and value of 224565.2 m3/day.

34. Select PRODUCER, and Add two constraints as shown below.

 Tophole Gas Rate SC Inst <= 0.00 STB/day

 Bottomhole Pressure >= 101.325 kPa

85
35. While you are selecting the Tophole Gas Rate SC Inst constraint of PRODUCER, add a row to the
time profile table. Set its date to December 1, 2022, and value of 400000 m3/day.

36. Click OK and click play to validate the stage.

37. Skip the Specify Perforations Event.

38. Go to Wells: INJnPROD  Specify Well Simulation Time Events  Edit . Here we will define
the followings:

 H2 injection cycles: will open the H2 injector every year at March 1st, and shuts it in 6 month
later at September 1st.

 Production cycles: will open the producer every year at December 1st, and shuts it in 3
month later at March 1st of the next year.

 Water rate monitor constraint for PRODUCER: will shut it in if its water production rate
gets larger than 200 m3/day.

86
39. For INJECTOR, Add a simulation time event, and name it as H2 Injection Cycle. Then, add a
simulation time event condition for the event that you just made.

40. Set the condition to Recurrent time and set the followings for it. Then, click OK.

41. Add a simulation time event action for the above event, and set it to Start up well.

87
42. Select the following options for this action. Then, click OK.

43. For PRODUCER, Add a simulation time event, and name it as Production Cycle. Then, add a
simulation time event condition for the event that you just made.

44. Set the condition to Recurrent time and set the followings for it. Then, click OK.

45. Add a simulation time event action for the above event, and set it to Start up well.

88
46. Select the following options for this action. Then, click OK.

47. For PRODUCER, add the second simulation time event, and name it as Water Rate Monitor. Then,
add a simulation time event condition for the event that you just made.

48. Set the condition to Well property monitor and set the followings for it. Then, click OK.

49. Add a simulation time event action for the above event, and set it to Shut-in well. Then, click OK.

89
 50. Click OK to close the Specify Well Simulation Time Events window. Then, click play to validate
the stage.

Coupling GEM Model Wells to CoFlow Model Wells

51. Go to Case Building  Wells  Create guided task . Select IMEX/GEM Well Coupling from
the Guided Task list and enter Coupled Wells for the Task Name. Click Ok.

52. Go to Wells: Coupled Wells  IMEX/GEM Well Coupling  Edit . The Well Coupling window
lists the available wells in CoFlow model (left table) and GEM model (right table).

90
53. Click on Auto Couple, Since the name of the wells are the same in both GEM and CoFlow models.
Couple option can be used in cases where the names of the wells are different between GEM and
CoFlow.

91
 54. Click Ok and then click play to validate the stage.

Creating Injection and Production Facilities

55. Go to Case Building  Facilities  Create guided task . Select Interactive Schematic and enter
the Facility Name and Task Name: Injection Production Facilities. Click OK.

56. Go to Facilities: Injection Production Facilities  Create Production Facility  Edit . This opens
up the canvas to design and create your facility schematic.

57. Set the Fluid Model to H2_EOS.

58. The Injection Production facilities will eventually look like this:

92
59. We will first create the injector well head. Click the Attach all Injector Wells to Injector Wellheads

icon. This will create well heads attached to the various injectors in the reservoir model.
60. Select the wellhead, and then right click the Compressor in the Equipment Palette and select
Create Outbound Connection for Compressor.

61. Select the compressor, and then right click the Gathering Center in the Equipment Palette and
select Create Outbound Connection for Gathering Center.

62. Drag and Drop two Sources from the Equipment Palette onto the canvas. Rename the one of the
Sources to H2 Source, and the other one to N2 Source. Connect the outlet of each source to the
Gathering Center as shown below.

93
63. Select the N2 Source, and input the followings for it:

 Shut-in after July 1, 2021 in the Operational Status section

 P=500 kPa, T=25 C in the Data section

 N2=1 in the Component Fractions section

64. Select the H2 Source, and input the followings for it:

 Shut-in until March 1, 2022 in the Operational Status section

 P=500 kPa, T=25 C in the Data section

 H2=1 in the Component Fractions section

94
65. Select the compressor, and only change the Thermal Option to isothermal. Make sure that
there is no value for the Pressure Change of the compressor (the pressure change cell should be
empty).

66. Click the Attach all Producer Wells to Producer Wellheads icon.

67. Select the producer wellhead, and then right click the Valve in the Equipment Palette and select
Create Inbound Connection for Valve. This will automatically connect the wellhead to the valve.

95
68. Drag and drop a 2 phase separator into the canvas and connect it to the valve.

69. Right click on the connector between the valve and separator and input a constraint of Pressure
= 101.325 kPa for it. Click OK

96
 70. From the equipment palette, drag and drop two sinks and connect them to the output ports of
the separator. Change the default names in the sinks to Water and Gas as shown below.

71. Click OK and click play to validate the stage.

Set-up Integrated Network Simulations

72. Go to Case Building  Simulations  Create guided task . Select Standard PSD and enter the
Task Name: H2 Storage. Click OK.

73. Go to Simulations: H2 Storage  Specify Problem Type and Solver  Edit . Set the Problem
Type to Time stepping problem.

74. Set the Start Date to January 1, 2021 and End Date to March 01, 2027.

75. Change the Maximum time step possible to 30 day, and the Time step passed in with initial value
to 0.5 day, and select the Well constrained option from the dropdown list for Prior Solve. Leave
the other Numerical Settings to default.

76. Go to the Numerical Settings tab  Auxiliary tab and change the default solver to IMEX/GEM
(Explicit Coupled). Make sure GEM Simulator Path has been set successfully (The simulator path
on your computer could be different from what it is shown in the following image, make sure you
set the correct path to GEM in this window. GEM can be found in two locations: CMG installation

97
 directory where the CMG package has been installed, or in the CoFlow installation folder). Parallel
simulation of GEM model can also be defined in Core Count section. We keep this to 1 core in this
exercise.

77. Click OK, then Click Play .

78. Skip the Specify Simulation Time Event, Specify Time Series Recording in Dynamic Models and
Specify Time Series Recording in Geomechanical Models stages.

79. Uncheck the skip for Specify Profile Recording stage, then Edit the data for this stage . Set the
Start Time to January 1, 2021, the Recurrence Interval to 3, and the Recurrence Type to Months
Check Step to, and Record all Profiles.

98
80. In the Time Series tab, click Record all time series. Click OK.

81. Put a stop action (red button) before Run Simulation to validate without running the model.

82. Click to play.

83. Remove the stop action by clicking on the red button.
84. Right Click on the case H2 FACILITY and choose Schedule Job, click Ok to run the simulation.
85. Go to Job Monitor to check the run progress.

86. You can go to the Visualization section in CoFlow to display the results. Working with this section is
pretty similar to how we work with Results. The following plot is the Tophole Gas Rate SC for
INJECTOR and PRODUCER.

99