Scholarlink

<p>Resolving molecular heterogeneity among crude oils is essential for understanding petroleum system evolution in structurally complex basins. This study applies a graph-based community detection framework to characterize molecular populations within Cretaceous reservoirs of the Abadan Plain (SW Iran) using integrated high-resolution gas chromatography (HRGC)-derived n-alkane ratios and biomarker parameters. Twenty-four crude oil samples were characterized using high-resolution gas chromatography (HRGC) and gas chromatography–mass spectrometry (GC-MS) analyses, and molecular similarities were represented as weighted correlation networks constructed through a Pearson correlation-based k-nearest neighbor approach (k = 5). Community structure was identified using the Leiden algorithm, thereby enabling detection of intrinsic molecular organization without imposing predefined cluster geometries. Independent analyses of HRGC-derived and biomarker datasets revealed distinct but complementary patterns of molecular variability. The HRGC network resolved four molecular communities, whereas the biomarker dataset identified three broader communities. Integrating both molecular domains produced a stable four-community framework that preserved key relational structures while incorporating complementary compositional and geochemical information. Sensitivity analyses demonstrated stable community assignments across graph densities (k = 3–10), with partition persistence converged to near-identical solutions for k ≥ 5. Additional validation using principal component analysis (PCA), K-means clustering, and Spearman rank correlation confirmed the robustness of the inferred molecular populations and indicated that the classification was not strongly dependent on a single similarity metric or clustering approach. Maturity-sensitive biomarker parameters, including Ts/(Ts + Tm), C29 sterane 20&#xa0;S/(20&#xa0;S+20R) and ββ/(ββ + αα) isomerization ratios, and C32 homohopane isomerization indices, show systematic differences among the identified molecular communities. These trends, together with variations in bulk molecular composition, suggest compositional heterogeneity and variable thermal evolution within the studied reservoirs. However, the observed molecular populations should be interpreted as geochemically coherent compositional groups rather than definitive evidence of discrete charging episodes. By modeling oil–oil similarity as a relational network rather than a purely distance-based problem, the proposed workflow captures complex molecular affinities and transitional relationships that may be overlooked by conventional clustering methods. The approach provides a reproducible and transferable framework for molecular classification and oil<b>–</b>oil correlation in geologically complex petroleum systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Graph-based community detection of molecular oil populations from cretaceous reservoirs of the Abadan Plain, SW Iran

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月31日

查看全文
  • Ahmad Batvandi,
  • Ali Shekarifard,
  • Golnaz Jozanikohan,
  • Asal Naseri

摘要

Resolving molecular heterogeneity among crude oils is essential for understanding petroleum system evolution in structurally complex basins. This study applies a graph-based community detection framework to characterize molecular populations within Cretaceous reservoirs of the Abadan Plain (SW Iran) using integrated high-resolution gas chromatography (HRGC)-derived n-alkane ratios and biomarker parameters. Twenty-four crude oil samples were characterized using high-resolution gas chromatography (HRGC) and gas chromatography–mass spectrometry (GC-MS) analyses, and molecular similarities were represented as weighted correlation networks constructed through a Pearson correlation-based k-nearest neighbor approach (k = 5). Community structure was identified using the Leiden algorithm, thereby enabling detection of intrinsic molecular organization without imposing predefined cluster geometries. Independent analyses of HRGC-derived and biomarker datasets revealed distinct but complementary patterns of molecular variability. The HRGC network resolved four molecular communities, whereas the biomarker dataset identified three broader communities. Integrating both molecular domains produced a stable four-community framework that preserved key relational structures while incorporating complementary compositional and geochemical information. Sensitivity analyses demonstrated stable community assignments across graph densities (k = 3–10), with partition persistence converged to near-identical solutions for k ≥ 5. Additional validation using principal component analysis (PCA), K-means clustering, and Spearman rank correlation confirmed the robustness of the inferred molecular populations and indicated that the classification was not strongly dependent on a single similarity metric or clustering approach. Maturity-sensitive biomarker parameters, including Ts/(Ts + Tm), C29 sterane 20 S/(20 S+20R) and ββ/(ββ + αα) isomerization ratios, and C32 homohopane isomerization indices, show systematic differences among the identified molecular communities. These trends, together with variations in bulk molecular composition, suggest compositional heterogeneity and variable thermal evolution within the studied reservoirs. However, the observed molecular populations should be interpreted as geochemically coherent compositional groups rather than definitive evidence of discrete charging episodes. By modeling oil–oil similarity as a relational network rather than a purely distance-based problem, the proposed workflow captures complex molecular affinities and transitional relationships that may be overlooked by conventional clustering methods. The approach provides a reproducible and transferable framework for molecular classification and oil–oil correlation in geologically complex petroleum systems.