There are many steps in a 4 -D feasibility study before reaching a final conclusion. One of the most important steps is the<br>prediction of the impedance and velocity variations due to changes in reservoir conditions(pore fluid saturation, pressure,<br>temperature), using Biot’s theory to make a fluid substitution. When dealing with old wells with a “limited” logs suites the lack of<br>shear wave velocity logs(VS) is a very common situation, a problem arises when there are no compressional wave velocity<br>logs(VP) available, or VP data is not reliable. There are many methods of generating VS from VP data, and the applications of the<br>knowledge of VS are many for example: borehole stability and DHI analysis. We chose 8 wells of Brazilian offshore reservoir<br>covering an area of 60 km2 of extent and simulated a fluid substitution study using sonic logs generated by artificial neural<br>networks and compared the impedance changes predicted by them to the impedance changes predicted by the fluid substitution<br>applied to the original curves. The pseudo-sonic well logs match very well the original ones for all the wells and the core velocity<br>measurements available. The core measurements were used first to calibrate and validate the training well and then, for evaluating<br>the quality of pseudo-sonic logs. The porosity and density logs were used as input curves for the artificial neural networks. The<br>examples presented in this paper shows that a fluid substitution study can be carried out on pseudo-sonic logs generated by<br>neural networks using density and porosity logs, although they do point out some critical points like geographic extent of training<br>data, amount of input parameters, etc. The particular features of each reservoir must be evaluated and analyzed before trying to<br>apply neural networks as a log generator for a 4 -D project, in this sense laboratory velocity measurements in cores are one of<br>the most important data in evaluation and calibration of the sonic, logs before fluid substitution and after that, in order to<br>evaluate the pseudo-sonic logs.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error