Electrofacies clustering and classification from the Ariri Formation in Santos Basin (southeastern offshore Brazil) involving unsupervised learning algorithms

Abstract

During oil well drilling, geological knowledge of the layers to be drilled is essential to dimension the drilling parameters and plan contingency operations in case of operational failures. The Ariri Formation, lying in the Santos Basin (southeastern Brazil), can reach a thickness exceeding 3000 m and consists of salts with a complex rheology, requiring some effort to understand these minerals’ vertical and lateral distributions. This study aims to classify the electrofacies of evaporitic sequences in a semi-automated way based on unsupervised algorithms applied to geophysical well-logs and drilling parameters. The results were validated based on the previous classifications performed by interpreters, in which six types of saline minerals predominate in the wells under study: halite, anhydrite, tachyhydrite, carnallite, sylvite, and sylvinite. The database consisted of a set of fourteen wells located in the offshore portion of the Santos Basin. Unsupervised analyses are developed using the multilayer perceptron with lateral connections (MPLC), k-means, and Self-Organising Maps (SOM) algorithms. The obtained clusters are classified according to their mineralogical composition and drilling resistance. The SOM and MPLC algorithms provide the best accuracy in segmenting the main evaporite groups and highlighting possible mixtures between them. In terms of facies, the clustering provides electrofacies with different levels of drilling resistance for the same mineral. The selection of the best grouping enables a detailed subdivision for the Ariri Formation, which will serve as a basis for future stratigraphic studies in the distal setting of the Santos Basin.

Appendix V: Summary table of the results of the three algorithms with the pros and cons of each application

MPLC 6 Clusters		MPLC 12 Clusters		MPLC Pros	MPLC Cons
Electrofacies	Accuracies (%)	Electrofacies	Accuracies (%)
AND(TAC)-H	Well-1: 91.9	AND(CRN)-M	Well-1: 89.4	Good segmentation of the main evaporites	Segmentation of only one potassium mineral (CRN)
AND-H	Well-2: 89.6	AND-H	Well-2: 88.4
CRN-L	Well-3: 92.2	HAL(AND)-M	Well-3: 90.7
AND-M	Well-4: 84.7	HAL-H	Well-4: 81.6	Accuracy greater than null accuracy in all wells	Less than 70% precision in TAC classification
TAC-L	Well-5: 85.9	CRN-L	Well-5: 89.7
HAL-M	Well-6: 90.8	AND-H	Well-6: 87.8
	Well-7: 82.9	TAC-H	Well-7: 85.4	Segmentation of HAL and AND with different drilling resistances
	Well-8: 88.3	CRN-L	Well-8: 85.3
	Well-9: 92.4	AND-M	Well-9: 91.9
	Well-10: 85.5	HAL-H	Well-10: 91.6	Lowest DBI and AWCD indexes
	Well-11: 86.3	TAC-L	Well-11: 88.3
	Well-12: 90.9	HAL-L	Well-12: 89.5
	Well-13: 89.7		Well-13: 89.6	High correspondence with the previously described lithologies
	Global average: 88.5		Global average: 88.4

K-means 6 clusters		K-means 12 clusters		K-means Pros	K-means Cons
Electrofacies	Accuracies (%)	Electrofacies	Accuracies (%)
HAL-M	Well-1: 80.9	HAL(CRN)-M	Well-1: 84.2	Segmentation of 3 potassium minerals	Did not segment AND clusters
CRN(HAL)-M	Well-2: 80.2	HAL-M	Well-2: 89.2
CRN-L	Well-3: 69.5	CRN-L	Well-3: 70.1
SVN(CRN)-L	Well-4: 77.1	SVI(HAL)-L	Well-4: 65	Segmentation of potassium mineral mixtures	Accuracy less than null accuracy in two wells
CRN-L	Well-5: 87.9	SVN-L	Well-5: 88.2
CRN-L	Well-6: 80	SVN(CRN)-L	Well-6: 86.1
	Well-7: 68.4	SVN(AND)-L	Well-7: 65.2	More than 70% classification precision for all minerals in the 12 clusters classification	Highest DBI and AWCD indexes
	Well-8: 70.2	TAC-L	Well-8: 86.3
	Well-9: 63.1	CRN-L	Well-9: 73.1
	Well-10: 62	CRN-L	Well-10: 60		Method most influenced by outliers
	Well-11: 78.8	CRN-L	Well-11: 88.3
	Well-12: 80.8	CRN-L	Well-12: 81.4
	Well-13: 79		Well-13: 87.1		Least correspondence with the previously described lithologies
	Global average: 75.2		Global average: 78.8

SOM 6 Clusters		SOM 12 Clusters		SOM Pros	SOM Cons
Electrofacies	Accuracies (%)	Electrofacies	Accuracies (%)
AND-H	Well-1: 85.6	AND-H	Well-1: 88.1	Good segmentation of the main evaporites	Segmentation of only one potassium mineral (CRN)
HAL(TAC)-M	Well-2: 85.6	AND(TAC)-H	Well-2: 89.2
HAL-L	Well-3: 90.6	AND-M	Well-3: 91.5
HAL-M	Well-4: 88.7	HAL-L	Well-4: 91.8	Accuracy greater than null accuracy in all wells	Classification in two steps demanding more computational time
CRN-L	Well-5: 87.9	HAL-L	Well-5: 90.3
TAC-L	Well-6: 83.9	HAL-M	Well-6: 87.5
	Well-7: 84.6	HAL-M	Well-7: 85.8	Classification precision of more than 80% on the main evaporites
	Well-8: 83.6	HAL-M	Well-8: 85.7
	Well-9: 81.6	CRN-L	Well-9: 91.9
	Well-10: 91	CRN-L	Well-10: 92.2	Segmentation of HAL with different drilling resistances
	Well-11: 85.4	CRN-L	Well-11: 89.3
	Well-12: 85.2	TAC-L	Well-12: 91.3
	Well-13: 74.2		Well-13: 82.3	Highest correspondence with the previously described lithologies
	Global average: 85.2		Global average: 89.0