GB2322197A

GB2322197A - Well log stacking

Info

Publication number: GB2322197A
Application number: GB9703521A
Authority: GB
Inventors: David Gordon Quirk
Original assignee: Individual
Current assignee: Individual
Priority date: 1997-02-18
Filing date: 1997-02-18
Publication date: 1998-08-19
Anticipated expiration: 2017-02-18
Also published as: GB9703521D0; GB2322197B

Abstract

A digital processing technique used to enhance well log trends and provide an objective measurement of the predictability of rock bodies in the sub-surface. Similar well log data in related wells are depth corrected using at least two datum levels in a reference well which represent the depth of correlation boundaries identified in all the wells. The depth corrected log data in each well is digitally resampled in the same way as the reference well. For each sampled depth, the log values in all the depth corrected wells plus the reference well are summed and then divided by the total number of wells 60 to produce a mean value. All of these mean values are plotted against depth to produce a stacked well log trace 64. Prior to applying the process, the data may be filtered and normalised.

Description

WELL LOG STACKING This invention relates to a well log processing technique known as well log stacking.

The correlation of well logs (or electric wireline logs) forms an essential part of the interpretation of geological strata in the sub-surface as applied in the exploration and production of hydrocarbons. It involves visually matching the log curves produced from digitally sampled data for one well (or borehole) with those from one or more adjacent wells and is used primarily to predict the lateral extent of rock bodies.

However, at present, the standard method of correlating logs is subjective in that it is based on a person characterising the appearance of an interval in each well by eye, defining boundaries and then connecting these boundaries between wells.

The invention described below is an automated technique which is unique in that it i) allows a quantative assessment of the predictability of geologic intervals in the sub-surface based on well log data; ii) can be used to objectively identify zones with high reservoir (or aquifer), sealing (or aquitard) and hydrocarbon source potential over the area of study; iii) condenses the number of well log traces used to characterise a geologic interval in a particular area; iv) allows regional well log trends to be enhanced and local effects to be suppressed; v) standardises well log interpretations and petrophysical evaluations; vi) can be used to produce average measurements of rock properties in other applications; vii) lessens the effect that a mis-correlated well has on the overall correlation; viii) allows well log correlations to be refined in a non-arbitary way by increasing the confidence and precision by which correlation boundaries are identified; ix) can be used to identify possible mis-correlations in a statistical manner; x) allows a quantative assessment of the differences in the relative amount of sediment accumulation and compaction between different wells According to the present invention there is provided a technique for processing digital log data from a number of wells involving identifying at least two initial correlation boundaries, then editing, smoothing, normalising, depth correcting and proportioning the data in order then to finally sum the data to produce a single stacked well log which represents an average trace of all the wells.

A specific embodiment of the invention will now be described by way of example with reference to the accompanying diagrams in which: Figure 1 illustrates an example of filtering well log data using a moving average algorithm to remove mechanical noise and smooth the trace.

Figure 2 illustrates the process of normalising well log data to a standard shale and sandstone base line determined from a reference well.

Figure 3 illustrates the process of depth correcting well log data relative to correlation boundaries using as datum levels the depths of these correlation boundaries in a reference well.

Figure 4 illustrates the process of resampling depth corrected well log data at the same depths as the levels sampled in a reference well.

Figure 5 illustrates the process of producing a stacked well log by calculating the mean value of the depth corrected log data from N- 1 wells plus 1 reference well for each sampled level.

Figure 6 shows a comparison between the depth corrected log trace for an individual well and a stacked well log trace in order to quantify the predictability of specific rock bodies identified in the sub-surface.

Figure 7 illustrates an example of a statistical measurement of the confidence in a well log correlation by calculating the degree of variation in log values from each sampled level in depth corrected data relative to the stacked well log.

Figure 8 is an example of filtered, normalised and depth corrected gamma ray log traces from six wells used to produce a stacked well log shown on the right.

Referring to the drawings, after first selecting the geological interval to be processed 10, the following steps are carried out separately for each type of well log 12.

In order to remove non-geologic noise from digital well log data 14, the data is usually filtered 16 so that random variation between digital samples is lessened 18 and the log trace is smoothed 20, as shown in Fig 1.

For some types of log data, the average log expression of two pure lithologies such as shale and sandstone in a reference well are used as standard base lines 22 to normalise 24 each well log by changing the value of each depth sample 28 by an amount proportional to their actual values and to the difference between the standard base lines 22 and base lines in the well 30, as shown in Fig 2. The normalised digital data 32 and the resultant trace 34 is similar to the original log except that the normalised log is shifted horizontally 35 and may be narrower or wider than the original trace 26.

The depths of at least two correlated boundaries 36 are each shifted in every well to datum levels 38, as shown in Fig 3. The datum level chosen is the depth of the correlated boundary in a reference well 40. In order to produce a depth corrected well log 45, the depth of every digital sample in each well 42 is corrected 44 by an amount proportional to the shift applied to the correlation boundaries 38, as shown in Fig 3. The depth corrected digital data 46 from each well is then resampled 48 at depth points 50 equivalent to the sampled levels for the log data in the reference well 52, as shown in Fig 4.

The digital value from every sampled level in the reference well 52 is summed 54 with the values corresponding with the same depth corrected point in all other wells 56 and then divided by the total number of wells 58, as shown in Fig 5. Each resultant value represents the average of all the wells 60 and all of these samples are used to form a new digital well log known as the stacked well log 62. On the resultant trace of the stacked well log 64 the original correlation boundaries can be identified visually 40 as well as alternative positions for these boundaries 66 and the positions of potential new correlation boundaries 68, as shown in Figs 5 and 6.

In order to measure the predictability of a rock type or its petrophysical expression on any correlated well log, a depth corrected log from the well 70 can be compared with the stacked well log 64, as shown in Fig 6. The difference between the two curves 72 relative to a background value or standard base line 22 is a quantative measure of the predictability of that lithology or its log character.

The confidence in any well log correlation is expressed as the statistical variation 74 in log values 52 56 around the stacked value 62 for each sample level which may be plotted as a histogram versus depth 76, as shown in Fig 7.

On the basis of trial tests shown in Fig 8, a stacked well log that is statistically robust 78 is produced when depth corrected data from a minimum of four different wells 80 are used in the stacking process 60.

Claims

1. A digital well log processing technique known as well log stacking which involves depth correcting each of at least two correlated geologic boundaries correlated in related boreholes to a datum equating with the depth of each boundary in a reference borehole. Thereafter, the depth corrected data in each well is digitally resampled at the same levels used in the reference well and then added to the corresponding value in the reference well at each sampled depth point. The resultant summation is then divided by the total number of wells, including the reference well, to produce a mean digital value for each sampled depth point. The mean values of all sampled levels represent the stacked well log which can then be plotted as a trace of the values against depth.

2. A digital well log processing technique known as well log stacking as claimed in claim 1 wherein the well log data may initially be mathematically filtered to suppress mechanical noise and smooth the traces.

3. A digital well log processing technique known as well log stacking as claimed in claim 1 wherein the well log may initially be normalised to two base lines representing the ideal log expression of two pure lithologies such as sandstone and shale.

4. A digital well log processing technique known as well log stacking as claimed in claim 1 and claim 3 wherein the predictability of the lithology of a geologic unit identified in a particular well can be measured as the difference between the depth corrected log expression of the unit in that well and its log expression in the stacked well log compared to a background value or base line.

5. A digital well log processing technique known as well log stacking as claimed in claim 1 wherein the resultant represents a standardised average trace in which regional log trends are enhanced and local effects are suppressed that can be used for other applications such as a sequence stratigraphic interpretation or the production of a synthetic seismogram.

6. A digital well log processing technique known as well log stacking wherein it can be used to refine well correlations by applying the process as claimed in claim 1 to a simple initial log correlation of a number of wells and then, by comparing the difference between the resultant and each depth corrected well log, problems with the initial correlation can be identified and new correlation boundaries defined. The new correlation boundaries can be used to repeat the well log stacking process in an iterative manner.

7. A digital well log processing technique known as well log stacking as claimed in claim 1 wherein the depth correction of correlation boundaries causes adjustments to the thicknesses of the intervals to be made between the boundaries. The amount of adjustment represents a measurement of the relative differences in the rate of sediment accumulation plus compaction which may be used in further applications such as the analysis of syn-sedimentary subsidence rates or burial history modelling.

8. A digital well log processing technique known as well log stacking as claimed in claim 1 wherein the confidence in a log correlation can be measured statistically as the variance or standard deviation between the log value in any or all the wells at any depth corrected sample point or set of points relative to the value in the stacked well log.

9. A digital well log processing technique known as well log stacking as described herein with reference to Figures 1-8 of the accompanying drawings.