An Optimization-Based Framework for Real-Time Power System Operation: The Colombian Experience

Colombia, as in many other countries, the day-ahead dispatch relies on discrete hourly forecasts, which naturally lead to imbalances in a continuously operating system subject to real-time fluctuations. These imbalances are further amplified by forecast errors, equipment outages and the stochastic behavior of renewable sources such as wind, solar and hydro inflows.

To address these challenges, electricity markets have adopted three main strategies. The first focuses on supply flexibility, including ancillary service markets, energy storage systems and regional interconnections, enhancing system stability despite high costs and technical challenges. The second involves demand-side management, leveraging demand response mechanisms to optimize consumption, requiring automation and advanced measurement technologies. Lastly, improved forecasting and real-time markets enhance operational efficiency and reduce uncertainty, but demand sophisticated infrastructure, active market participation and significant regulatory changes.

This paper presents a novel real-time optimization framework developed collaboratively by

XM S.A. (the Colombian system and market operator) and the Universidad Tecnológica de

Pereira. The proposed solution introduces a Minimum Market Deviation Security-Constrained

Unit Commitment (MMD-SCUC) model, which performs 5-minute resolution dispatches aimed at minimizing deviations from the day-ahead schedule while ensuring compliance with operational security constraints.

The methodology employs a mixed-integer programming (MIP) approach that captures key technical characteristics of generation units, including ramping limits, minimum up/down times and complex configurations such as hydro cascades and combined-cycle plants. It integrates real-time SCADA data and network constraints using enhanced DC power flow models based on Power Transfer Distribution Factors (PTDFs) and Line Outage Distribution Factors

(LODFs). A key innovation is the development of a parameter estimation technique that improves the accuracy of the DC model, enabling more reliable dispatch decisions.

The framework also features a robust control and data management architecture capable of integrating diverse information sources—such as market bids, planned outages and real-time measurements—into a situational-awareness interface that supports operator decision-making under uncertainty.

Implementation of this mechanism in the Colombian power system has demonstrated significant improvements in operational efficiency and reliability. Simulation results show reduced manual interventions, improved alignment between scheduled and actual generation and enhanced integration of VRE and DER. Furthermore, the model supports the evolution toward more dynamic market structures, such as intraday and real-time markets, by enabling finer temporal granularity in dispatch decisions.

This contribution details the model’s mathematical formulation, implementation strategy and operational outcomes. It offers a replicable and scalable solution for power systems worldwide facing similar challenges and represents a meaningful step toward more resilient, flexible and sustainable grid operations.