|
Begin this tutorial by establishing a new project called Gridding demo. Start with one Generic folder.
The grid window is important when gridding. In this tutorial we select a grid window that has realistic coordinates i x, y and z:
xmin=450000 xmax=460000 ymin=6450000 ymax=6460000 zmin=1000 zmax=4000
This grid window is for convenience established using the command: win demo. Do that command in the command shell.
Go to the Project Settings on the project tool bar and under Data and click Use current. Likewise, under Graphics select white as background color and check in the two boxes Erase graphics when loading project and Apply settings when loading project.
We use these options in our tutorial as they are convenient when loading the project.
At this step the task is to generate a few random points with the grid window. These points would then represent surface data or well data. In the demo case 11 random points were generated using the shell command: mak ran 11. Create a folder called XYZ data in the project and apply New -> Workspace Data -> Active. Rename the dataset to well data.
Check that the random points have a reasonable distribution across the graphical window. If not repeat the generation.
In order to be more realistic we want to generate a boundary for the reservoir surface. This is done using the digitizer under Tools -> Quick Digitizing.
Digitize a closed boundary going fairly close to the edges and including all points to get a good representation. In the default case when ending the digitizing the digitized line is saved in workspace as digitized_line.
Establish a new folder called Boundary and put the digitized line into that folder using New -> Workspace Data -> digitized_line. Rename this dataset to border line.
For convenience establish a folder called Images to capture screenshots along the development of the project and the tutorial. For later use a folder for Grids and a folder for Fault lines are created.
It all looks like this:
Dispay of project and graphics when staring the tutorial
|
The color legend was displayed using the shell command cco s 1000 500 4000 rev col bla. For convenience this color legend command was placed in an item script for the grid surface and called Color legend 500 inc.
Establishing small scripts are effective in real production cases and is part of this tutorial. The way it is done is as follows:
|
The gridding was repeated with two other algorithms: Moving average and Surfit. Read more about gridding algorithms in the user guide.
The three surfaces were collected in the Grids folder and renamed to: gridmav gridpara gridsurfit. A command script located on the Grids folder and called Grids in viewport was created to display them in a connected viewport presentation. The script is simple and explained below.
The script is executed on the Grids folder. That means it will execute the contents within the Prestep just once. The contents within Main is executed on every dataset in the folder. The contents in Poststep is executed finally just once.
Here is the contents in Prestep. An initial variable is allocated and the viewport arrangement is achieved. This is done before any execution on the datasets.
# This is the contents in Prestep variable init 0 ;# initialize an init variable vie def ;# reset viewports to default vie 3 1 1 ;# allocate 3 x 1 viewports |
Here is the contents in Main. It selects the viewport number and performs the mapping and other graphical stuff. Beware that the well data is copied and pasted to workspace where it get the name welldata having no blanks in the name. Then it is easy to apply the well points in the script.
# This is the contents in Main incr init ;# increment init variable spe amb 0.3 ;# specify overall ambient light 0.3 vie $init con ;# go to viewport number in variable and connect to previous spe bgc ran lig ;# specify random background light with a light value col bro ;# select brown color dra win ;# draw graphical window map rng 1000 4000 ;# map using range values col blu ;# select blue color bol 2 ;# draw border of the grid with linewidth 2 mlo welldata ;# move welldata to active which has been placed in workspace col bla ;# select black color poi ;# display the points if {$init == 1} { ;# chech on init variable tx2 lle col bla txt "Moving average" ;# the first is moving average cco s 1000 500 4000 rev col bla ;# display color legend } elseif {$init == 2} { ;# the second is parabolic tx2 lle col bla txt "Parabolic" ;# display text in 2d mode } else { ;# last is surfit tx2 lle col bla txt "Surfit" ;# display text in 2d mode mak com ;# make the compass and display } |
No statements are necessary in Poststep.
The user may copy these statements and put them correctly into the script and reproduce the display.
The viewport display is taken out as a screenshot and saved. The result shows that the algorithms solve the interpolation task slightly different. Such a difference is more obvious for few points where the degree of freedom is greater.
Viewport presentation of gridding with different algorithms
To go back to one viewport use the viewport icon on the toolbar or type vie def in the shell. To set white background go the the project toolbar and activate the project settings or type spe bgc whi in the shell.
|
To generate a grid surface using faults as extra input one has to apply the menu Gridding, points, lines and faults found under Gridding.
The menu settings should be fairly straight forward. The two menu displays shows the Grid Window page and the Fault Input page settings. Otherwise it is default settings with the exception that smoothing is turned off on the Result page. The result grid will by default be saved in workspace resultgrid so we pick it up from there after the gridding.
Gridding is performed using the single stick fault lines as blocking lines so that the evaluation of any grid node is only affected by input points that are within non blocking connection with the node.
After gridding transfer the resultgrid in workspace over to the Grids folder and rename it to resultgrid_faults.
The result grid with faults shows a good sharp edge where the stick fault lines were placed. The algorithm generates an upper and lower fault trace that is saved in workspace closeGenFaults. That dataset is transferred to the Fault lines folder and displayed in black in the display.
The mapping was done with the command Additional Display -> General display applying a combination of Color bands and Line contours. Color bands is applied with increment 500 (and no checkbox for Draw contour lines) while Line contours applies an increment of 100.
The right display shows also the input points and tells that they are honoring the derived surface.
|
A new exercise was performed which is displayed in the viewport arrangement below. The result was not good enough right away so Two step gridding had to be applied.
|
Three stages of grid exercise: 1: Input data. 2. Grid result. 3. Improved grid result using Two step gridding
This tutorial is based on creating input data along with the tutorial steps. This goes also for gridding of line data.
|
|
The digitized seismic 2D lines is so far not quite suitable to be used. It requires more points; i.e. seismic 2D lines have a shot point spacing that we will introduced. A fairly adequate spacing is 100 meter. This is achieved quite simple using the following procedure:
|
The faulted grid surface generated as the first grid from surface data and two stick faults will be used as a master surface to be reproduced. Assume this grid is called resultgrid_faults. Do Copy and paste it into workspace.
|
|
A critical parameter in line data gridding is the grid cell size value in Use two gridding steps. If it were blank the algorithm used 200 (= 4 * 50, where 50 is the result grid increment). Some other values were tried and it turned out that 200 gave the best result.
Two steps gridding works by letting the first step generate some surface points that are appended with the input lines to represent input data in the final gridding process. That dataset is present in workspace as input_data and can be inspected and displayed.
Below is a viewport representation of two result grids with different 1. step grid cell size and the original surface. A quick view valuates the first grid as the best.
Result grid for two different cases of 1. step grid cell size and original surface
|
All the issues in the exercise are cases that can occur in realistic situations. Gridding is dependent of input data and the parameters used in the algorithm. Good knowledge about gridding will benefit a good grid result.