Preprocessing of logging data
|Development Geology Reference Manual|
|Series||Methods in Exploration|
|Chapter||Preprocessing of logging data|
|Author||J. G. Patchett|
The preprocessing or data preparation step is potentially the most time-consuming part of a large logging project. It is estimated that 60% to 80% of the project time is spent on editing and other data preparation tasks. These activities are key to a successful logging project.
Data preparation steps
Data preparation is usually conducted in a prescribed order. The order may vary slightly because of personal preference and the nature of available computer programs. The order recommended here is as follows:
- 1. Move digital log data from logging field tapes or digitized logs to the log processing environment
- If paper copies of logs were digitized, all digital log data should be plotted and compared with the original log hard copies.
- 2. Display all logging data
- This way the user can become familiar with typical responses for the different lithologies and/or formations in the study area. This is an initial quality control step and is also used as input to the merging of logging runs.
- 3. Collect all information related to the logs
- Locate the headers for all logs, and identify the logging tool model used on each well and logging run. Collect all temperature data and mud properties. Determine the top and bottom of the valid logging measurements for each curve to be utilized for each logging run. The stacking of logging tools can lead to large differences (several tens of feet) in the starting depth of individual measurements.
- 4. Merge logging runs (if required)
- When merging different logging runs in a single well, the first problem is that the depths may be different (for equivalent positions in the wellbore) from run to run. Depth shifts between runs should be made when necessary. Run overlaps should be noted because they allow the user to compare common-point depth measurements from different logging runs. This comparison of two logging measurements at the same depth aids in quality control.
- 5. Edit logs to eliminate invalid data
- Erroneous data can be recorded in casing at the top of the logging run and/or recorded with the tools setting on bottom before pick up. These data should be deleted (replaced with a missing data flag) on the edited copy of the logging curve. Logging data that are invalid because of environmental conditions (such as hole washouts or gas in the drilling mud) should also be deleted. This will result in data gaps, but these are preferable to erroneous data. If the data gaps occur within potential reservoirs, a replacement value of all reservoir parameters (such as porosity and water saturation) should be estimated from other logging measurements if at all possible. Logging data can also be incorrect due to incorrectly calibrated or malfunctioning logging tools (see ).
- 6a. Depth shift all log curves not recorded with the base curve or log
- When logging tools are run in sequence, differences always occur in depth from tool to tool and from run to run. Even when the logging tools are run in a single string there are potential depth differences due to differential cable stretch. Stretch can be pronounced when the logging tool string sticks or temporally hangs up in the hole. All logging measurements must be adjusted to a common depth reference before data processing can continue. A depth shift of 3 ft0.914 m
36 in can destroy an otherwise good correlation among logging measurements or between well logs and cores.
- All depths should be referenced to what is termed a base log. The base log is selected from a logging tool where strong or forceful tool positioning is not used. Free-moving tools travel through the borehole more smoothly than tools that are pushed with great force against the borehole wall, such as the density log. For this reason, strongly centralized tools are not selected as the base log. A resistivity log (induction or laterolog) is usually selected as the base log. For example, if gamma ray logs are available from both the density tool and induction tool strings, it is wise to select the gamma ray from the induction tool as the base log. The gamma ray from the density curve and all curves recorded with the density are then shifted to match the induction log depths. The base curve should also be selected based upon its expected strong correlation with the curves to be depth matched.
- Depth-shifting programs are commonly of two types: (a) automatic depth-shifting programs in which mathematical correlations are made among curves from different tool strings and the shifting is accomplished without user input, or (b) visual correlation programs in which the curves to be shifted are laid beside or on top of the base curve, allowing the user to instruct the program by noting correlative points on each log and calculating the depth offset. With older programs, the correlations can be made by using log prints and the shifts input to the screen or a file. Most programs allow the user to carry or cause the same shift to be performed on other curves recorded on the same tool.
- The depth-shifting operation necessarily stretches or shrinks the curve being shifted, thus it should be kept in mind that data are both created and lost in the process. For this reason, subsequent depth shifts (corrections) should start with the original raw logs, not with a previously depth-shifted copy.
- 6b. Depth shift core data to logs
- The depth correlation of log data to core data is frequently characterized by numerous abrupt changes in the amount to be shifted. Every trip with the core barrel is potentially a change in the relative core or log depth, even if continuous cores are taken. Zones flagged as lost core zones often are not where they were interpreted to be. Because of this, automatic depth shift procedures generally do not work when shifting core data to well log data. An overlay procedure is recommended where the core is segmented by core run and again where missing data occurs within a core. The core is then usually shifted by segments. Segments can separate or overlap. Separation is caused by incomplete or poor core recovery, and overlap can be caused by poor core-handling procedures. Review of the field core description can help clarify some of these problems. (For more information on cores, see Conventional coring, Core handling, and Core description.)
- A core gamma ray can be a valuable aid in establishing depth correlations between core and logs. Boyle's law core porosity and core grain density can be used to construct a core bulk density curve to correlate with log bulk density to determine the amount of depth shifts required. Core bulk density usually correlates well with the density log because lithologic variations are eliminated, resulting in two similar curves being correlated.
- While interpolation is a necessary step in the depth matching of wireline logs, it is highly undesirable when shifting cores. Interpolation should not be done when a core segment is shifted. Also, core data should not be resampled if least squares correlations are planned for calibrating logs or developing porosity and permeability relationships. Linear resampling of permeability destroys porosity and permeability relationships and can make statistical inference incorrect when making core to core or log to core data correlations. It is recommended that in any of these correlations the logs be resampled, not the core data.
- 7. Perform environmental corrections on logs
- Well logs are recorded in the hostile borehole environment where borehole size, temperature, pressure, mud properties, and other environmental factors affect logging tool responses. Logging tools are calibrated to make correct measurements only when specific environmental conditions exist (that is, an 8-in. borehole diameter at standard temperature). The purpose of environmental corrections is to correct the logging measurements to these standard conditions. Environmental corrections can be large. Some logging tools work over a much broader range of environmental conditions than others (see Table 1 or service company charts for more details).
|Gamma ray||H: GEN-7 S: POR-7 W:3-3||Mud weight, hole size, and tool position||Becomes significant as hole size and mud weight increase; charts also available for potassium, uranium, and thorium curves.|
|Guard logs||S: HR-8 S: RCOR-2A W:6-6||Mud resistivity, hole size, and tool position||—|
|Spherically focused||S: RCOR-1 W:6-6||Mud resistivity, hole size, and tool position||—|
|Induction||H:R-1E S: RCOR-4A W:6-2||Mud resistivity, hole size, and tool position||—|
|Microlaterolog microspherically focused proximity microguard||H.R-11 S: RXO-3 W:4-1||Mudcake thickness, Rxo zone resistivity, mudcake resistivity, and hole size||Mudcake thickness is difficult to determine|
|Compensated neutron logs||H: POR-14C S: POR-14C W:5-13||Hole size, temperature, pressure, borehole, and formation salinity; mud weight, mud type, mudcake thickness, and tool position||Charts tend to overcorrect in elliptical boreholes; some or all corrections may have been applied at the time of recording.|
|Density logs||S:POR-15A W:5-11||Hole size, formation density, and mud weight||Charts tend to overcorrect in elliptical boreholes|
The logging service company chart books contain correction charts for most logging tools and instructions for their use. Users of well log data should become familiar with these so that they can either apply the corrections or recognize the environmental conditions where corrections are not significant. Table 1 gives a summary of environmental corrections for common logging tools emphasizing the conditions for which corrections can be large. Table 1 also contains a partial list of recent service company charts used for typical corrections.a, b;  Older tools may require the use of older chart books. Slim hole models available for some tools may require their own specific charts. The user must become familiar with log headings to identify tool models so that the correct chart can be selected.
Environmental corrections are often performed serially. For example, Figure 1 is a flowchart illustrating the steps used to determine true resistivity from a dual induction log. It should be noted that shoulder bed effect corrections are difficult and require complex algorithms. Also, it is possible that not performing all corrections (e.g., invasion correction without borehole corrections) may be worse than making no corrections at all.
The SP curve presents special environmental problems that are not addressed by charts. Since the SP responds in part to the mud filtrate resistivity, no two SP curves will be recorded under exactly the same environment. (For more information on SP logs, see Basic open hole tools and Determination of water resistivity.)
- Difficult lithologies
- Formation evaluation of naturally fractured reservoirs
- Basic open hole tools
- Basic tool table
- Determination of water resistivity
- Wireline formation testers
- Basic cased hole tools
- Standard interpretation
- Quick-look lithology from logs
- Borehole imaging devices
- Lang, W. H., Jr., 1980, Porosity log calibration: The Log Analyst, v. 21, n. 2.
- Neinast, G. S., Knox, C. C., 1973, Normalization of well log data: SPWLA 14th Annual Logging Symposium Transactions Paper I.
- Patchett, J. G., Coalson, E. B., 1979, The determination of porosity in sandstones and shaly sandstones, Part 1— quality control: SPWLA 20th Annual Logging Symposium Transactions Paper QQ.
- Welex, 1985, Welex log interpretation charts: Houston, TX.
- Dyos, C. J., 1987, Inversion of induction log data by the method of maximum entropy: SPWLA 28th Annual Logging Symposium Transactions, Paper T.
- Western Atlas International, 1985, Atlas log interpretation charts: Houston, TX.
- Bateman, R. M., 1985, Open-hole log analysis and formation evaluation: Boston, MA, IHRDC, p. 15-189.