Add Absorbing Boundary #692 #703

cgeudeker · 2020-12-15T08:06:42Z

Describe the PR
Adding an absorbing boundary for earthquake simulations.

Related Issues/PRs
Part 2 of #692 (Most of the below information can be found here)

Additional context
Dashpots provide normal and shear traction along boundaries to dissipate the ground motion. Traction values are computed according to the following equations from Lysmer and Kuhlemeyer (1969).

Traction values are included in the governing equation for calculating acceleration. The direction input refers to the normal direction.

bool assign_absorbing_constraint(unsigned dir, double pwave_v, double swave_v);

codecov · 2020-12-15T08:23:45Z

Codecov Report

Merging #703 (7210c21) into develop (c9630b3) will increase coverage by 12.87%.
The diff coverage is 99.00%.

@@             Coverage Diff              @@
##           develop     #703       +/-   ##
============================================
+ Coverage    83.91%   96.78%   +12.87%     
============================================
  Files          169      130       -39     
  Lines        34176    25934     -8242     
============================================
- Hits         28676    25098     -3578     
+ Misses        5500      836     -4664

Impacted Files	Coverage Δ
include/node.h	`100.00% <ø> (ø)`
include/node_base.h	`100.00% <ø> (ø)`
include/node.tcc	`95.99% <96.15%> (ø)`
tests/node_test.cc	`99.79% <100.00%> (ø)`
...cbgeo/project/include/materials/linear_elastic.tcc
.../project/tests/materials/modified_cam_clay_test.cc
/home/cbgeo/project/tests/mesh_test_2d.cc
/home/cbgeo/project/include/io/io_mesh_ascii.tcc
/home/cbgeo/project/tests/nodal_properties_test.cc
/home/cbgeo/project/include/node.tcc
... and 293 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c9630b3...7210c21. Read the comment docs.

jgiven100 · 2020-12-15T08:36:55Z

@kks32 Maybe this would be a good PR to start pushing to a new earthquake branch?

kks32

@cgeudeker Thanks for your first PR. Nice work!

If the boundary nodes are only supported by the Lysmer-Kuhlmeyer dashpots, it would result in creep because of the stresses from the soil body and will continue to deform. I would suggest adding a spring + dashpot to avoid this effect similar to a Kelvin-Voigt type boundary. Thanks for the unit test. However, I suggest a benchmark problem, create a linear elastic solid block supported on just the dashpots on one end and free on the other. At the free surface, apply an instantaneous vertical load and see what happens to the displacements at the dashpot end, you'll notice creep. This issue needs to be fixed to be able to use a good absorbing boundary.

kks32 · 2020-12-16T04:45:54Z

include/node.h

@@ -299,6 +308,10 @@ class Node : public NodeBase<Tdim> {
  Eigen::Matrix<double, Tdim, Tnphases> acceleration_;
  //! Velocity constraints
  std::map<unsigned, double> velocity_constraints_;
+  //! Absorbing Constraints
+  std::tuple<unsigned, double, double> absorbing_constraints_;
+  //! Absorbing Traction


This variable is not needed

kks32 · 2020-12-16T04:46:12Z

include/node.h

@@ -201,6 +201,15 @@ class Node : public NodeBase<Tdim> {
  //! Apply velocity constraints
  void apply_velocity_constraints() override;

+  //! Assign absorbing constraint
+  //! \param[in] pwave_v P-wave velocity


p-wave and s-wave should be material properties too, not nodal properties.

Hi, so I've been able to understand and make most of your other suggestions barring this one. How would I be able to access these values from the material properties? I had thought it would be through property_handle_ but I'm currently getting an error about segmentation violation when I run ./mpmtest and it calls the function where I do this. Perhaps this is the right method but I need to initialize the properties somewhere. Thanks for your help

kks32 · 2020-12-16T04:46:51Z

include/node.tcc

+          static_cast<unsigned>(dir), static_cast<double>(pwave_v),
+          static_cast<double>(swave_v));
+    } else
+      throw std::runtime_error("Constraint direction is out of bounds");


Please be specific on the constraint

kks32 · 2020-12-16T04:48:07Z

include/node.tcc

@@ -346,6 +350,62 @@ void mpm::Node<Tdim, Tdof, Tnphases>::apply_velocity_constraints() {
  }
 }

+//! Assign absorbing constraints
+template <unsigned Tdim, unsigned Tdof, unsigned Tnphases>
+bool mpm::Node<Tdim, Tdof, Tnphases>::assign_absorbing_constraint(


Let's change the function to void

kks32 · 2020-12-16T04:48:17Z

include/node.tcc

+    unsigned dir, double pwave_v, double swave_v) {
+  bool status = true;
+  try {
+    if (dir < Tdim) {


Check this condition as assert

kks32 · 2020-12-16T04:51:38Z

include/node.tcc

+
+    // Calculate Traction Forces
+    if (dir == dir_n) {
+      traction = -velocity * mass_density * pwave_v;


I'd suggest adding a dimensionless damping co-efficient.

kks32 · 2020-12-16T04:58:51Z

include/node.tcc

+  // Create Traction Value
+  double traction;
+
+  for (unsigned dir = 0; dir < Tdim; ++dir) {


Instead of a for loop create an Eigen::Matrix<double, Tdim, 1> wave_velocity = swave_velocity and set the normal direction to pwave_velocity then use a cwiseProduct to do the multiplication so you don't need a for loop.

You probably need absorbing boundaries on all phases

include/node.tcc

kks32 · 2020-12-16T05:01:45Z

tests/node_test.cc

+        node->apply_absorbing_constraint();
+
+        // Define traction variables
+        double traction_normal;


No need to predefine variables

kks32 · 2020-12-16T05:02:18Z

tests/node_test.cc

@@ -683,6 +683,93 @@ TEST_CASE("Node is checked for 2D case", "[node][2D]") {
      // TODO Assert: REQUIRE_NOTHROW(node->update_external_force(false, 1,
      // force_bad));

+      SECTION("Check absorbing constraint") {
+        // Set external force to 0


Could you use an already existing function, so you don't have to redefine and check all the momenta and forces?

ezrayst · 2020-12-16T07:35:51Z

Hi @kks32, thanks so much for your thoughtful comments! Really appreciate it. We are making the benchmark as well, but do you have a reference to a problem? I am struggling to make a problem because the answer is not quite analytical. In addition, does it mean you do not want the unit test (I hope you still want it :P). Thanks once again!

kks32 · 2020-12-16T12:21:54Z

Hi @ezrayst What benchmark problem are you creating? Could you please elaborate.
It would be good to see the results of an elastic bar subjected to an instantaneous load, like you mentioned in #692

I was thinking a linear elastic rectangular block which is 1 m thick and 1 m long supported on dashpots. Apply an instantaneous load and see how the displacement gets damped. You can compare the results against (a) fixed displacement boundary (in this case the stress will get doubled, once the wave reflects), (b) dashpot-alone, I expect this to creep, and (c) spring-dashpot system.

kks32 · 2021-03-02T21:25:58Z

Closing this PR as #703 supersedes this PR.

cgeudeker added 7 commits November 9, 2020 19:30

initiate absorbing constraint

32fdba3

initiate absorbing constraint

2360285

initiate bounding constraint update

10c6415

Test Case Attempt

cee7b4e

Working Test Case

04e38ef

3D test case

7e7ebfc

3D test case, clang-format

7210c21

jgiven100 requested review from bodhinandach, kks32, thiagordonho, jgiven100 and ezrayst December 15, 2020 08:33

jgiven100 added Priority: Medium Status: Review needed Type: Feature request labels Dec 15, 2020

kks32 requested changes Dec 16, 2020

View reviewed changes

kks32 self-assigned this Dec 16, 2020

kks32 closed this Mar 2, 2021

kks32 added Type: Wontfix and removed Status: Review needed Type: Feature request Type: Wontfix Priority: Medium labels Mar 2, 2021

kks32 removed the request for review from ezrayst March 2, 2021 21:26

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add Absorbing Boundary #692 #703

Add Absorbing Boundary #692 #703

cgeudeker commented Dec 15, 2020 •

edited

Loading

codecov bot commented Dec 15, 2020

jgiven100 commented Dec 15, 2020

kks32 left a comment

kks32 Dec 16, 2020

kks32 Dec 16, 2020

cgeudeker Dec 29, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

kks32 Dec 16, 2020

ezrayst commented Dec 16, 2020

kks32 commented Dec 16, 2020 •

edited

Loading

kks32 commented Mar 2, 2021

Add Absorbing Boundary #692 #703

Add Absorbing Boundary #692 #703

Conversation

cgeudeker commented Dec 15, 2020 • edited Loading

codecov bot commented Dec 15, 2020

Codecov Report

jgiven100 commented Dec 15, 2020

kks32 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

ezrayst commented Dec 16, 2020

kks32 commented Dec 16, 2020 • edited Loading

kks32 commented Mar 2, 2021

cgeudeker commented Dec 15, 2020 •

edited

Loading

kks32 commented Dec 16, 2020 •

edited

Loading