Both sides previous revision
Previous revision
Next revision
|
Previous revision
|
solvers:glads [2017/06/22 13:21] gag [SIF contents] |
solvers:glads [2019/09/06 13:17] (current) gagliar [Known bugs] |
| |
==== General Informations ==== | ==== General Informations ==== |
* **Solver Fortran File:** ''GlaDSCoupledSolver.F90'' and ''GlaDSchannelSolver.F90'' | * **Solver Fortran Files:** ''GlaDSCoupledSolver.F90'' and ''GlaDSchannelSolver.F90'' |
* **Solver Name:** ''GlaDSCoupledSolver'', ''GlaDSsheetThickDummy'' and ''GlaDSchannelOut'' | * **Solver Names:** ''GlaDSCoupledSolver'', ''GlaDSsheetThickDummy'' and ''GlaDSchannelOut'' |
* **Required Output Variable(s):** ''Hydraulic Potential'', ''Sheet Thickness'' and ''Channel Area'' | * **Required Output Variable(s):** ''Hydraulic Potential'', ''Sheet Thickness'' and ''Channel Area'' |
* **Required Input Variable(s):** None | * **Required Input Variable(s):** None |
* **Optional Output Variable(s):** ''Vclose'', ''Wopen'', ''Water Pressure'', ''Effective Pressure'', ''Sheet Discharge'', ''Sheet Storage'' and ''Channel Flux'' | * **Optional Output Variable(s):** ''Vclose'', ''Wopen'', ''Water Pressure'', ''Effective Pressure'', ''Sheet Discharge'', ''Sheet Storage'', ''Flux from Moulins'' (nodal variables) and ''Channel Flux'' (edge variable, to be exported in ''GlaDSchannelOut''). |
* **Optional Input Variable(s):** ''Zb'' | * **Optional Input Variable(s):** ''Zb'' |
| |
| ==== Known bugs ==== |
| If in parallel a moulin belongs to two partitions, the flux from the moulin is taken into account twice. This is may be not a bug, but the partitioning should avoid to have moulins at the boundary of partitions. The python tool ''makemoulin.py'' (in ''[ELMER_TRUNK]/elmerice/Meshers'') take care to have no duplicated moulins on the boundaries of partitions. |
| |
| If running the solver on a 3d internally extruded mesh, one should specify ''Preserve Edges = True'' and ''Preserve Baseline = True'' in the Simulation section. |
| |
| If running on a "true" 3d mesh, the ''GlaDSchannelsSolver'' has to be executed on the 3d body (not the bottom surface body as for the two other GlaDS solvers). |
| |
| |
| |
* ''GlaDSCoupledSolver'' is the main solver and couple the solve of the 3 main variables ''Hydraulic Potential'', ''Sheet Thickness'' and ''Channel Area''. Detail on the keywords for this solver are given below. | * ''GlaDSCoupledSolver'' is the main solver and couple the solve of the 3 main variables ''Hydraulic Potential'', ''Sheet Thickness'' and ''Channel Area''. Detail on the keywords for this solver are given below. |
* ''GlaDSsheetThickDummy'' is just a solver to declare the ''Sheet Thickness'' variable as a primary variable. | * ''GlaDSsheetThickDummy'' is just a solver to declare the ''Sheet Thickness'' variable as a primary variable. |
* ''GlaDSchannelOut'' has two functions: declare that the ''Channel Area'' variable is an edge variable (''Element = "n:0 e:1"'') and create output vtk files for edge variables. | * ''GlaDSchannelOut'' has two functions: declare that the ''Channel Area'' variable is an edge variable (''Element = "n:0 e:1"'') and create output vtu files for edge variables. |
| |
Currently (June 2017), ''GlaDSchannelOut'' doesn't support parallel simulation. These solvers only work in transient. They can be executed either on a 2d plane view mesh defining the bedrock or on the boundary of a 3d mesh. More details about the specificity of the solvers are given below. | Since version Version 8.3 (Rev: b213b0c8), ''GlaDSchannelOut'' works for parallel simulation (no more vtk or acscii output, only vtu). These solvers only work in transient. They can be executed either on a 2d plane view mesh defining the bedrock or on the boundary of a 3d mesh. If using internal extrusion within Elmer see the [[mesh:structuredmesh|structured mesh]] page for essential keywords to preserve baseline and edges. More details about the specificity of the solvers are given below. |
| |
| |
Neglect sheet Thickness in Potential = Logical True | Neglect sheet Thickness in Potential = Logical True |
| |
! choices are EXPLICT, CRANK-NICHOLSON, IMPLICIT | ! choices are EXPLICT, CRANK-NICOLSON, IMPLICIT |
Channels Integration method = String "Crank-Nicholson" | Channels Integration method = String "Crank-Nicolson" |
Sheet Integration method = String "Implicit" | Sheet Integration method = String "Implicit" |
| |
Exported Variable 1 = -dofs 1 "Vclose" ! closure velocity of the water sheet layer | Exported Variable 1 = -dofs 1 "Vclose" ! closure velocity of the water sheet layer |
Exported Variable 2 = -dofs 1 "Wopen" ! opening velocity of the water sheet layer | Exported Variable 2 = -dofs 1 "Wopen" ! opening velocity of the water sheet layer |
Exported Variable 4 = -dofs 1 "Water Pressure" ! water pressure at the base | Exported Variable 3 = -dofs 1 "Water Pressure" ! water pressure at the base |
Exported Variable 5 = -dofs 1 "Effective Pressure" ! effective pressure at the base | Exported Variable 4 = -dofs 1 "Effective Pressure" ! effective pressure at the base |
Exported Variable 6 = -dofs 2 "Sheet Discharge" ! water discharge (vector) in the water sheet layer | Exported Variable 5 = -dofs 2 "Sheet Discharge" ! water discharge (vector) in the water sheet layer |
Exported Variable 7 = -dofs 1 "Sheet Storage" ! storage in the water sheet layer | Exported Variable 6 = -dofs 1 "Sheet Storage" ! storage in the water sheet layer |
| Exported Variable 8 = -dofs 1 "Zs" |
| Exported Variable 9 = -dofs 1 "Zb" |
| Exported Variable 10 = -dofs 1 "Flux from Moulins" ! flux of water from the moulins (Qm) |
| |
Linear System Solver = Direct | Linear System Solver = Direct |
</code> | </code> |
| |
''GlaDSchannelOut'' allows to declare the ''Channel Area'' variable as an edge variable and to save edge variables. Output file formats are vtk or ascii. | ''GlaDSchannelOut'' allows to declare the ''Channel Area'' variable as an edge variable and to save edge variables. Output file formats is VTU. |
<code> | <code> |
Solver 3 | Solver 3 |
Exported Variable 1 = -dofs 1 "Channel Flux" | Exported Variable 1 = -dofs 1 "Channel Flux" |
| |
VTK OutPutFile = Logical True | VTU OutPutFile = Logical False ! set to TRUE to have VTU output |
ASCII OutPutFile = Logical False | VTU BinaryFile = Logical False |
| |
| Channels OutPut Directory Name = String "results" |
Channels OutPut File Name = String "$namerun"_channels" | Channels OutPut File Name = String "$namerun"_channels" |
End | End |
End | End |
</code> | </code> |
* the possibility to impose water flux at some nodes in the domain (moulins type inflow): | * the possibility to impose water flux at some nodes in the domain (moulins type inflow). The nodes have to be declared as node element (101) in the mesh (''mesh.header'' and ''mesh.boundary'' files have to be modified by hand for that, or using the python tool ''makemoulin.py'' - see below). |
<code> | <code> |
Boundary Condition 4 | Boundary Condition 4 |
</code> | </code> |
| |
| |
| ==== Making a mesh with Moulins ==== |
| Moulins are 101 boundary elements. ElmerGrid does not export correctly 101 boundary elements from gmsh or when partitioning a mesh. To add 101 boundary elements to an existing mesh, thanks to Mondher Chekki (IGE), one can use the python tool ''makemoulin.py'' (in ''[ELMER_TRUNK]/elmerice/Meshers''). |
| |
| Usage: |
| <code> |
| python makemoulin.py --meshdir mesh_dir --moulin moulin_file --partition number_of_partition |
| </code> |
| where ''moulin_file'' is an ascii file which contains the (x,y) coordinates of the moulins. The same file has to be used in gmsh so that nodes with the moulin coordinates already exist. |
| |
==== Examples ==== | ==== Examples ==== |
An example using the //GlaDS// Solver can be found in ''[ELMER_TRUNK]/elmerice/examples/GlaDS''. | Examples using the //GlaDS// Solver can be found in ''[ELMER_TRUNK]/elmerice/Tests/''. |
| |
==== Reference ==== | ==== References ==== |
The description of the GlaDS model is in:\\ | The description of the GlaDS model is in:\\ |
Werder M.A., I.J. Hewitt, C.G. Schoof and G.E. Flowers, 2013. Modeling channelized and distributed subglacial drainage in two dimensions. Journal of Geophysical Research: Earth Surface, 118(4), 2140-2158. | * Werder M.A., I.J. Hewitt, C.G. Schoof and G.E. Flowers, 2013. Modeling channelized and distributed subglacial drainage in two dimensions. Journal of Geophysical Research: Earth Surface, 118(4), 2140-2158.\\ |
| The implementation of the GlaDS model in Elmer/Ice is described here:\\ |
| * Gagliardini O. and M. Werder, 2018. Influence of increasing surface melt over decadal timescales on land-terminating Greenland-type outlet glaciers, Journal of Glaciology, 64(247), 700-710, doi:10.1017/jog.2018.59\\ |
| Results of the SHMIP experiments using Elmer/Ice are discussed in the SHMIP paper:\\ |
| * De Fleurian, B., M. Werder, S. Beyer, D. Brinkerhoff, I. Delaney, C. Dow, C., J. Dows, O. Gagliardini, M.J. Hoffman, R. LeB Hooke, J. Seguinot, A.N. Sommers, 2018. SHMIP The subglacial hydrology model intercomparison Project. Journal of Glaciology, 1-20. doi:10.1017/jog.2018.78 |
| |