Determining Down-hole Tool Orientation by PCA

The geometry will be complete once the down-hole tool horizontal component orientation has been established as described in Michaels (12). We begin by running the genbhod program which builds scripts that control the hodogram analysis by program bhod. The following is the dialog between genbhod and the user (boxed). The program is run from inside the directory with the newly created L*.seg files (created from Bison format):

 |---------------------------------|
 | Copyright (C) 2017  P. Michaels |
 |       All rights reserved       |
 |see GNU General Public License   |
 |---------------------------------|

 WARNING: !!
 See Source Code, genbhod.f, or BSU documentation
 (man pages and BSU user Guide)
 before you use this program. It is hardwired for
 a specific type of acquisition.

 enter 1char_ALPHA PREFIX

  enter FIRST FILE NUMBER (<=3digits)
  for which source polarization is 270 deg.

  enter LAST FILE NUMBER (<=3digits)
  for which source polarization is 270 deg.

146

  enter UP/DOWN SWITCH
    -1= 90 Azimuth File Number 1 LESS than 270 Az
    +1= 90 Azimuth File Number 1 MORE than 270 Az

-1
enter azimuth of bowspring(R-comp)
240

  OUTPUT====> Downhole: gobhodo
  OUTPUT====> Reference: gobhodoR
  OUTPUT====> Downhole: gorunbhod
  OUTPUT====> Reference: gorunbhodR
  ----------------------------------------
 REMEMBER to change permissions on the
 above files to execute.
  ----------------------------------------
  IF examining the Down-hole Phone

  1. Run gobhodo in directory with 6 chan
  records (3 down, 3 reference phones)

  2. Run gorunbhod in directory with files
  that are named hxxxyyy.seg

  ----------------------------------------
  IF examining the Reference Phone

  1. Run gobhodoR in the directory with
  the 6 channel records.

  2. Run gorunbhodR in the directory with
  files that are named rxxxyyy.seg

  ----------------------------------------

Because the tool is fixed for both source polarizations, there will be half as many orientation determinations as there are seismic files. The tips at the bottom of the dialog are reminders about which scripts are for what purpose. In a normal case, one generally is only interested in the down-hole tool orientation, and will only run scripts gobhodo and gorunbhod. If you are interested in viewing the actual rotation of the radiation from the source with time, then you will run the other two scripts, gobhodoR and gorunbhodR (the upper case “R” being a reminder that these are for the fixed reference phone). The following steps are recommended:

Make a new directory for the hodogram analysis under the current “seg” directory. For example,

mkdir hodo

will do this nicely.
Run the script, gobhodo in the current “seg” directory. This creates a lot of files.
Move the files beginning with “h” to the directory “hodo” created in step (1)

mv h*.seg hodo
Remove the bscl generated files.

rm bscl*
Copy gorunbhod to the “hodo” directory.

cp gorunbhod hodo
Change into the hodo directory and run gorunbhod. The important file to save is bhod.lst. It contains a list of the file numbers with tool orientations. Copy bhod.lst back to the “seg” directory.
Run the script, mergeplots, provided in the script directory of the distribution (see 6.7.4). This merges all the Postscript (*.ps) files into a single PDF file for viewing in a program like acroread (Adobe Acrobat).
Remove the *.ps files, and view the file, merge.pdf.
Look for $180^{o}$ reversals in polarity, particularly if the tool had to be released during any part of the survey (and hence was free to spin). Sometimes, a slight change in the guide vector azimuth will help on a second attempt.

Figure 12 shows the plot for the files L141.seg and L142.seg. The SH-wave enhancement involved subtracting the even from the odd files (see gobhodo script). The result was file h141.seg. It is this file, h141.seg which was then subjected to PCA analysis by program bhod.

**Figure 12:** PCA result (file h141plt.ps) for near surface geophone station.
$\includegraphics{Figure3}$

Since this is a shallow station, it corresponds to the observed orientation of the bow spring when the tool exited the hole. Without the guide vector (set at $240^{o}$ in the dialog above), the resulting R-component direction could easily have been rotated by $180^{o}$ due to the inherent ambiguity in the eigenvector solution. The line in the file bhod.lst corresponding to this station is: 00141 209.5 299.5 from which you can see the meaning of the 3 columns. The first column corresponds to the file number, the second column the R-component azimuth, and the third column is the T-component azimuth. The values will be truncated to integers when written to the headers by program btor.

There is one case in which bhod returned a tool orientation rotated by $180^{o}$ . This is for the deepest level. The tool twisted enough during the survey to make the single guide vector insufficient. With only one problem, it is easier to edit the bhod file directly.

Figure 13 shows the problem with the deepest level. In the figure are two hodograms, one for h001.seg, and the other for h003.seg (the next shallower station). You will see from examining file merge.pdf that h003.seg and shallower levels resulted in a consistent set of tool orientations.

**Figure 13:** Deepest level (A) is 180 degrees off from desired as shown in (B)
$\includegraphics{Figure4}$

The way to correct the deepest level here is to change the line in the bhod.lst file. The current first few lines are:

00001 155.9 245.9 <--this one is in error by $180^{o}$

00003 328.6 58.6

00005 323.1 53.1

00007 314.1 44.1

The first line should be manually edited so that these lines become:

00001 335.9 65.9 <--fixed, rotated by $180^{o}$

00003 328.6 58.6

00005 323.1 53.1

00007 314.1 44.1

The result is a consistent set of orientations, and one that continues to the surface and agrees with the observed orientation of the tool as it exited the hole.