Individual log curves are often off-depth with respect to each other. The problem is worse for older logs, because it may have taken several runs into the wellbore to obtain all the log curves required. Modern logs can also suffer from this problem due to mis-typed parameters in the logging software or logging speed changes. Even if the log was on-depth to begin with, the data we use may be "second hand", that is not the original digital data, but digitized manually or semi-automatically from paper copies.

Shifts are made with respect to a reference curve.
The choice of the reference curve is somewhat arbitrary but the first choice is often the gamma ray log, as it is the most common correlation curve. Resistivity is sometimes preferred. If a curve needs a shift, it is likely that all curves recorded on that logging run will need the same shift. For example, if the GR curve on a density neutron log is used as a reference and a resistivity curve is off-depth relative to this GR, then all the resistivity curves and the SP probably need the same shift.

There are numerous other kinds of edits that may be needed. Some examples are curve re-scaling or normalization, SP base line straightening, eliminating overlaps and bottom hole effects, filling gaps, and patching bad data. These are aided by appropriate editing functions in most software. A little common sense and observation of offset logs will be useful too.

There are two styles of depth shifting in most commercial software packages. The block shift moves a portion of a log curve up or down by a fixed amount. If the top and bottom boundary of the block are close in intervals with relatively constant log readings, the result at the boundaries of the block will not look too ugly. It is often used to place core gamma ray or core porosity and permeability on depth with log data.

Example of a Block Shift: top left shows correlation line between two curves, middle shows proposed shift, bottom shows final result.

The better method is to stretch and squeeze the log curves using some form of interpolation to obtain new data points at each data level that fit the original curve shape. The procedure is usually to pick peaks and/or valleys (or obvious bed boundaries) on two curves with a mouse click, where it is assumed that these points should be at the same depth. After picking enough such points, the software is asked to perform the interpolation. It can be undone and redone until satisfied.

Some software packages perform stretch and squeeze automatically using various forms of cross-correlation between the curves.

Example of a Stretch and Squeeze Shift: top left shows correlation lines between two curves, middle shows proposed shifts, bottom right shows final result.

Page Views ---- Since 01 Jan 2015
Copyright 1978 - 2017 E. R. Crain, P.Eng. All Rights Reserved