PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 12, 2026Energies0 citationsOpen Access

Sectoral Forecasting of Natural Gas Consumption in Colombia: A Structural and Seasonal Analysis Using Holt–Winters Models

APAlexander D. Pulido-RojanoNCNeyfe Sablón CossíoAVArnaldo Verdeza

Key Points

  • The central aim is to analyze sectoral natural gas consumption in Colombia to improve forecasting models.
  • Applied Holt-Winters exponential smoothing models (additive and multiplicative)
  • Analyzed monthly data from Colombian Mercantile Exchange (April 2020 - July 2025)
  • Optimized model configurations by minimizing forecasting errors (MAE, MAPE, MSE)
  • Natural gas consumption patterns are not uniform across sectors, showing distinct seasonal dynamics.
  • The Thermal Generation sector was best modeled with a multiplicative approach, reflecting varying demand.
  • Other sectors (Industrial, Residential, GNVC, Commercial, SNT Compressor Stations) performed better with an additive model.
  • Petrochemical and Refinery sectors exhibited short-term cyclical behavior, with accuracy varying by metric.

Abstract

This study examines the sectoral dynamics of natural gas consumption in Colombia by applying additive and multiplicative Holt–Winters exponential smoothing models. The analysis covers the main demand segments (Thermal Generation, Industrial, Residential, Refinery, Compressed Natural Gas for Vehicles (GNVC), Commercial, Petrochemical, and SNT Compressor Stations) using official monthly data from the Colombian Mercantile Exchange for the period April 2020 to July 2025. Model configurations were optimized by minimizing the Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Mean Squared Error (MSE) to identify the most appropriate structure for each sector. The results confirm that natural gas consumption in Colombia does not follow a uniform seasonal pattern. Instead, each segment exhibits distinct dynamics shaped by operational conditions, production schedules, mobility-related behavior, or logistical planning. The Thermal Generation sector was best represented by the multiplicative model, reflecting proportional variability associated with electricity dispatch and system-level operational changes. In contrast, the Industrial, Residential, GNVC, Commercial, and SNT Compressor Stations sectors showed superior performance under the additive model, consistent with relatively stable or constant-magnitude seasonal effects. The Petrochemical and Refinery sectors displayed short-term cyclical behavior, with model accuracy depending on the performance metric prioritized. These findings demonstrate that energy forecasting must incorporate the structural heterogeneity of demand systems rather than treating natural gas consumption as a homogeneous aggregate. Practically, the results provide insights for improving supply planning, contract allocation, and regulatory segmentation. The study also offers a replicable methodological basis for forecasting in emerging economies characterized by diverse consumption profiles.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Pulido-Rojano et al. (2026) studied this question.

synapsesocial.com/papers/698d6d9f5be6419ac0d52aa9https://doi.org/10.3390/en19040915
Ask AI
Helpful
Bookmark
Share
View Full Paper