[GeoMechanicsApplication] Added a 3D dynamic test case #12656

mnabideltares · 2024-09-02T06:52:02Z

📝 Description
A 3D dynamic test case is added to the integration tests.

🆕 Changelog

Added a test case
Added documentation

WPK4FEM

Dear Mohammed,

Thank you for the elaborate work on testing here. Even if we are not miraculously close to the semi-analytical formulation I think these tests are valuable.

Regards, Wijtze Pieter

WPK4FEM · 2024-09-02T08:42:01Z

...Application/tests/test_dynamic/documentation_data/test_constant_point_load_3d_conditions.svg

The direction of the force seems really vertical, while the block looks to be at an angle with the vertical. Force arrow should be aligned with the edge of the block.

WPK4FEM · 2024-09-02T08:44:23Z

...eoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/ProjectParameters.json

+        "body_domain_sub_model_part_list"    : ["soil_block"]
+    },
+    "output_processes": {
+            "gid_output" : [{


I wonder if this output should be asked for when only the next output process gives the things that are actively checked.

WPK4FEM · 2024-09-02T08:51:10Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+- $\nu$		= Poisson's ratio  $\mathrm{[-]}$
+- $\rho$	= Mass density $\mathrm{[kg/m^3]}$
+
+In this test case, $E = 91800 \mathrm{Pa}$ and $\nu = 0.25$. This leads to shear and compression wave velocities of $6 \mathrm{m/s}$ and $10.39 \mathrm{m/s}$, respectively. Hence, in order to avoid the effects of reflecting waves from the boundaries, considering a domain of $10 \mathrm{m} × 10 \mathrm{m} × 10 \mathrm{m}$ with a simulation time of 1 second is sufficient.


With 10.39 m/s and a distance of 10m, the wave starts to reflect at 10/10.39 s which is less than the computed 1 s.

You are right, there may happen a slight reflection at the end, but as it is just a fraction of time, we don't observe it. The reflection may affect the very end points of the figure.
Tried 20m domain, but the results did not change much around 1s

WPK4FEM · 2024-09-02T08:54:12Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+## Case Specification
+In this test case, a three-dimentional soil block with dimensions of 10 m x 10 m x 10 m is considered. This is a quart of the full domain, as this case is axisymmetric. The pressure is fixed at 0 Pa throughout the entire simulation to eliminate the effect of water pressure. This is done to allow comparison of our results with published semi-analytical solutions. The bottom side of the domain is fixed, while the vertical sides are allowed to move only in the tangential directions. A sudden force of -250 N is then applied in the vertical direction at the top surface of the block. The simulation spans 1 second. This test is conducted on a mesh with triangular elements of type 3D4N. The deformation at the surface is then compared with semi-analytical results.
+
+The geometry and boundary conditions are shown below:


the Dirichlet boundary conditions are not in the picture ( and would be a lot of work to show clearly ). Therefore I propose to only mention the geometry and loading conditions. Not the boundary conditions.

WPK4FEM · 2024-09-02T08:59:46Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+- $\nu$		= Poisson's ratio  $\mathrm{[-]}$
+- $\rho$	= Mass density $\mathrm{[kg/m^3]}$
+
+In this test case, $E = 91800 \mathrm{Pa}$ and $\nu = 0.25$. This leads to shear and compression wave velocities of $6 \mathrm{m/s}$ and $10.39 \mathrm{m/s}$, respectively. Hence, in order to avoid the effects of reflecting waves from the boundaries, considering a domain of $10 \mathrm{m} × 10 \mathrm{m} × 10 \mathrm{m}$ with a simulation time of 1 second is sufficient.


91800 seems a strange number in [Pa] I would expect something in the order of 80 MPa.

I do agree that a such number does not make much sense practically. We used it just to compare.

WPK4FEM · 2024-09-02T10:57:25Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+**Source files:** [Dynamic solution in 3D](https://github.com/KratosMultiphysics/Kratos/tree/master/applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d)
+
+## Case Specification
+In this test case, a three-dimentional soil block with dimensions of 10 m x 10 m x 10 m is considered. This is a quart of the full domain, as this case is axisymmetric. The pressure is fixed at 0 Pa throughout the entire simulation to eliminate the effect of water pressure. This is done to allow comparison of our results with published semi-analytical solutions. The bottom side of the domain is fixed, while the vertical sides are allowed to move only in the tangential directions. A sudden force of -250 N is then applied in the vertical direction at the top surface of the block. The simulation spans 1 second. This test is conducted on a mesh with triangular elements of type 3D4N. The deformation at the surface is then compared with semi-analytical results.


triangular elements are 2D. Here tetrahedral elements are used.

WPK4FEM · 2024-09-03T07:07:14Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+$$ M = \frac{E \left(1 - \nu \right)}{\left( 1 + \nu \right) \left(1 - 2 \nu \right)} $$
+
+- $G$		= Shear modulus $\mathrm{[Pa]}$
+- $M$		= P-wave modulus  $\mathrm{[Pa]}$


I'm guessing that the usual name for this is bulk modulus ( mostly denoted as K ) or is it one of the 2 constants of Lame that usually have Greek symbols lambda and mu? If we are following the notation used by Verruit.we shouldn't change it.

WPK4FEM · 2024-09-03T08:53:55Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+
+In the semi-analytical solution, there is a singulair point around $\tau = 1.1$. However, this behavior, which is caused by the arrival of the Rayleigh wave, is captured by a peak in the numerical solution. 
+
+Note: To avoid numerical oscillations, the force is gradually applied over a time span of 0.1 seconds.


On a total time of 1 s the 0.1 for the increase seems rather long. That may have an influence on what you describe as shift in time for the peak behavior. Is the same ramping time used in the 2D computation?

In the case of < 0.1, the results even shift more to the left side. for the case of > 0.1, the peak becomes smoother but not much shifts to the right

rfaasse

Thank you for adding this test case with a clear readme! I left some suggestions/questions, let me know what you think!

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

...eoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/ProjectParameters.json

rfaasse · 2024-09-03T09:13:41Z

...eoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/ProjectParameters.json

+                    "output_file_name"             : "json_output.json",
+                    "output_variables"             : ["DISPLACEMENT_Y"],
+                    "gauss_points_output_variables": [],
+                    "time_frequency"               : 0.0099999999


This seems very specific, what is the reason it can't be 0.01?

In the case of 0.01, it somehow skips some of the outputs (don't know the reason). In order to be certain for a correct output, we make it slightly less than 0.01.

...Application/tests/test_dynamic/documentation_data/test_constant_point_load_3d_conditions.svg

...icsApplication/tests/test_dynamic/documentation_data/test_constant_point_load_3d_results.png

mnabideltares

@WPK4FEM and @rfaasse
Thank you for review. These are my changes and comments

mnabideltares · 2024-09-05T09:14:55Z

...eoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/ProjectParameters.json

+                    "output_file_name"             : "json_output.json",
+                    "output_variables"             : ["DISPLACEMENT_Y"],
+                    "gauss_points_output_variables": [],
+                    "time_frequency"               : 0.0099999999


In the case of 0.01, it somehow skips some of the outputs (don't know the reason). In order to be certain for a correct output, we make it slightly less than 0.01.

...eoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/ProjectParameters.json

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

mnabideltares · 2024-09-05T09:23:32Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+- $\nu$		= Poisson's ratio  $\mathrm{[-]}$
+- $\rho$	= Mass density $\mathrm{[kg/m^3]}$
+
+In this test case, $E = 91800 \mathrm{Pa}$ and $\nu = 0.25$. This leads to shear and compression wave velocities of $6 \mathrm{m/s}$ and $10.39 \mathrm{m/s}$, respectively. Hence, in order to avoid the effects of reflecting waves from the boundaries, considering a domain of $10 \mathrm{m} × 10 \mathrm{m} × 10 \mathrm{m}$ with a simulation time of 1 second is sufficient.


You are right, there may happen a slight reflection at the end, but as it is just a fraction of time, we don't observe it. The reflection may affect the very end points of the figure.
Tried 20m domain, but the results did not change much around 1s

mnabideltares · 2024-09-05T09:24:43Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+- $\nu$		= Poisson's ratio  $\mathrm{[-]}$
+- $\rho$	= Mass density $\mathrm{[kg/m^3]}$
+
+In this test case, $E = 91800 \mathrm{Pa}$ and $\nu = 0.25$. This leads to shear and compression wave velocities of $6 \mathrm{m/s}$ and $10.39 \mathrm{m/s}$, respectively. Hence, in order to avoid the effects of reflecting waves from the boundaries, considering a domain of $10 \mathrm{m} × 10 \mathrm{m} × 10 \mathrm{m}$ with a simulation time of 1 second is sufficient.


I do agree that a such number does not make much sense practically. We used it just to compare.

mnabideltares · 2024-09-05T09:26:14Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+
+In the semi-analytical solution, there is a singulair point around $\tau = 1.1$. However, this behavior, which is caused by the arrival of the Rayleigh wave, is captured by a peak in the numerical solution. 
+
+Note: To avoid numerical oscillations, the force is gradually applied over a time span of 0.1 seconds.


In the case of < 0.1, the results even shift more to the left side. for the case of > 0.1, the peak becomes smoother but not much shifts to the right

mnabideltares · 2024-09-05T09:28:19Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+## Case Specification
+In this test case, a three-dimentional soil block with dimensions of 10 m x 10 m x 10 m is considered. This is a quart of the full domain, as this case is axisymmetric. The pressure is fixed at 0 Pa throughout the entire simulation to eliminate the effect of water pressure. This is done to allow comparison of our results with published semi-analytical solutions. The bottom side of the domain is fixed, while the vertical sides are allowed to move only in the tangential directions. A sudden force of -250 N is then applied in the vertical direction at the top surface of the block. The simulation spans 1 second. This test is conducted on a mesh with triangular elements of type 3D4N. The deformation at the surface is then compared with semi-analytical results.
+
+The geometry and boundary conditions are shown below:


rfaasse

Thank you for incorporating the changes and diving into this problem in the first place! From my perspective this is good to go!

WPK4FEM

Thank you Mohamed,
Almost there. I'm afraid that I caused some confusion about bulk modulus and p wave modulus.

Did we try the computation with both Rayleigh parameters 0? Verruijt's equations have no damping and I'm not sure which frequencies we are damping here. Adding damping usually smears out the peak behavior.

Wijtze Pieter

WPK4FEM · 2024-09-06T08:50:43Z

...Application/tests/test_dynamic/documentation_data/test_constant_point_load_3d_conditions.svg

Now it is consistent, thank you.

WPK4FEM · 2024-09-06T08:56:14Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+$$ M = \frac{E \left(1 - \nu \right)}{\left( 1 + \nu \right) \left(1 - 2 \nu \right)} $$
+
+- $G$		= Shear modulus $\mathrm{[Pa]}$
+- $M$		= Bulk modulus  $\mathrm{[Pa]}$


M is the P wave modulus. See e.g. https://en.wikipedia.org/wiki/Lam%C3%A9_parameters that is different from the bulk modulus usually denoted with K

WPK4FEM · 2024-09-06T08:57:38Z

applications/GeoMechanicsApplication/tests/test_dynamic/test_constant_point_load_3d/README.md

+
+$$ c_p = \sqrt{\frac{M}{\rho}} $$
+
+where $G$ and $M$ are shear modulus and bulk modulus, respectively. They are defined as:


bulk modulus -> P wave modulus.

WPK4FEM

Thank you for adding the test and documentation.

Added a 3D dynamic test case

bab9f68

mnabideltares linked an issue Sep 2, 2024 that may be closed by this pull request

[GeoMechanicsApplication] The 3D dynamic test case results do not match the analytical solution #12653

Closed

mnabideltares self-assigned this Sep 2, 2024

mnabideltares requested review from WPK4FEM, avdg81 and rfaasse September 2, 2024 06:53

README modified

5b499e9

WPK4FEM reviewed Sep 3, 2024

View reviewed changes

rfaasse reviewed Sep 3, 2024

View reviewed changes

mnabideltares added 2 commits September 5, 2024 11:09

Changes based on review comments

396eec2

fixed figure

332cb1e

mnabideltares commented Sep 5, 2024

View reviewed changes

mnabideltares requested review from WPK4FEM and rfaasse September 6, 2024 07:06

Readme is modified

ef03f3b

rfaasse previously approved these changes Sep 6, 2024

View reviewed changes

WPK4FEM reviewed Sep 6, 2024

View reviewed changes

fixes in Readme

5cb294f

mnabideltares dismissed rfaasse’s stale review via 5cb294f September 6, 2024 09:04

WPK4FEM approved these changes Sep 6, 2024

View reviewed changes

rfaasse approved these changes Sep 6, 2024

View reviewed changes

rfaasse merged commit 4ef6e05 into master Sep 6, 2024
9 of 11 checks passed

rfaasse deleted the geo/12653-3d-dynamic-test branch September 6, 2024 15:22

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[GeoMechanicsApplication] Added a 3D dynamic test case #12656

[GeoMechanicsApplication] Added a 3D dynamic test case #12656

mnabideltares commented Sep 2, 2024 •

edited by rfaasse

Loading

WPK4FEM left a comment

WPK4FEM Sep 2, 2024

WPK4FEM Sep 2, 2024

mnabideltares Sep 5, 2024

WPK4FEM Sep 2, 2024

mnabideltares Sep 5, 2024

WPK4FEM Sep 2, 2024

mnabideltares Sep 5, 2024

WPK4FEM Sep 2, 2024

mnabideltares Sep 5, 2024

WPK4FEM Sep 2, 2024

WPK4FEM Sep 3, 2024

WPK4FEM Sep 3, 2024

mnabideltares Sep 5, 2024

rfaasse left a comment

rfaasse Sep 3, 2024

mnabideltares Sep 5, 2024

mnabideltares left a comment

mnabideltares Sep 5, 2024

mnabideltares Sep 5, 2024

mnabideltares Sep 5, 2024

mnabideltares Sep 5, 2024

mnabideltares Sep 5, 2024

rfaasse left a comment

WPK4FEM left a comment

WPK4FEM Sep 6, 2024

WPK4FEM Sep 6, 2024

WPK4FEM Sep 6, 2024

WPK4FEM left a comment


		In the semi-analytical solution, there is a singulair point around $\tau = 1.1$. However, this behavior, which is caused by the arrival of the Rayleigh wave, is captured by a peak in the numerical solution.

		Note: To avoid numerical oscillations, the force is gradually applied over a time span of 0.1 seconds.


		$$ c_p = \sqrt{\frac{M}{\rho}} $$

		where $G$ and $M$ are shear modulus and bulk modulus, respectively. They are defined as:

[GeoMechanicsApplication] Added a 3D dynamic test case #12656

[GeoMechanicsApplication] Added a 3D dynamic test case #12656

Conversation

mnabideltares commented Sep 2, 2024 • edited by rfaasse Loading

WPK4FEM 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

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

rfaasse left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mnabideltares 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

rfaasse left a comment

Choose a reason for hiding this comment

WPK4FEM 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

WPK4FEM left a comment

Choose a reason for hiding this comment

mnabideltares commented Sep 2, 2024 •

edited by rfaasse

Loading