Blocking of Wells.docx

Blocking of Wells Before modeling can be performed based on well information, the well data must be upscaled to the resolution of the 3D grid. The cells intersected by the well tracks are identified, and each cell is given an average value for each selected log property. The scaled-up well data is s...

Blocking of Wells Before modeling can be performed based on well information, the well data must be upscaled to the resolution of the 3D grid. The cells intersected by the well tracks are identified, and each cell is given an average value for each selected log property. The scaled-up well data is stored in an object called a Blocked wells parameter. The following topics are covered in this section: Introduction to Blocking of Wells The Block Wells Dialog Box Data Selection Tab Parameters Tab Report Logs Tab Quality Control of Well Upscaling Introduction to Blocking of Wells Both petrophysical modeling and facies modeling are performed using a 3D grid cell structure to represent the value distribution within the grid model volume. Each cell represents one value corresponding to the petrophysical/geological model. As most well logs are measured every half foot (every 15 cm), the cell thickness in a 3D grid is normally greater than the well data sample density, as shown in the figure below. Accordingly, well data must be scaled up to the resolution of the 3D grid layout. Well data density and 3D grid layout The process of scaling up well data is performed in one operation, involving the following two steps: Blocking the well. This is the process by which all the grid cells penetrated by a well trajectory are identified, to create a cell-based version of the well path. Scaling up the selected log data belonging to each cell. Each cell is treated individually and given an upscaled value depending on the specified parameter settings. The grid layout defines the shape and size of the different cells and is therefore the most important input to the upscaling of well data. The orientation of the different grid layers has a direct impact on the number of possible well-to-well correlations, as shown in the following figure. Well-to-well correlation versus grid layout A single blocked well parameter can contain the geometry information and averaged log data for all the wells in that grid model. However, you can also create multiple blocked well parameters to try out different upscaling options for different sets of wells. Note You can merge new well logs into an existing set of blocked well data. This means that it is easy to update the existing blocked well data parameter when new data has been collected. A discrete blocked well log can be divided into several Well segments. A well segment is defined here as a sequence of neighboring cells having upscaled cells with the same code. For example, the same facies. In most cases, a well segment will correspond to the well interval occupied by a single facies body in the facies logs. The figure below shows how the start, end, and thickness of a well segment are dependent on the grid layout. Definition of a well segment in the blocking process The Block Wells Dialog Box The Block Wells dialog box is divided into three separate tabs: Data selection, used to select wells and associated logs to be scaled up Parameters, used to specify parameter settings for the upscaling process Report logs, used to define report logs generated in the process Each tab is described in the following sections. Data Selection Tab The Data selection tab (see the following figure) is used to select the name of the output object, together with a selection of wells and associated logs to be scaled up. Block Wells dialog box, Data selection tab Output Block Wells The Name of output blocked wells field provides a drop-down list of all existing blocked wells parameters for the current grid. In new jobs, the default name BW is displayed. Select or enter the name of the blocked wells parameter to which the upscaled data will be output. Enter a new name or select an existing blocked wells object. Keep Old Wells and Keep Old Logs If the output blocked wells object already exists, the new data will be added to the existing object. The original data can be retained by selecting one of the following check boxes: Keep old wells or Keep old logs. New data will then be added to the existing object. Expand Truncated Grid Cells The Expand truncated grid cells (using simbox definition) option defines handling of grid cells that have been truncated in the vicinity of a horizon or unconformity surface. The option can be used if the blocked wells object is to be used as input to both facies modeling and stochastic petrophysical modeling. The blocking is performed on a modified version of the ordinary grid. All truncated grid cells are stretched out, giving a new grid called "Full". This grid is not available and only used internally in the blocking algorithm. The algorithm creates a blocked well parameter that belongs to the original truncated grid. The following figures illustrate the blocking process that was performed, where the first figure shows the original grid, and the second figure illustrates the blocking process both with and without the expansion option. Original grid with truncated grid cells at the base of a zone Result of a blocking process without and with the expansion option One of the following four alternatives is selected for each zone, based on how the grid is created. For more information on how to change these settings manually, see Setting Simbox Thicknesses. Bottom - This option can be used for bottom-conformable zones with constant cell thickness and grid cells truncated against an upper surface. For each pillar the starting point is the lowest cell corner. All the grid cells are expanded upwards such that the thickness of a grid cell equals the cell thickness in simbox (given implicitly by the information in No. of layers and Simbox thickness). Top - This option can be used for top-conformable zones with constant cell thickness and grid cells truncated against a lower surface. For each pillar the starting point is the highest cell corner. All the grid cells are expanded downwards such that the thickness of a grid cell equals the cell thickness in simbox (given implicitly by the information in No. of layers and Simbox thickness). Estimate - This option can be used for any zone with constant cell thickness and truncated against bounding surfaces. This will be typically used for grids with truncated grid cells both in Top and Bottom or imported grids with a special structure. Proportional - This option can be used for all other grids, for example in proportional grids with no top or bottom erosion. This does not perform any expansion of grid cells in the blocking. The result of a property modeling process will generally differ depending on the blocking process that has been performed: an ordinary blocking process or expansion-based blocking process. A question arises: should the final simulation result match the blocked wells from the simbox or the ordinary method? Seen from the simulation algorithm's point of view, the simbox method is better. It gives correct conditioning and no bias in facies thickness, ranges etc. As the simulation is performed in the rectangular simulation box, one can argue that the simbox blocking algorithm is in fact the only correct algorithm. However, as the simulation results are transformed back into the truncated grid, inconsistency may be introduced. For example, where the bottom of one zone has truncated grid cells, the simulation could give a barrier toward the zone below, even if the well information says there is connectivity between the two grid models. It is possible to create an IPL script that changes the values in the simulation result to match the blocked wells using the ordinary method. Note The Expand truncated grid cells option cannot be used for repeat section grids. For this type of grid, a warning is displayed, stating that this setting will be ignored. Well Selection Click the Wells selection bar to open the selection pane (see the following figure). Selection pane, opened from the Block Wells dialog box This pane displays a list of all wells in the project. Choose required wells to be included in the new data set individually or, alternatively, click Open selector window... to open the Select Data dialog box and select an RMS Data Explorer collection (see also RMS Data Explorer Collections). If required, multiple jobs can be used to create multiple blocked wells objects, where each object contains a separate subset of the available wells. Log Selection The Block Wells job uses logs from the selected Default log run. Logs available for the selected wells are displayed in the list of Available logs. Select the logs you want to upscale by clicking them and transferring them to the Scale up logs list by clicking the right-arrow button. If required, you can return logs to the Available logs list by selecting them in the Scale up logs list and clicking the left-arrow button. When averaging log data, it is important that the raw log data is associated with the correct zone. The upscaling result can be improved by specifying a Zone Log, which will be used to adjust the upscaling of both continuous and discrete logs. A zone log contains information about which zones are found at various depths; see the Creating Zone Logs section for a description of how to create a zone log. If available, select the discrete zone log in the Available logs list, and click the upper-right arrow button to transfer it to the Zone log field. Algorithm Two algorithm options are available for upscaling the selected log data. You can switch between both options by selecting, or clearing, the Use deprecated (pre RMS 11) check box as appropriate. Select the check box for the job to use the upscaling algorithm of RMS versions prior to RMS 11. This is the default setting for blocking where it is restricted to block logs from only the default log run. So, the Available logs only shows a list of logs in the default log run. This option is dependent on log run MDs and the resolution of the MDs affect the blocking result. Clear the check box to apply the newer upscaling algorithm (RMS 11 and later). Here, there is no restriction for default log run and the Available logs list shows all log instances present in the project (all logs and their corresponding log runs). The log instances are presented in the format "Log run. Log", for example, log.Poro. See the following figure. For example,If Poro log is selected as the scale up log for blocking, Poro log is upscaled for each selected well using the default trajectory and the specified log run. Note You can select logs from any log run, a selected log can only be blocked for one log run only. For example, it is not possible to simultaneously block Poro log from log run log1 and Poro log from log run log2. Block Wells dialog box, Data selection tab illustrating newer upscaling algorithm (RMS 11 and later) selected As the new algorithm does not depend on the log run MDs, the trajectory geometry will be slightly different from that of the earlier Use deprecated (pre RMS 11) algorithm (see the following figure). a) Use deprecated (pre RMS 11) and b) New upscaling algorithm (RMS 11 and later) In the preceding figure, the blue trajectories are displayed using curvature interpolation and the red trajectories as linear interpolation and the colored grid cells (bright yellow, pale yellow and green) of the blocked wells represent each blocked well interval along the trajectory. The figures show that for (a), the trajectory sample points do not specifically coincide with the blocked well intervals like that of (b). Note For users with projects created in RMS version prior to RMS11, it is recommended to use the earlier RMS blocking algorithm Use deprecated (pre RMS 11) for data consistency. For setting up new Block Well jobs in RMS 11 or later, it is recommended to use the newer algorithm. It is also recommended to use only the new algorithm method for blocking of interval logs. Parameters Tab For each selected log, details of the upscaling method are specified in the Parameters tab. The top section Logs contain a list of all logs defined in the Data selection tab. Each log in the list can be selected and the lower part of the tab will update with the available parameter settings for the log. The options depend on the log type, which can be Discrete, Continuous, or Zone. The Settings section in this tab displays information according to the selection of Algorithm in the Data selection tab (see the following figure). Block Wells dialog box, Parameters tab, for a continuous log, Use deprecated selected for Algorithm in the Data selection tab A measured-depth thickness-weighted average can be applied to these log types by selecting MDT from the Thickness weighting drop-down list (see the following figure); this is applicable only for the newer upscaling algorithm (RMS 11 and later) as specified in the Data selection tab. Block Wells dialog box, Parameters tab, for a continuous log, newer algorithm selected (RMS 11 and later) The MDT option is a weighted arithmetic mean, wherein some data points contribute more than others to the final average. It uses the Measured depth thickness (denoted as MDT) as the weighting; note that the MDT is reduced to the actual thickness inside the cell (for more information on averaging methods, see Weighted Arithmetic Average in Appendix K - Rescaling Theory). The MDT option is particularly useful for blocking of interval logs where intervals (that is, MDTs) are non-regular, and a single interval can span a large section of a cell or even multiple cells. When selecting MDT, each data point is weighted by its effective length inside the cell; this cannot be achieved using a weight log, because the effective length inside the cell is calculated for each data point at the time of the blocking operation. The MDT option should be selected for Thickness weighting when blocking a log with non-regular and/or sparsely sampled MDTs. If all MDTs in the log are equidistant and densely sampled (as is the requirement for the deprecated pre RMS 11 algorithm), then the result for selecting MDT or None option for Thickness weighting is expected to be identical, as the effective length inside the cell will be equal for all the points inside the cell. However, the MDT option can be selected regardless of the input data. Note The Thickness weighting selection of MDT and Cell layer averaging option are the only settings shared for all log types. Discrete Log Parameter Settings Family Log and Object Log The facies modeling methods Facies Channels and Facies Composite can both use a Family log and/or an Object log. These modeling methods are described in detail in Facies Channels and Facies Composite. A set of a Family log and an Object log must be selected together with a discrete facies log, and they are all upscaled together in a unified manner. The families represents a subdivision of the facies log, where each family contains objects associated with one generation of meander belts, with the same orientation and large-scale sinuosity. Within a family, independently modeled objects such as channels or splays represent yet another subdivision. A Family log describes observations of facies objects belonging to the same family, subordinate to facies. Similarly, an Object log classifies observations of objects within a family. To select the required logs in the Data selection tab, first select a facies log in the list of Available logs and move this to the list of Scale up logs. From this list of Scale up logs, the internal order Facies/Family/Object can subsequently be seen in the Logs list of the Parameters tab, as illustrated in the following figure: Block Wells dialog box, Parameters tab In the facies modeling jobs, an upscaled facies log, family log, and object log will be available as alternatives for the definition of a Facies log. Cell Layer Averaging This technique is similar to simple zonal average calculations and is only applied when neighboring cells shift from being vertical to horizontal and vice versa. The blocking will be performed by integrating the available information of cell values between the top and base of one single cell layer. This will result in only one scaled cell for each layer. Before enabling Cell layer averaging, check that the data conform to the following assumptions: The variation of the properties exists in the vertical direction, or direction normal to the layering. This should at least be true within the well paths intersecting interval. The well path is a straight line through the cell layer. The well path never stays within the same layer for a long horizontal distance. The sample interval is constant. The intersected cells (in the same layer) are treated as a single cell during upscaling. The scaled value is assigned to the cell which contains the longest well path, and the other cells are set to undefined (see Well A in the following figure). If the well path is U-shaped and does not intersect the cell layer completely, all intersected cells will be set to undefined (see Well B in the following figure). Examples of cell layer averaging Treatment of Original Log The discrete log(s) to be scaled up can be treated in two different ways: Intervals - In this method, some markers are defined in the raw log. These markers are put halfway between all the existing data points (refer to the figure below). This will give a special treatment, where all point values will be "valid" within a certain interval. This interval is defined by the distance between two markers. The marker thus represents two values, from the points on both sides. If the cell density is denser than the measured points from the raw log, cells will still be assigned an interpolated value by using the Intervals method. Points - In this method, the log(s) are treated as Points. It means that only the point value within the actual cell boundary is used in the calculation of the upscaled value. Cells without any measurements will always be set to undefined (-999). The Interval method for a continuous log Priorities Discrete logs have integer values, typically facies or zone codes. In the upscaling, the dominant code from the raw logs is used as value for the grid cell. The dominant code is determined from a set of weights, where the facies code with largest weight is defined as dominant. The weight is defined as the product of two terms, a geometric weight GW and a user-defined weight UW: W = GW • UW When treating logs as Intervals, the geometric weight will be the total length of the well path of the given log type inside a cell. When treating logs as Points, GW will be the total number of data points of a given log type inside a cell. A user-defined weight can be set for each facies type in the interface. Select a facies type from the list and use the right arrow to define UW for this. If two or more facies have the same calculated weight, the one with priority highest on the list will be picked. Use the arrows to the right of the list to change the priority of the facies types. Continuous Log Parameter Settings When continuous logs are scaled up, the upscaled values are a result of harmonic, arithmetic, geometric, or power averaging methods. Average From the Average drop-down list, select one of the five averaging methods for scaling up the log values of the current continuous log. Available methods are: Arithmetic average Geometric average Harmonic average Power average Weighted arithmetic average For a detailed description, refer to Simple Averaging Methods in Appendix K - Rescaling Theory. For the first four alternatives, no weights are used in the calculations. In case of the last alternative, weights are defined from the length of the interval within the cell where the observed value is representative. The interval lengths are defined as described for Intervals in Treatment of Original Log. The weighted arithmetic average is illustrated in the following figure: Weighted arithmetic averaging For discrete and interval logs, the original log values are treated as constant forward sections, meaning that each original value defined at one MD is valid throughout the next defined MD. For continuous linear logs, the original values are treated as piecewise linear sections (see the following figure). In this example, Perm is an interval log and Poro is a linear log. Perm log values are treated as constant forward, while that of Poro are treated as piecewise linear. Plot illustrating an interval log (Perm) as constant forward section and a linear log (Poro) as a piecewise linear section Bias Log The blocking of the well data often results in petrophysical log measurements from different facies types (that is, lithologies) being lumped together and averaged. This can result in unrealistic average calculations, compared to taking the average of only those measurements belonging to the same facies type. A discrete log (normally a facies log) can be used to filter out data points coming from facies bodies lost in the upscaling process. This type of discrete log is called a Bias log, and the effect is illustrated in the figure below. The log must have been selected for upscaling in the Data selection tab and can then be selected from the drop-down Bias log list. Note If the bias log is missing for one or more wells, the associated log will not be blocked for these wells. Note If a zone log has been specified and an adjustment method has been set, all other logs will automatically be biased to it. If a Bias log is also selected, the continuous parameter will be biased to both the bias log and the zone log. Upscaling with Bias log Weight Log When blocking a continuous log, a weight log can be used to give relative weights to each observation. For instance, this can be used for a saturation log, where a porosity log can be an appropriate weight log. Cutoff When scaling up continuous logs, you can remove unreal/unwanted values. Toggle the buttons associated with Lower limit and Upper limit to specify the cutoff values. By default, no cutoffs are used. Cell Layer Averaging Refer to the description for a discrete log in the Cell Layer Averaging section. Interpolate This option corresponds to the discrete log option Treat original log as intervals. When choosing weighted arithmetic, Interpolate is always used. This option incorporates the intersection entry and exit points when calculating the average value for each cell (see the following figure). 3D grid illustrating intersection Entry and Exit points used for averaging Note The Interpolate option is only available for the earlier upscaling algorithm when selecting the Use deprecated (pre RMS 11) check box. Zone Log Parameter Settings Cell Layer Averaging Refer to the description for a discrete log in Cell Layer Averaging section. Treatment of Original Log Refer to the description for a discrete log in the Treatment of Original Log section. Zone Mapping The Zone mapping area is used to specify zone log values corresponding to each zone. This option is available for single-zone grids as well as for multi-zone grids. The zones are listed along with an additional entry for log values below the lowest zone. Three algorithms are available for controlling the way in which the zone log is used to bias the upscaling of all the other logs: Normal scale up - No adjustment of the logs according to the zone log will be performed. This is the default. Scale up biased to zones - This option will initiate a bias treatment of all the logs to be scaled up. If the raw zone log does not correspond to the zone layout, the cells in the overlapping area will not be given a value. This is only true if the whole cell holds only incorrect raw zone-log values. Shift and scale logs to match zones - Raw logs are shifted and scaled to match the zone. This option will ensure that all cells are given a value, even when biasing the upscaling to a zone log. This is done by shifting the raw log values to correspond to the zone layout. The three alternatives are illustrated in the following figure. Effect of three alternatives for zone mapping using the zone log Report Logs Tab The Report logs tab (see the following figure) is used to define a set of blocked logs, which can be generated as output, containing information about the upscaling process. Block Wells dialog box, Report logs tab Six different logs can be generated: Layer intersection angle - This will generate a blocked log with default name Angle, which describes the angle between the well trajectory and the dip of each cell. Angles are defined in the interval [-90,90] degrees, where 90 degrees is perpendicular and zero is parallel. Distance along well - This will generate a blocked log with default name Distance, which contains the absolute depth within a zone. Note that this is not the depth from the start of the well. Number of points in cell - This will generate a blocked log with default name Number_of_points, which lists the number of observations in the cell for each grid cell. Number of wells passing through cell - This will generate a blocked log with default name Number_of_wells, which indicates whether cells have been penetrated by more than one well. Cell entry measured depth - This will generate a blocked well log with the default name MD_Entry, which reports the MD entry information from the original logs. Cell exit measured depth - This will generate a blocked well log with default name MD_Exit, which reports the MD exit information from the original logs. Both MD_Entry and MD_Exit report logs also support raw log data that can be shown in scatterplots and any other relevant views. The default names can be changed in the field Log name. The option to Create raw logs is provided to generate a new log run in the raw Wells folder. Default step increment is 0.15. It can be changed by toggling on Constant step increment. Quality Control of Well Upscaling It is important to check that the averaging of the raw well data has been correctly performed and that the statistics of the data set has not been biased during the upscaling process. There are many ways in which quality control of the upscaled result can be performed, and the following is not meant as an extensive list: Use visual inspection of the blocked wells parameter, for example, in a 3D view (as illustrated in the following figure) a Chart view for a Blocked well data item (including Scatterplot, Histogram, and Rose Diagram) a Correlation view the Well Log Editor dialog box The Visual Settings dialog box controls display options for blocked wells in all three views. The upscaling can be run with alternative bias logs and results can be compared. Check statistics and view information, using the commands available from the task group in the Task pane associated with the blocked wells parameter. Carry out some simple data analyses. A wide range of data analysis options are available for blocked wells. Blocked wells data (porosity log) displayed in a 3D view Selection of Algorithm for Blocking Wells For the later algorithm used for blocking well (RMS 11 and later), you can manage the resolution of the output Blocked well raw logs from the User Preferences dialog box (see the Data Analysis section in the Setting User Preferences Help page). Setting and applying a value for the Number of samples per meter in the User Preferences dialog box denotes coarse or fine Blocked wells raw log data. This automatically updates blocked well raw log data in the following dialog boxes and views: Well Log Editor dialog box for a Blocked well data item (see also the Raw Logs section in the Blocked Well Log Editor Help page) Statistics dialog box for a Blocked well data item (see also Blocked Well Statistics) Charts views for a Blocked well data item (including Scatterplot, Histogram, and Rose Diagram) These updates are visible in the blocked well chart view and in the filter information for charts. It is recommended to refresh these chart views, if open, to see these updates. Note Raw log data is not stored in RMS; it is created on-the-fly when requested by jobs and by certain views. In addition, raw log resampling does not affect the blocking and upscaling of the original logs in any way; these are only used to handle the original logs, for example, in scatterplots and other places where a common set of MD values is needed.