Prediction of permeability and effective porosity values using ANN in Maleh field
This study presents the development of an intelligent system designed to predict permeability and effective porosity in wells where core samples are unavailable. An artificial neural network (ANN) was constructed with three hidden layers—comprising 15, 10, and 4 neurons, respectively—utilizing well...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Universidad Nacional Autónoma de México, Instituto de Geofísica
2025-07-01
|
| Series: | Geofísica Internacional |
| Subjects: | |
| Online Access: | https://revistagi.geofisica.unam.mx/index.php/RGI/article/view/1830 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study presents the development of an intelligent system designed to predict permeability and effective porosity in wells where core samples are unavailable. An artificial neural network (ANN) was constructed with three hidden layers—comprising 15, 10, and 4 neurons, respectively—utilizing well logging parameters (CAL, VCL, NPHI, RHOB, DT) as inputs. The ANN outputs predicted permeability and effective porosity values with remarkable accuracy. The network was optimized with a learning rate of 0.05, a momentum coefficient of 0.95, and the LOGSIG activation function, applied across layers. Input values were normalized to the range of 0 to 1, and training was performed using the sequential forward backpropagation algorithm (newcf). The training phase achieved a minimum mean square error of 0.00001 within 58 seconds over 12,000 cycles, delivering a 100% recognition rate for the training data. The ANN was tested on independent data and demonstrated exceptional performance, achieving 96% accuracy for effective porosity and 98% for permeability predictions in sandstone formations. This efficient algorithm eliminates the need for core sample analysis, reducing costs and time while improving prediction reliability, making it a valuable tool for subsurface characterization and resource exploration. |
|---|---|
| ISSN: | 0016-7169 2954-436X |