Grid Project:   Wyoming CERC using Petrel / GSLIB

Grid Team Member:	Terry Miller
Grid Versions:	Subset NS, box, box2
Data Source:	PETREL Grid in GSLIB format
Delivered To:	John Jiao
Date Completed:	Dec 2013

---

Purpose of Project

Build a mesh from GSLIB format data representing a grid point distribution from PETREL. Data includes point xyz and material ID. Reformat into simple data field format for import into LaGriT. Subset and write FEHM files, evalute quality of resulting mesh. Note original point distribution is used and connected into Delaunay mesh. Note the material interfaces are not Voronoi conforming. Problems do occur do to the point distribution not being Delaunay where extreme aspect ratio (xy to z) and sloping interfaces cause connections across the interfaces.

Mesh Subset Box2 (Dec 2013)

LaGriT Files for creating subset box2.

Images for subset box2.

Mesh for FEHM:
number of nodes =       117450          
number of elements =    662382   

ATTRIBUTE         MIN               MAX         DIFFERENCE    LENGTH  
 xic      4.201000000E+05  4.299000000E+05   9.800000000E+03    117450  
 yic      4.307100000E+06  4.316900000E+06   9.800000000E+03    117450  
 zic     -2.208199951E+03  1.081999969E+02   2.316399948E+03    117450  

Material           1 has     12500 nodes. #nodes/nnodes is   0.106428265572     
Material           2 has     25000 nodes. #nodes/nnodes is   0.212856531143     
Material           3 has     12500 nodes. #nodes/nnodes is   0.106428265572     
Material           4 has      5000 nodes. #nodes/nnodes is   0.425713062286E-01 
Material           5 has      7500 nodes. #nodes/nnodes is   0.638569593430E-01 
Material           6 has      7500 nodes. #nodes/nnodes is   0.638569593430E-01 
Material           7 has     25000 nodes. #nodes/nnodes is   0.212856531143     
Material           8 has     15000 nodes. #nodes/nnodes is   0.127713918686     
Material           9 has      7450 nodes. #nodes/nnodes is   0.634312480688E-01 

Input Data:
-rw-r--r--. 1 tamiller sft 7863227 Nov 26 10:55 petrel_grid.dat

Data and FEHM files: /scratch/fwo/tam/wy_petrel_v01/subset_box2

Mesh for Subset Box (Sep 2013)

LaGriT Files to subset box 1.

Images for subset box 1.

Mesh for FEHM:

number of nodes =       17200        
number of elements =    94792    

ATTRIBUTE         MIN               MAX         DIFFERENCE    LENGTH  
 xic      2.410500000E+05  2.885500000E+05 4.750000000E+04     17200  
 yic      4.044700000E+06  4.092200000E+06 4.750000000E+04     17200  
 zic     -3.430279602E+03  1.523439774E+03 4.953719376E+03     17200  

Material           1 has      4000 nodes. #nodes/nnodes is   0.232558146119     
Material           2 has       800 nodes. #nodes/nnodes is   0.465116277337E-01 
Material           3 has       400 nodes. #nodes/nnodes is   0.232558138669E-01 
Material           4 has      1200 nodes. #nodes/nnodes is   0.697674453259E-01 
Material           5 has      6000 nodes. #nodes/nnodes is   0.348837196827     
Material           6 has       800 nodes. #nodes/nnodes is   0.465116277337E-01 
Material           7 has      4000 nodes. #nodes/nnodes is   0.232558146119     

Input Data:
-rw-r--r-- 1 tam sftus 84020735 Sep  3 12:32 test_grid_Gslib.dat

Data and GSLIB format work: /scratch/fwo/tam/wy_petrel_v01
Mesh and FEHM files: /scratch/fwo/tam/wy_petrel_v01/subset_box

Mesh for Subset NS Slice (Sep 2013)

LaGriT Files for subset NS slice.
Images for subset NS slice.

Mesh for FEHM:
This is a NS slice, 2 elements wide. 30x exaggeration in Z.
I had to cut the north end off where layers go up to the surface and can not connect delaunay.
This mesh has a few neg ccoef, but they are on the outside of the mesh so should be ok.

ATTRIBUTE         MIN               MAX         DIFFERENCE    LENGTH  
 xic      2.410500000E+05  2.885500000E+05 4.750000000E+04     17200  
 yic      4.044700000E+06  4.092200000E+06 4.750000000E+04     17200  
 zic     -3.430279602E+03  1.523439774E+03 4.953719376E+03     17200  

number of nodes =       27348        
number of elements =   110452    

Material           1 has      6360 nodes. #nodes/nnodes is   0.232558146119     
Material           2 has      1272 nodes. #nodes/nnodes is   0.465116277337E-01 
Material           3 has       636 nodes. #nodes/nnodes is   0.232558138669E-01 
Material           4 has      1908 nodes. #nodes/nnodes is   0.697674453259E-01 
Material           5 has      9540 nodes. #nodes/nnodes is   0.348837196827     
Material           6 has      1272 nodes. #nodes/nnodes is   0.465116277337E-01 
Material           7 has      6360 nodes. #nodes/nnodes is   0.232558146119     

Input Data:
-rw-r--r-- 1 tam sftus 84020735 Sep  3 12:32 test_grid_Gslib.dat

Regular grid 139x230x43 with null -99.00 x,y around irregular shape
Spacing is 2500x2500 in x,y with variable z (spacing near 60-125 vertical).

The outline boundary is irregular and can not be captured with a
simple connect mesh method. With more work quad surfaces can be created
from this point set and stack methods can be used.

Note that top and bottom layers are not flat.
Top layer has variable topology that pinches out some lower layers.
The points still exist but are squished together close to top surface
which causes neg ccoef in these areas.

The extreme aspect ratio of xy to z combined with sloping interfaces
causes difficulties with Delaunay connections such that nodes may be
removed. The material interfaces are not guaranteed where slope is
steep enough to "pop" elements up or down during connect.

created awk and lagrit scripts to correct point values, subset,
then connect into an FEHM mesh.


Data and GSLIB format work: /scratch/fwo/tam/wy_petrel_v01
Mesh and FEHM files: /scratch/fwo/tam/wy_petrel_v01/subset_NS

Grid Team Contacts