more revisions

wpbonelli · Dec 11, 2024 · 297b082 · 297b082
1 parent fd98950
commit 297b082
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 19 deletions.
diff --git a/src/Solution/ParticleTracker/MethodDis.f90 b/src/Solution/ParticleTracker/MethodDis.f90
@@ -192,18 +192,22 @@ subroutine load_particle(this, cell, particle)
 
     select type (dis => this%fmi%dis)
     type is (DisType)
-      ! compute and set reduced/user node numbers and layer
+      ! compute reduced/user node numbers and layer
       inface = particle%iboundary(2)
       inbr = cell%defn%facenbr(inface)
       idiag = this%fmi%dis%con%ia(cell%defn%icell)
       ipos = idiag + inbr
       ic = dis%con%ja(ipos)
+      icu = dis%get_nodeuser(ic)
+      call get_ijk(icu, dis%nrow, dis%ncol, dis%nlay, &
+                   irow, icol, ilay)
 
-      ! if returning to immediately previous cell
-      ! through its bottom, we've entered a cycle
-      ! as can occur e.g. in the bottom of wells.
-      ! terminate in the previous cell.
-      if (ic == particle%icp .and. inface == 7) then
+      ! if returning to a cell through the bottom
+      ! face after previously leaving it through  
+      ! that same face, we've entered a cycle
+      ! as can occur e.g. in wells. terminate
+      ! in the previous cell.
+      if (ic == particle%icp .and. inface == 7 .and. ilay < particle%ilay) then
         particle%advancing = .false.
         particle%idomain(2) = particle%icp
         particle%istatus = 2
@@ -215,9 +219,7 @@ subroutine load_particle(this, cell, particle)
         particle%izp = particle%izone
       end if
 
-      icu = dis%get_nodeuser(ic)
-      call get_ijk(icu, dis%nrow, dis%ncol, dis%nlay, &
-                   irow, icol, ilay)
+      ! update node numbers and layer
       particle%idomain(2) = ic
       particle%icu = icu
       particle%ilay = ilay

diff --git a/src/Solution/ParticleTracker/MethodDisv.f90 b/src/Solution/ParticleTracker/MethodDisv.f90
@@ -155,12 +155,15 @@ subroutine load_particle(this, cell, particle)
       inbr = cell%defn%facenbr(inface)
       ipos = idiag + inbr
       ic = dis%con%ja(ipos)
+      icu = dis%get_nodeuser(ic)
+      call get_jk(icu, dis%ncpl, dis%nlay, icpl, ilay)
 
-      ! if returning to immediately previous cell
-      ! through its bottom, we've entered a cycle
-      ! as can occur e.g. in the bottom of wells.
-      ! terminate in the previous cell.
-      if (ic == particle%icp .and. inface == 7) then
+      ! if returning to a cell through the bottom
+      ! face after previously leaving it through  
+      ! that same face, we've entered a cycle
+      ! as can occur e.g. in wells. terminate
+      ! in the previous cell.
+      if (ic == particle%icp .and. inface == 7 .and. ilay < particle%ilay) then
         particle%advancing = .false.
         particle%idomain(2) = particle%icp
         particle%istatus = 2
@@ -172,10 +175,6 @@ subroutine load_particle(this, cell, particle)
         particle%izp = particle%izone
       end if
 
-      icu = dis%get_nodeuser(ic)
-
-      call get_jk(icu, dis%ncpl, dis%nlay, icpl, ilay)
-
       particle%idomain(2) = ic
       particle%icu = icu
       particle%ilay = ilay

diff --git a/src/Solution/ParticleTracker/vertical.md b/src/Solution/ParticleTracker/vertical.md
@@ -130,4 +130,4 @@ flowchart LR
 
 In MF6.5, behavior was as described by `DROP`, with one major exception: lack of an exit face (i.e. any face with outgoing flow) took precedence over cell saturation; a particle finding itself in a dry cell with no outgoing flow would previously terminate, where if `DROP` is selected (or a dry tracking method unspecified) the pass-to-bottom method will now be applied instead.
 
-With this change, it also becomes necessary to prohibit particle backtracking (i.e. re-entering the previous cell) within the same time step, in order to avoid the possibility of infinite loops. For instance, a particle might otherwise be passed endlessly between e.g. the bottom face of a cell containing a pumping well and the top face of the cell below. 
+With this change, it also becomes necessary to prohibit backtracking between vertically adjacent pairs of cells within the same time step, in order to avoid the possibility of infinite loops &mdash; a particle might otherwise be passed endlessly between e.g. the bottom face of a cell containing a pumping well and the top face of the cell below. Note that this limitation applies only to vertically adjacent cells, and only to the immediately previous cell &mdash; a particle may re-enter a cell after entering a third cell.
Original file line number	Diff line number	Diff line change
Expand Up		@@ -130,4 +130,4 @@ flowchart LR

		In MF6.5, behavior was as described by `DROP`, with one major exception: lack of an exit face (i.e. any face with outgoing flow) took precedence over cell saturation; a particle finding itself in a dry cell with no outgoing flow would previously terminate, where if `DROP` is selected (or a dry tracking method unspecified) the pass-to-bottom method will now be applied instead.

		With this change, it also becomes necessary to prohibit particle backtracking (i.e. re-entering the previous cell) within the same time step, in order to avoid the possibility of infinite loops. For instance, a particle might otherwise be passed endlessly between e.g. the bottom face of a cell containing a pumping well and the top face of the cell below.
		With this change, it also becomes necessary to prohibit backtracking between vertically adjacent pairs of cells within the same time step, in order to avoid the possibility of infinite loops — a particle might otherwise be passed endlessly between e.g. the bottom face of a cell containing a pumping well and the top face of the cell below. Note that this limitation applies only to vertically adjacent cells, and only to the immediately previous cell — a particle may re-enter a cell after entering a third cell.