SPE 164420 Petrel Workflow for Adjusting Geomodel Properties Propert ies for Simulation David R. Hoffman, Tatweer Petroleum
Copyright 2013, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Middle East Oil and Gas Show and Conference held in Manama, Bahrain, 10–13 March 2013. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract During the initialization phase of a simulation modeling project, one of the most critical problems to overcome is the inevitable disconnect between the geomodel properties and reservoir performance in completion intervals. These disconnects can be the result of incorrect perforations, simple data errors, and/or poor reservoir characteristics. Poor reservoir characteristics, in turn, can be the result of sparse or clustered well control, erroneous petrophysical interpretations, or improper model property distribution. A simple Petrel-based workflow is described which can be used to automatically modify geomodel properties in model cells cells associated with completion intervals. Using the modified model properties, the numerical simulation model can be initialized and the simulation engineer can work on other early-modeling issues, while the modified property can be used to isolate property modeling errors for further evaluation and correction in the geomodel . The workflow makes the basic assumption that if an interval has been perforated at any time in the well history, then that portion of the reservoir is assumed to meet minimum reservoir criteria for pay. Using the historical well events, a pseudo-log is created and used to generate a Petrel model property. This property is then used to evaluate and/or modify reservoir properties in completion zones. If a completed zone has reservoir properties which do not meet pay cut-off criteria, that interval is modified by the workflow so that it passes the pay cut-off without changing other parts of the geomodel. In the Awali Field model tested, the modified cells constituted less than 1% of the total model volume and the modified geomodel eliminates “non-connection” errors in the simulation model (however, the reservoir quality issues still exist). Using property filters, the modified cells can then be used to by the geomodeler to isolate the problem areas of the model for further evaluation and correction. Although this methodology is presented using Petrel as the geomodeling platform, the same approach could be easily adapted to most geocellular modeling applications. Introduction The use of 3D geocellular models has become both commonplace and necessary to conduct numerical simulation of reservoir performance. Although most geoscientists and engineers agree that the process of numerical simulation is iterative, the “feedback loop” between reservoir engineers and geoscientists is often times less than perfect. One of the most common problems experienced in the early stages of a simulation project is the disconnection between actual production and perforated intervals in the geocellular model. Producing zones which are not “connected” in the model can be caused by several problems (or a combination of problems), such as: § § §
Incorrect Perforations Production Allocation Problems Poor Reservoir Characterization
Each of these problems must be evaluated and solved before an accurate numerical simulation will be meaningful. The first two issues can be solved through careful, though tedious, checking of completion and workover records and historical production data. The problem of poor reservoir characterization brings into question the fundamental geological and petrophysical assumptions which were used to generate the geomodel in the first place. These problems are often the most time-consuming and difficult to resolve, as they can result from a lack of sufficient
2
SPE 164420
local well log control, problems with individual or local petrophysical interpretations, and/or incorrect statistical distribution of model properties. Regardless of source of the problem(s), delaying the initial steps in the numerical simulation can be very detrimental to the overall simulation project effort. Ideally, then, it would be very beneficial to have a method which would allow modification of geomodel properties, but only in zones associated with confirmed production zones. The workflow presented here provides that methodology, and results in a simulation-ready geomodel where connection problems have been greatly reduced or eliminated altogether. Assumptions and Limitations The underlying and fundamental assumption employed in this workflow is that any completed interval in a well which has been confirmed productive (including water-producing intervals) should meet minimum reservoir-quality requirements. While there are several geocellular modeling and data management applications available, the workflow process presented in this paper assume the use of Schlumberger’s Petrel modeling software and Landmark OpenWorks for data management. Workflow The workflow consists of six basic processes, not all of which will be required in all cases depending on the data source available, or the stage in the overall geomodel construction: § § § § § §
Export of completion data from database Reformatting of data for Petrel Loading of production log data to Petrel Upscaling of production logs Creation of pseudo properties Modification of geomodel properties
Export of Completion Data – Even if completion intervals are present in the geomodel, it is sometimes easier to start with the raw data directly from the database. Using Landmark OpenWorks, raw completion data are exported using the Well Data Export utilities included in OpenWorks. Regardless of the source of the data, the goal is to generate an ASCII-format text file which contains all the completions for each well in the field. The format of this file can vary, but in this example has a format as shown in Figure 1.
Figure 1 - Format of exported completion data
In this example, the output data includes the well name, UWI, measured depth of the top and base of the perforated interval, the type of well event (perforation, squeeze, etc.), the event data and event sequence, and the perforated interval or zone.
SPE 164420
3
Reformatting of Data for Petrel – This workflow requires that the raw completion interval data be converted into “pseudo-logs” for each well, where completions are designated with values of “1” and the remainder of the well is designated with “0” values (i.e. binary curves). Because there is no commercial software which can easily reformat the data, the author developed a simple Visual Basic application which is used to do the reformatting. Figure 2 shows the user interface and summarizes the features and options of this program.
CUSTOM UTILITY TO RE-FORMAT COMPLETION DATA PROGRAM FEATURES: §
Input & Output File Name Selection §
§
Standard file selection dialogs
Output Filename & Extension §
Allows removal of header lines
§
Preview of Input Data File
§
UNIX-DOS Conversion Option
§
Input File Format Options §
Custom formatting of data columns
Figure 2 - Reformatting completion data for Petrel
Once the data have been reformatted using the utility, “pseudo-logs” will have been created for each well, although for convenience all the logs are included in a single multiplexed file to facilitate loading to Petrel. Figure 3 shows the reformatted data ready to be loaded to Petrel:
4
SPE 164420
COMPLETIONS CONVERTED TO COMPLETION LOGS
ORIGINAL INPUT DATA
After processing the input data file using the custom utility, all completions are exported to a multiplexed set of data which are in standard Petrel completion log format. The log is a simple, binary “switch” with “1” representing a perforation and “0” indicating no perforation.
Figure 3 - Data reformatted as production logs, and ready to be loaded to Petrel
Loading Production Logs to Petrel – Now that the “pseudo” production logs have been generated they are loaded to the Petrel project database. It is assumed that the reader is familiar with the basic operation of Petrel with regards to data loading, but Figure 4 illustrates the basic steps in Petrel which are used to load production logs
SPE 164420
5
IMPORTING COMPLETION LOGS TO PETREL
The completion logs are imported by selecting the “Import (on selection)” option from the context menu on the “Wells” folder, then selecting the “Multiple well logs” option from the import dialog. On the Input Data tab of the import dial og, set the column and file format to match the input file format. If necessary, Units or Other settings are modified on the Coordinates and Units tab.
Figure 4 - Loading completion interval data to Petrel as production logs
Upscale Production Logs – The next step in the workflow is very important, as the completion information (as production logs) must now be upscaled into the model so that model cells which are directly associated with completions will have a unique value. As before, it is assumed that the reader is familiar with this process, but Figure 5 shows the basic steps in this process.
6
SPE 164420
UPSCALING COMPLETION LOGS Before the completion data can be used, the completed intervals (represented by “1” on the completion log) must be upscaled into the Petrel model as a new model property using the Scale Up Well Logs process in Petrel. Since the completion logs consist of “0” and “1” values, use the “Maximum” method for scale-up averaging. Any model cell intersecting a completion interval will be assigned a value of “1”
Figure 5 - Upscaling completion data into the geomodel from production logs
Creating Pseudo-Properties in Geomodel – Now that the model cells which intersect completed intervals have been designated, the next steps in the workflow populate the model with reservoir-quality values for each property to be modified, but only in completed intervals (the remainder of the model is filled with “0” values). Using the Petrel property calculator (Figure 6), the entire model volume is filled with default values in completed intervals and “0” for all other model cells.
SPE 164420
7
CREATING PSEUDO-PROPERTIES Pseudo-properties are created in the model using the Petrel property calculator. A conditional statement is used to populate the model with “default” property values for completed zones, and values of “0” where there are no completions The conditional statement in the property calculator sets the model property to the chosen “default” in cells intersecting completions and “0” values for the remainder of the model volume.
Figure 6 - Using the property calculator to create pseudo-properties
It is important to choose the “default” values for each property carefully, as these values can later be used to filter the model for further quality control and evluation. In this example, for instance, the default value of porosity for a completed interval is 0.1234. This value was chosen as it is extremely unlikely to occur in reality, but will still be considered “minimum reservoir quality” in this particular reservoir. As discussed in the example later in this paper, the default values can be used as a filter to identify portions of the model where artificial modifications have been made. Modification of Model Properties – In the final workflow step, the Petrel property calculator is used to modify the actual model properties. A logic statement compares each cell and replaces the actual property with the pseudoproperty if the pseudo-property is greater than the actual property. Since the pseudo-property can only exceed the actual property in completed intervals (and only if the pseudo-property is larger), all the remaining model volume will be unchanged. Figure 7 is an illustration of this process.
8
SPE 164420
FINAL REMODELED PROPERTY In most cases, the “remodeled” property will be visually indistinguishable from the original property. Remodeled cells will generally comprise less than 1% of the total model volume, as only cells intersecting completion intervals with values below reservoir quality minimum will be changed
Figure 7 - Final remodeled property
Application of Methodology Using this workflow to temporarily modify basic model properties can speed and simplify model initialization. The “improved” reservoir properties should eliminate many simple problems where reservoir properties are marginal, where wells do not have valid petrophysical data or logs, and out-of-zone completions. Of course, this method does not actually correct problems, but in many cases can be used to isolate parts of the model which need further review, or provide a means to “correct’ model problems where conventional modeling approaches cannot be used. The following examples illustrate the practical application of the workflow. Example 1 – Filtering Model to Isolate Possible Problems In the first example (Figure 8), the model is filtered to show only the areas where corrections to the model property were needed to meet minimum reservoir criteria. Using the property settings, the model is filtered by setting the filter maximum and minimum values equal to the “default” value for the pseudo-property. When the filter is applied and visualized in a 3D window, these areas can then be evaluated to determine what corrective action is needed to permanently “fix” the initial problem.
As discussed earlier, default values should be chosen to represent minimum r eservoir quality, while retaining a unique numerical sequence which is unlikely to occur in nature. In the example, the default value for porosity is 0.1234. Since the minimum reservoir-quality porosity is 0.10, the default will slightly exceed the minimum cut-off value while still being “unlikely” as valid log data. Similar values for water saturation (e.g. 0.5678), clay volume (0.4567), or NTG (0.9876) could be used, depending on the actual reservoir cut-off values for reservoir quality.
SPE 164420
9
APPLICATION OF WORKFLOW
EXAMPLE OF MODEL FILTERING FOR QC Setting the property filter MIN and MAX values to the pseudo-property default value will allow display of only those portions of the model which have been modified. This allows much easier QC of areas of the model where completion and production data are inconsistent with petrophysical data
Figure 8 - Filtering to determine portions of the model for review and correction
Example 2 – Adjusting Models for Missing or Insufficient Petrophysical Data Many older fields have wells which have been completed and produced, but do not have sufficient well log data to use in a conventional petrophysical interpretation. Although sophisticated methods using neural networks and similar approaches are available to estimate log data in these older wells, we can use this workflow to “remodel” areas of the property model where log data do not provide accurate property distribution. While this is not a substitute for a comprehensive petrophysical evaluation, this workflow provides the simulation engineer with a quick means of initializing a model.
In this example (Figure 9), older wells have been completed and produced in a reservoir, but do not have modern log data to incorporate into the field-wide property distribution. As a result, the statistical distribution of the property in the vicinity of these wells yields values which do not meet minimum reservoir cut-off criteria. Although the property modeling is statistically valid, and honors all the actual well data, there is no simple way to account for the lack of well log control for these wells.
10
SPE 164420
APPLICATION OF WORKFLOW
EXAMPLE OF SIMPLE PROPERTY REMODELING In this example, two wells are completed in a reservoir, but have no valid porosity logs (old wells). Although both wells penetrate cells that are “close” to reservoir quality, they do not meet the minimum cut-off value (in this case, 10%) Property modeling is statistically valid, but the lack of well control at the completed wells does not adjust local properties for likely reservoir quality (based on completions and production history
ORIGINAL POROSITY PROPERTY
Figure 9 - Model corrections for missing or invalid log data using simple property remodeling
The first attempt to correct for this situation is to use the simple property remodeling workflow. As shown in Figure 10, the resulting model has been adjusted in such a way that the cells penetrated by the completed wells now meet the minimum reservoir quality criteria.
SPE 164420
11
APPLICATION OF WORKFLOW
EXAMPLE OF SIMPLE PROPERTY REMODELING After “remodeling” the porosity property, the cells associated with the completed intervals have been increased to the “default” value (which meets reservoir minimum cut-off value) Lack of well control in this case may require that a manual local adjustment to the reservoir quality be made, or possibly a pseudolog approach could be used to remodel local properties near the well
REMODELED POROSITY PROPERTY
Figure 10 - Resulting model using simple property remodeling method
Example 3 – Adjusting Models Using a Pseudo-Log Approach While the simple property remodeling method discussed above will eliminate connection errors when a simulation model is initialized (in most cases), as the simulation engineer starts comparing model volume with produced volumes, additional problems will arise in areas of the model where there is insufficient volume to produce the actual produced volumes associated with a completion.
Figure 11 illustrates a situation where several wells (all older wells lacking modern petrophysical log data) are located in a portion of the model with very poor reservoir quality, as defined by the actual log data from surrounding well control. Although these wells have produced significant volumes of hydrocarbons, the simulation model will produce errors due to the very low model volume in that region.
12
SPE 164420
APPLICATION OF WORKFLOW
EXAMPLE OF PSEUDO-LOG REMODELING Creating “pseudo-logs” for perforated intervals can help remodel properties in areas surrounding completions if indicated by production history
In this example, three wells have completions in the zone, but no valid log data. A workflow can be used to remodel properties in the region surrounding the wells for a more realistic depiction of local reservoir properties
ORIGINAL POROSITY PROPERTY
Figure 11 - Using pseudo-log remodeling; initial property conditions
The first step in this approach is to apply the simple property remodeling workflow as discussed in the previous example. As before, the model properties intersected by the wells will be improved (Figure 12), but because the overall reservoir quality is so low in that region of the model there will undoubtedly be problems with the numerical simulation. To correct the surrounding areas of the model requires an additional workflow.
SPE 164420
13
APPLICATION OF WORKFLOW
EXAMPLE OF PSEUDO-LOG REMODELING Using the workflow described, the model properties can be modified in the cells penetrated by the completed wells ….
…unfortunately, production data indicates that the local reservoir volume is insufficient to match produced volumes. Another workflow is needed to modify the properties locally
REMODELED POROSITY PROPERTY
Figure 12 - Property adjusted using simple remodeling method
To modify the local model properties in the region near the productive wells, it is necessary to create “pseudo-logs” from the model for the specific wells which need to be modified. The first step in this process is to generate synthetic logs for the remodeled property using the Petrel “Make Logs” process. Figure 13 shows the basic steps in the creation of the synthetic logs from the remodeled property.
14
SPE 164420
APPLICATION OF WORKFLOW
EXAMPLE OF PSEUDO-LOG REMODELING WORKFLOW §
Create Remodeled Property Log
§
Upscale Remodeled Property Log
§
Re-Run Petrophysical Modeling
Right-click on the WELLS folder, open Settings dialog, select Make Logs tab, select property to create a log from, and click the Make Logs button
Figure 13 - Creating remodeled property pseudo-logs
Next, the synthetic logs are upscaled into the model on a selective basis. The Petrel well log scale-up process has the option to scale-up logs into an existing property, but the user must be very careful to not to delete the already scaled-up cells from wells with valid petrophysical data. In this example, the three wells are selectively upscaled into the existing property (Figure 14).
SPE 164420
15
APPLICATION OF WORKFLOW
EXAMPLE OF PSEUDO-LOG REMODELING WORKFLOW §
Create Remodeled Property Log
§
Upscale Remodeled Property Log
§
Re-Run Petrophysical Modeling
Open the Scale Up Well Logs process, select the property to be edited, select the individual wells to upscale, and change settings to leave existing upscaled cells unchanged IMPORTANT!
Change these settings to avoid deleting all the “real” upscaled cells in the model!
Figure 14 - Upscaling the pseudo-property log from the remodeled property
The final step in the process is to re-run the petrophysical modeling for the entire model. Since only selected wells have been upscaled (using the remodeled property pseudo-logs), the model will be adjusted in the vicinity of the newly-upscaled wells, leaving the remainder of the model relatively unchanged. This will produce a final result which will more closely represent model properties by incorporating a combination of actual petrophysical data with pseudo-data justified with historical production data. The final model (Figure 15) shows the local modification to the property using the pseudo-log approach.
16
SPE 164420
APPLICATION OF WORKFLOW
EXAMPLE OF PSEUDO-LOG REMODELING Using the pseudo-log workflow described, the remodeled properties have been used to locally modify a portion of the model …
…it is important to remember, however, that these are pseudo-properties, and are ONLY to be used when adjusting simulation models where all other data are lacking!
FINAL POROSITY PROPERTY
Figure 15 - Final property model using the pseudo-log remodeling workflow
Conclusions Based on the assumption that completion intervals should have at least minimum reservoir properties, the workflow presented utilizes completion data to directly modify geomodel properties in zones with confirmed production. This is a very rapid and simple method which can at least temporarily eliminate the majority of initialization errors for numerical simulation. The resulting modifications can also be used to isolate areas of the geomodel for further quality control and correction, or to correct local model properties to adjust the model volumes to better match volumes indicated by historical production. However, the workflow does not actually correct problems, it simply “postpones” them until more rigorous reservoir characterization and petrophysical corrections can be made. Acknowledgement The author would like to thank Tatweer Petroleum for permission to publish and present this material, and for the suggestions and critiques from Tatweer Subsurface Department staff members for improvements to the workflow and this manuscript.
