Free surface tractions #3058

fionaclerc · 2019-05-27T22:29:02Z

This pull request add an exception to allow boundary tractions to be applied on the free surface. It also includes benchmarks for the surface deformation of viscous and viscoelastic materials in response to surface loading.

naliboff

Outstanding, thank you for contributing these benchmarks! I've done an initial review and on the whole this is PR is in good shape.

A few general comments and requests.

Please remove the example_1.prm
Would you mind include a python script in the benchmarks folder that calculates the analytical solution?

Nice work and great to see the viscoelasticity implementation is working at intended!

naliboff · 2019-05-27T22:58:40Z

benchmarks/free_surface_tractions/viscoelastic/README

+ solution found in the 'NL82_soln' directory, using the provided
+ gnuplot script ('compare_VE_def' for maximum surface deflection
+ through time, 'compare_VE_def_profile' for deflection of profile
+ through time).


Would you be willing to put a python script in this directory that calculates the analytical solution and writes out these files?

naliboff · 2019-05-27T22:59:30Z

benchmarks/free_surface_tractions/viscoelastic/compare_VE_def

@@ -0,0 +1,7 @@
+set title "Depression of surface in response to loading (inst. from 0 to 1000 years), over viscoelastic medium"


Can you add a brief comment here explaining what this file does?

naliboff · 2019-05-27T22:59:38Z

benchmarks/free_surface_tractions/viscoelastic/compare_VE_def_profile

@@ -0,0 +1,11 @@
+set title "Depression of surface in response to loading (inst. from 0 to 1000 years), over viscoelastic medium"


Same as above.

naliboff · 2019-05-27T23:01:37Z

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_2D_loading.prm

+
+# Mesh refinement specifications 
+subsection Mesh refinement
+  set Initial adaptive refinement        = 1


Do the results change significantly if you do not do the adaptive refinement? If so, it might be worth putting in a brief comment mentioning this.

naliboff · 2019-05-27T23:02:51Z

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_2D_loading.prm

+        # r0 is load radius, H0 is load height, t1 is time load is removed, 
+	# rhoi is density of ice/load, g is gravitational acceleration 
+    set Function expression = 0; if (x<r0 && t<t1,-g*H0*rhoi*(1-t/t1), 0)
+    #set Function expression = 0; if (x<r0 ,if(t<t0,-g*H0*rhoi,if(t<t1,-g*H0*rhoi*(1-(t-t0)/t1),0)), 0)


Remove this line.

naliboff · 2019-05-27T23:04:43Z

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_2D_loading.prm

+# Temperature boundary conditions
+subsection Boundary temperature model
+  set Fixed temperature boundary indicators = bottom, top, left, right
+  set List of model names = box


You can change this line to set List of model names = initial temperature, which would allow deleting the Box subsection below.

naliboff · 2019-05-27T23:07:57Z

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_3D_loading.prm

+# Material model
+subsection Material model
+
+   set Model name = simple


For simplicity, you could use the viscoelastic material model here and set the Elastic shear moduli to something like 1e50, which for all intensive purposes produces viscous behavior (e.g., negligible viscoelastic contribution).

naliboff · 2019-05-27T23:08:31Z

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_3D_loading.prm

+  end
+end
+
+# Number and name of compositional fields


Unless you use the Viscoelastic material model below, there is no need to specify these compositional fields.

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_3D_loading.prm

naliboff · 2019-05-27T23:09:42Z

source/simulator/helper_functions.cc

@@ -2040,7 +2040,7 @@ namespace aspect
    std::set<types::boundary_id> boundary_indicator_lists[6]
      = { boundary_velocity_manager.get_zero_boundary_velocity_indicators(),
          boundary_velocity_manager.get_tangential_boundary_velocity_indicators(),
-          parameters.free_surface_boundary_indicators,
+      //    parameters.free_surface_boundary_indicators,


Please delete this line.

naliboff · 2019-05-27T23:15:33Z

/rebuild

gassmoeller · 2019-05-31T14:28:57Z

@fionaclerc do you want to talk about how to address the comments?

naliboff · 2019-05-31T23:41:24Z

@fionaclerc - is this ready for another review?

naliboff

Thanks for updating and reorganizing all of this @fionaclerc! I think this is very close to being ready.

After you address the comments here, I would like to take one last look over to check for minor spelling errors, etc. Overall, the PR is in great shape and will be a really useful contribution to the code.

After merging, I think a cookbook on viscoelasticity is in order.

General comments and questions:

Does example_1.prm in the main folder still need to be included? If so, can you add some brief documentation (and perhaps rename)? If not, go ahead and delete this file.
Please add a changelog entry (doc/modules/changes) documenting the changes to the source code and benchmarks!

naliboff · 2019-06-16T16:57:27Z

benchmarks/free_surface_tractions/viscoelastic/README

+Changing the stress instantaneously as the load is applied/removed
+ leads to oscillations in the elastic response. This can be addressed
+ by setting a fixed elastic timestep larger than the numerical 
+ timestep and turning on stress averaging:


Do the oscillations still appear if you use a small CFL value and a Maximum time step that is equal to the elastic time step? In other problems, that scenario has produced reasonable results.

Using a smaller CFL value does not reduce the oscillations. Setting the maximum time step equal to the elastic time step does work, as in the parameter files “free_surface_VE_cylinder_*D_loading_fixed_elastic_dt.prm”. I am correcting/clarifying the README file.

naliboff · 2019-06-16T17:09:14Z

benchmarks/free_surface_tractions/viscoelastic/README

+ r0 is the load radius), for t>0. The load is fully removed by t=t1,
+ and the surface relaxes. This is done both in a 2-D and 3-D geometry
+ (by symmetry the load is centered on the left boundary or left/front
+ corner). The input files are:


Can you provide a bit more detail in this paragraph about the different loading/unloading scenarios (e.g., instantaneous verse linear)?

...free_surface_tractions/viscoelastic/free_surface_VE_cylinder_2D_loading_fixed_elastic_dt.prm

naliboff · 2019-06-17T13:46:43Z

benchmarks/free_surface_tractions/viscous/README

@@ -0,0 +1,35 @@
+Benchmark of infinite viscous half-space loaded/unloaded by an


Consider adding a note here and in the parameter files that the viscous benchmark is identical to the viscoelastic benchmarks, with the exception of the rheology (e.g., viscous verse viscoelastic).

benchmarks/free_surface_tractions/viscous/free_surface_viscous_cylinder_2D_loading.prm

fionaclerc · 2019-12-19T22:48:20Z

I have updated the README files and some comments to address the review. I also added a note in the README files about the mismatch between the analytical and numerical solutions at the far (right) boundary due to the free-slip condition.

tjhei · 2019-12-20T22:05:56Z

thanks for continuing the work on this! Sadly, there have been edits in free_surface.cc merged since you started your branch. You will need to rebase your changes to the current master. Note that the free_surface.cc got moved into a different folder, so you likely have to fix the conflicts manually. Let us know if you need help.

naliboff · 2020-01-30T20:27:38Z

@fionaclerc and I are working on fixing the rebase, will rebuild and test after that is done.

naliboff · 2020-02-04T00:32:42Z

@fionaclerc - This looks great, well done and thanks for sticking through all of the revisions! These benchmarks will be incredibly useful for a large group of users.

The last step is to do a final commit adding a changelog entry in the doc/modules/changes directory. A good example for this type of addition is: doc/modules/changes/20190529_euen.

Let me know if you have any questions about this and as soon the change log file is added, this will be ready to merge.

Co-Authored-By: naliboff <john.naliboff@gmail.com>

naliboff added the ready to test label May 27, 2019

fionaclerc force-pushed the free_surface_tractions branch from a717281 to 7b8f04c Compare May 27, 2019 22:37

naliboff reviewed May 27, 2019

View reviewed changes

fionaclerc force-pushed the free_surface_tractions branch from fbf3085 to 234d7cc Compare May 31, 2019 21:19

naliboff requested changes Jun 17, 2019

View reviewed changes

fionaclerc closed this Dec 19, 2019

fionaclerc reopened this Dec 19, 2019

fionaclerc force-pushed the free_surface_tractions branch from 5a5d294 to c3e20eb Compare January 30, 2020 18:17

add free surface traction benchmarks

3034e23

fionaclerc force-pushed the free_surface_tractions branch 2 times, most recently from 9dbcf15 to 9dc0599 Compare January 30, 2020 22:03

naliboff approved these changes Feb 4, 2020

View reviewed changes

address comments

1346f4b

Co-Authored-By: naliboff <john.naliboff@gmail.com>

fionaclerc force-pushed the free_surface_tractions branch from 9dc0599 to 1346f4b Compare February 5, 2020 22:59

naliboff merged commit ef542ec into geodynamics:master Feb 6, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Free surface tractions #3058

Free surface tractions #3058

fionaclerc commented May 27, 2019

naliboff left a comment

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff May 27, 2019

naliboff commented May 27, 2019

gassmoeller commented May 31, 2019

naliboff commented May 31, 2019

naliboff left a comment

naliboff Jun 16, 2019

fionaclerc Dec 19, 2019

naliboff Jun 16, 2019

fionaclerc Dec 19, 2019

naliboff Jun 17, 2019

fionaclerc commented Dec 19, 2019

tjhei commented Dec 20, 2019

naliboff commented Jan 30, 2020

naliboff commented Feb 4, 2020

		@@ -0,0 +1,7 @@
		set title "Depression of surface in response to loading (inst. from 0 to 1000 years), over viscoelastic medium"

		@@ -0,0 +1,11 @@
		set title "Depression of surface in response to loading (inst. from 0 to 1000 years), over viscoelastic medium"

		@@ -0,0 +1,35 @@
		Benchmark of infinite viscous half-space loaded/unloaded by an

Free surface tractions #3058

Free surface tractions #3058

Conversation

fionaclerc commented May 27, 2019

naliboff left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

naliboff commented May 27, 2019

gassmoeller commented May 31, 2019

naliboff commented May 31, 2019

naliboff left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fionaclerc commented Dec 19, 2019

tjhei commented Dec 20, 2019

naliboff commented Jan 30, 2020

naliboff commented Feb 4, 2020