Upscaling your well log

With the Upscale Well Log form (model > Rock Properties > Upscale Well Log or model > Facies > Upscale Well Log) you can upscale both discrete ('facies') and continuous ('rock properties') well logs to the resolution of your grid. Well log data is normally at a higher vertical resolution than the resolution of the cells in the 3D grid. Upscaling translates the higher resolution well log data to lower resolution grid cell values. A rock property log can (optionally) be biased to a facies log, to enforce consistency between the upscaled facies cells and their corresponding rock properties. The output of the upscaling is an upscaled grid property with upscaled data in the 3D grid cells that are penetrated by the well trajectories (i.e. the Blocked wells item in the JewelExplorer). All other 3D grid cells are as yet undefined.

The principle of well log upscaling.    click to enlarge

On the form, you can select which wells you want to include, the upscaling method and whether you want to bias your rock property log to a facies log, in which case you need to have an upscaled facies log available in your grid. The result is an upscaled grid property, which will be stored in the Properties > Upscaled Properties folder of your grid in the JewelExplorer.

You can visualize the upscaled property in the 3D View as a property of your Blocked wells (Blocked wells are automatically created during upscaling) or directly on your grid. You can also display your upscaled property in the Well View by selecting 'Add continuous grid property' or 'Add discrete grid property' as the log type. You can find the upscaled grid property in the property drop-down list as 'Upscaled Properties\ <name of the upscaled property>'.

To upscale your well log

1. At the top of the form, select the Grid to which you want to upscale the log.
2. Under Upscaled property, select Create new and type the name in the field below.
3. (Optional)  Autofill upscaled well log setting  With this option you can derive the upscaling settings of an existing 'well log upscaling case' from another 3D grid.
When the existing well log upscaling case (from which you want to derive settings) used a bias log (under Bias log on the form), make sure to derive the bias log first.

To do this:

• Click on the icon to open the Autofill Upscaling Settings form.
• Grid  Select the 3D grid that was used for the existing upscaled well log.
• Upscale well log  Select the log that was used for the existing upscaled well log.
• Click Apply to derive the settings and keep the form open, or OK to derive the settings and close the form.
4. In the Well log drop-down menu, first select the log type (left field). The field on the right filters all logs in your solution based on the selected log type. Select the log you want to upscale. The selected log will be upscaled to the cell resolution of the selected grid. You can also select an upscaled well log property that is created using the Property Tool.
If a log is created via the Property Calculator, not the log but the property calculator model is available in the drop-down list, displayed with its specific icon, . Select a model to use its associated well logs.
The calculation performed by well log upscaling depends on the 'Measurement type' of the selected well log. You can review this information on the Logs form.
Data with the measurement type 'Local measurement' (see image 2 below) are upscaled with equal weighting of all data points inside the cell. Data in adjacent cells have no effect on the result. Cells penetrated by the wellbore, but containing no data, remain undefined.
Data with the measurement type 'Local measurement with continuity' or 'Boundary measurement' ( see image 1 and image 3 below) are upscaled with weights determined by integrating the curve inside each cell (there is a continuous line between the data points which is used for upscaling). Data in adjacent cells affect the shape of the curve and influence the upscaled result. Cells penetrated by the wellbore but containing no data may nevertheless contain an interpolated curve and might receive an upscaled result, depending on the Minimum data coverage setting.





Schematic representation of the different 'Measurement type' settings for logs which impact the upscaling result (colored bars): 1) Local measurement with continuity, 2) Local measurement, 3) Boundary measurement.    click to enlarge

5. In the Wells section, check the boxes of the wells for which you want to upscale the selected log.
Important   Only select those wells that you want to use as input for your Facies or Rock Property Model. Once they are included in the upscaled property, you cannot exclude them anymore during facies or rock property modeling.
6. In the Upscaling Settings section, set your preferred settings with the following options:

Bias log  Only available when you are upscaling a rock property log (i.e. a continuous log). When you check the box, you can select a (discrete) upscaled log to bias the upscaling calculations of your rock property log. Upon selection, all options except 'Method' and 'Weight by' get grayed-out and will be automatically set to match the settings of the selected (discrete) upscaled well log.

The bias log restricts the input values for the upscaling to only those intervals that coincide with the class of the upscaled bias log. This way you can, for example, eliminate the influence of thin layers (for example, thin shale layers) on the calculation of the upscaled log value. The upscaled class (of the bias log) is considered the most important class, and using the bias log can provide an upscaled log value that is more realistic and suitable for rock property modeling.

Upscale to  The log is upscaled to either the K-layers or Zones of your selected grid. If you upscale to a zone that has more than one k-layer, all cells within that zone receive the same (upscaled) value.

Stratigraphic level  Select the stratigraphic level that contains the zones to which you want to upscale (only available when you selected to upscale to Zones). The stratigraphic levels in the drop-down are the 'levels' as shown in the Model Definition View, which can be opened by clicking Show model definition view on the Upscaling & Layers form (model > 3D Grid). Note that in most cases, the levels in the Model Definition View are similar to the levels of the Stratigraphic View, however, the Model Definition View can potentially contain more levels as it allows for adding and duplicating levels. If a zone contains more than one K-layer, all cells within that zone receive the same (upscaled) value.

Method  Select one of the following upscaling methods. Availability of methods depends on whether you are upscaling a discrete, or continuous log, and the log measurement type (local measurement, local measurement with continuity or boundary measurement):

• Sum  The upscaled value is the sum of the log values of the grid cell interval.
• Min  The upscaled value is the minimum value of the log values of the grid cell interval.
• Max  The upscaled value is the maximum value of the log values of the grid cell interval.
• Arithmetic mean  The upscaled value is the arithmetic average (the sum of the log values divided by the number of log values) of the grid cell interval.
• Most occurring  (Only available for discrete logs.) The upscaled value is the most occurring of all values in the grid cell interval.
• Nearest neighbor  The upscaling value is the nearest value to the center of the upscaled grid cell interval.
• Geometric mean  The upscaled value is the geometric average (the n^th root of the product of n number of values) of the log values of the grid cell interval. This averaging method 'normalizes' the range of values, in the sense that it reduces the effect of exceptionally high or low values on the mean. All of the well log values inside the grid cell must be non-negative. If this is not the case, the upscaled value for the cell will be set to undefined. If there are any zero log values in the cell, the upscaled cell value will be zero.
• Harmonic mean  This averaging method is often applied to data with log normal distributions, making it well suited for permeability which can exhibit several orders of magnitude difference between facies. All of the well log values inside the grid cell must be non-negative. If this is not the case, the upscaled value for the cell will be set to undefined. If there are any zero log values in the cell, the upscaled cell value will be zero.
• Root Mean Square  The upscaled value is the square root of the arithmetic mean of the squares of the log values of the grid cell interval.
• Median  The upscaled value is the median of the log values of the grid cell interval. The median is the value separating the higher half of the data sample, from the lower half. This averaging method reduces the importance of data outliers, for example caused by measurement errors.
• 1/3 Power average  The upscaled value is a power mean where the exponent p=1/3. This method is recommended for properties with a logarithmic scale and extreme range, such as permeability in carbonates. All of the well log values inside the grid cell must be non-negative. If this is not the case, the upscaled value for the cell will be set to undefined.
• Global distribution reproduction  Sample points are randomly picked from anywhere within the cell and the picked value is assigned as the upscaled value to the cell. Type a seed number (under Method Settings) if you want to exactly reproduce the results.

Strategy  Options vary depending on your selection under 'Upscale to':

• K-layer averaging  When a well penetrates more than one cell with identical k-layer ID, these 'neighboring' cells are averaged (upscaled) together and receive the same (upscaled) value.
• Zone averaging  When a well penetrates more than one cell with identical zone ID, these 'neighboring' cells are averaged (upscaled) together and receive the same (upscaled) value.
• Single cell  Each grid cell is upscaled individually.

Minimum data coverage per cell  Specify the minimum required percentage of the well trace to be covered by values, for the cell to be upscaled. For example, if the 'Minimum data coverage per cell' is set to 60%, this means that gaps in the log values (i.e. intervals where the log is non-existent, or 'undefined') are not allowed to exceed 40%; if the gap(s) exceed 40% of the well trace, the cell is excluded from upscaling. The resulting grid property for that cell will be 'undefined'. Right-click on 1, and you can switch between fraction and %.

Leave cells undefined where angle is smaller than  By checking the box and filling in a threshold angle, you can define which parts of a (horizontal) well are excluded from upscaling: all cells within those sections will get the value 'undefined'. Threshold angle is the angle between the grid cell and the wellbore trajectory.

7. Depending on the type of log you selected for upscaling and the upscaling method, different settings are available in the Method Settings section:
• Weight by   (only available when you are upscaling a continuous log in combination with method Arithmetic mean, Geometric mean, Harmonic mean, Root mean square or 1/3 Power average)  From the Weight by drop-down menu, select another continuous log as weight. For example, you can use a net to gross log as a weight when upscaling a porosity log to calculate the average porosity in the net bulk volume.

A weighted average is a type of mean that gives differing importance to the log values of the log being upscaled. Calculating the weighted average involves multiplying each data point by its weight and summing those products, then summing the weights of all data points and finally dividing the weight * value products by the sum of the weights. If the weights are all the same then the result is the same as an unweighted average.

A weighted arithmetic average of x, for example, is:



x = log value of log being upscaled

weight = log value of the 'weight by' log

The 'Measurement type' of the weighting log needs to be set to either 'Local measurement with continuity' or 'Boundary measurement' (not 'Local measurement') for the log to be used as weighting log. You can change the Measurement type of the weighing log on the Logs form.

Select one of the three options on how undefined and zero values in the 'weight by' log should be handled:

• Undefined when all the weights are zero or undefined   (default) The upscaled value of a cell will be undefined when all the weights of the log used for weighting are zero or undefined.
• Unweighted when all the weights are zero or undefined   The upscaled value of a cell will be unweighted when all the weights in the log used for weighting are zero or undefined.
• Unweighted when <percentage> or more of the weights are zero or undefined   Fill in a percentage. The upscaled value of a cell will be unweighted when the percentage of zero or undefined values in the log used for weighting is greater than or equal to the percentage you entered.
You can simultaneously use the Bias log option (under Upscaling Settings) with the Weight by option. The bias option restricts the number of samples used for the upscaling; the weighting option subsequently weights the remaining samples.
• Seed   (only available when you are upscaling a continuous log in combination with method Global distribution reproduction)  The seed number determines the starting point of the data sampling. You have two options:
• Click the dice to fill in a random seed number.
• Type a seed number in the field if you want to reproduce the results of the global distribution reproduction. If you manually typed the seed number, it will be stored with the upscaled well log.
• Proportion reproduction   (only available when you are upscaling a discrete log)
• Weighted   Type the weight (value between 0 and 10) that you want to apply to each class in the Weight column. For each facies class, the occurrence within each cell is multiplied by this value, and subsequently divided by the total of all weights within the cell. This way, you can allow a class to 'win' a cell, even when its fraction is low.
• Global   Use this option when two (or more) facies are present in almost the same quantity and you want to prevent that one facies class is ignored. The higher you set the percentage, the more the original distribution of the facies classes is adhered to during upscaling.

Random selection   Move the slider to change the percentage.

Seed   The seed number determines the starting point of the data sampling. You have two options:
• Click the dice to fill in a random seed number.
• Type a seed number in the field if you want to reproduce the results of the global distribution reproduction. If you manually typed the seed number, it will be stored with the upscaled property.
8. With the Depth Alignment option you can choose to use markers or grid elements (zones or k-layers) to define the log interval to be used for upscaling. Marker depth alignment might be preferred in case the markers have a depth mismatch with the grid element tops and bases. To QC this mismatch with the 'Create All Residual Log Shift Logs' option, see Visualizing the residual log shift on how to do this.  Note that when the grid is perfectly aligned with the markers, these two depth alignment options produce identical upscaling results.
• Marker depth alignment  (default) The log interval between the markers is squeezed or stretched to align with the grid element, then upscaled to the grid element. When you are upscaling to k-layers (where markers only exist for the top of the shallowest and base of the deepest k-layer), this means the squeezed or stretched log interval is equally divided over the k-layers, before being upscaled to each k-layer.
• Grid depth alignment  The log between the grid element (zone or k-layer) is upscaled directly to the grid element.




Depth alignment options: (1) Marker depth alignment, (2) Grid depth alignment.    click to enlarge

9. Click OK to calculate and create the upscaled grid property and close the form. The upscaled grid property is added to the Upscaled Properties folder under your 3D Grid in the JewelExplorer.

Note: If not already present, the process of well log upscaling creates blocked wells immediately after the upscaled property is created. The blocked wells are located under the 3D Grid in the JewelExplorer. For more information see Blocked well.