Finite difference modelling and 2D model building can be found under the "Modeling/Survey Planning" button on the lower right side of the main window.
These processing options are not allowed by default, and will need to be activated first.
The current version only supports data with metric depth.
The available Model building option is for a flat model only.
9.1.1 Input datasets and Pre-Processing
The following data is required.
- Open hole log data that includes VP (compressional velocity) and RHOB curves, indexed in TVD_Datum.
- If VP is not available, this can be easily functioned from a DT curve using the "Sonic-to-Velocity" process.
The following criteria apply
- Only metric data is supported, for FD modelling.
- All the open hole log data should be in a single file. This can be done by using merging and splicing as required.
- Null values are not tolerated in either of the DT or RHOB curves, and Null values should be removed as discussed in sections 4.5 and 4.6.
The data must be indexed in TVD-Datum. It is easy to toggle between TVD and TVD_Datum, if the input data was in TVD.
If the data is in Measured Depth (Depth or MD), then the proper way to get it to TVD is to apply Deviation Survey processing, as discussed in Section 8. If a deviation survey is not available and the data is from a vertical well, then a quick (and nasty) way to get it to TVD index is to save the dataset to a LAS file, open the LAS file with a text editor and rename the Depth curve to "TVD" in the Curve Table, then reload the LAS file. Check the APD parameter is correctly filled and toggle to TVD_Datum.
Ideally the data should be available all the way to the surface. If this is not the case, then the model building process will automatically extrapolate to the surface using water velocity and water density.
Logs can be manually extrapolated using the "Extend Data Range" process. This will extend all logs; the extended depth interval will be padded with Null Values. Next goto the data editor and manually edit the extended sections using the curve editor (see Section 4).
9.1.2 Building a Flat 2D Model
The 2D model for finite difference modelling is a regular grid model. We need to specify the horizontal and vertical extent of the model, and cell size.
Accurate FDM is a tradeoff between cellsize and processing time. A large cell size will give poor results, while a small cell will take a lot of processing time. In addition, large cell size will only suport a very low frequency wavelet. The default cell size is 5 metres, which is about as large as what is practically useful for a 40 to 50 Hz wavelet. A 10 metre cell size could be used for a test run that will be quick to process, but results wont be too good.
The X and Z extent are selected to keep the model boundarys (which can also cause reflections) to be sufficiently far away. But keeping them close enough to minimize processing time.
Model building procedure:
- Pre-processing on the open hole log data, as discussed above in section 9.1.1.
- Goto the Modelling menu and select "Make 2D Log Grid". Choose the dataset and the VP adn RHOB log curves.
- Select the model size and cell size and click Compute. The Log Grids for each curve can now be seen in the datalist
- To view the results, right click in the Log Plot windows and select "Add Log Curve/2D Grid". Only one log-grid can be plotted. Plotting a second log-grid will cause the first one to be removed.
The 2D Log-grids are 2 axis-arrays, so if they are to be saved to a disk file, then need to save as a JSON file, since LAS or CSV do not support array data.
9.1.3 Add Source- Receiver location
- Select Make Source-Receiver Arrays.
- Fill the parameters and click Compute.
The results can now be seen in the data-list. Note that these channels are 2 element arrays, so if they are to be saved to a disk file, then need to save as a JSON file, since LAS or CSV do nor support array data.
Two curves corresponding to the X and Z values can be seen in the Quickview window. To see the numerical values, hold down the Alt key down before clicking on the curve name in the data-list. Or save the dataset to a JSON file, and view the resutls with text editor.
9.1.4 Finite Difference Modelling.
Only Acoustic Finite Difference modelling is currently available. Select this processing option, choose the Log Grid and Source/Receiver array data, set the parameters parameters and select compute.
A 5 metre cell model may take between 15-60 seconds depending on the model extent and CPU power. A smaller model with larger cell size could be done as a test to see how it works, before re-running with a better model.
Two results are returned:
- a Synthetic VSP dataset, with waveforms at each receiver depth. Acoustic FDM returns a Scalar wavefield, so is a single component.
- Snapshots of the expanding wavefield. The snapshots are added together to give a single image. The snapshots can be found in the same folder as the 2D log Grids.
The 2D flat model
Synthetic VSP data are FDM
Stacked Snapshots after FDM