Application of Log Lithofacies Classification Model Based on Clustering-Support Vector Classification Method
The carbonate reservoir in study area A of the Middle East has good reservoir quality. However, in this area, the pore types are numerous, the pore structure is complex, and the heterogeneity difference between wells is really strong. So it is difficult to complete the lithofacies identification and...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Office of Well Logging Technology
2023-04-01
|
| Series: | Cejing jishu |
| Subjects: | |
| Online Access: | https://www.cnpcwlt.com/#/digest?ArticleID=5467 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The carbonate reservoir in study area A of the Middle East has good reservoir quality. However, in this area, the pore types are numerous, the pore structure is complex, and the heterogeneity difference between wells is really strong. So it is difficult to complete the lithofacies identification and physical property evaluation accuratly in this area. At present, the lithofacies division of wells in this area is mainly based on tiny amount drilling coring data, which costs too much and has no ability to completely reflect the vertical and horizontal change information of the stratum. It is also difficult to identify the lithofacies and establish an accurate porosity and permeability model in this area by using conventional log identification methods. In order to solve those problems, the wavelet transform method is used to complete the division of different lithofacies stratigraphic boundaries according to the morphological characteristics of logging curves. The hierarchical clustering and K-means clustering methods are used to cluster the logging series corresponding to each small layer, and clustering results are rectified by the observation results of coring. Finally, based on the constrained clustering results and lithofacies classification labels in the study interval, the mapping model between them is established under the supervised learning of support vector machine. By applying the model in the coring well, the verification result shows a match of up to 92% with the core recognition. The application of the method in other wells shows a significant inprovement on the correlation between the permeability and porosity of the same type of lithofacies in the rock identification results, which indicates a good effect of the model in lithofacies identification. This can provide theoretical basis and data support for accurate rock types division and physical parameters calculation. |
|---|---|
| ISSN: | 1004-1338 |