Cluster-based Geospatial Optimization of Gas Flaring Sites in the Niger Delta for Enhanced Gas Recovery
Abinye Chimdia Nwankwo
*
University of Port Harcourt, Choba, Port Harcourt, Nigeria.
Bourdillon Odianonsen Omijeh
Department of Electronic & Computer Engineering, University of Port Harcourt, Choba, Port Harcourt, Nigeria.
Matthew Ehikhamenle
University of Port Harcourt, Choba, Port Harcourt, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
Aims: The Niger Delta experiences high levels of routine gas flaring. This leads to wasted associated gas, economic losses, and contributes to environmental degradation. The fragmented spatial distribution of flare sites further complicates gas capture and infrastructure planning. This study aims to apply a geospatial cluster-based optimization approach to group twenty-four (24) onshore gas-flaring sites in the Niger Delta. The objective is to improve flare-gas recovery potential and guide the design of centralized gas-gathering infrastructure.
Study Design: It follows a quantitative geospatial clustering using Python-based K-means analysis, supported by internal cluster-validation metrics.
Place and Duration of Study: The study was carried out using flare-volume records and GPS data from 24 onshore flowstations across the Niger Delta, covering approximately 100 days of operational reporting between July to December 2024.
Methodology: Daily flare-volume datasets were pre-processed to compute average active-day flare rates for each flowstation. Latitude and longitude coordinates were compiled using Google Earth Pro and field entries. The Elbow Method was used to determine the optimal number of clusters (K) based on inertia values. K-means clustering was then applied to group the flowstations into distinct geospatial clusters. Internal cluster quality was evaluated using the silhouette coefficient. Aggregated flare volumes were computed for each cluster to assess recovery potential.
Results: The Elbow Method identified four clusters as the optimal configuration. K-means clustering produced coherent spatial groupings reflecting natural geographic alignments within the region. Cluster 0 recorded the highest aggregated average flare volume (≈ 45.99 mmscf/day), followed by Cluster 3 (≈ 30.22 mmscf/day), Cluster 1 (≈ 24.56 mmscf/day), and Cluster 2 (≈ 22.55 mmscf/day). Silhouette analysis confirmed strong internal cohesion and clear separation between clusters, with no misclassified points. Together, Clusters 0 and 3 accounted for approximately 63% of total aggregated flare volume; this indicates priority zones for possible centralized gas-gathering development.
Conclusion: Geospatial clustering provides a robust foundation for designing shared-infrastructure flare-gas recovery systems in the Niger Delta. The four-cluster model presented two priority hubs suitable for centralized infrastructure development. This can reduce total pipeline distance. The cluster model forms a baseline for subsequent techno-economic feasibility studies.
Keywords: Gas flaring, geospatial clustering, K-means algorithm, elbow method, silhouette coefficient