A HYBRID SCENARIO-BASED PROBABILISTIC FRAMEWORK FOR LONG-TERM ELECTRICITY DEMAND FORECASTING IN PAKISTAN (2025-2060)

Abstract

Pakistan’s electricity sector faces chronic challenges, including a persistent supply demand gap, circular debt exceeding PKR 2.5 trillion, and the imperative to integrate variable renewable energy sources. Accurate long-term demand forecasting is critical for infrastructure planning; however, existing deterministic approaches consistently overestimate demand and do not account for structural changes, such as behind-the-meter (BTM) solar adoption and efficiency improvement. This study presents a hybrid scenario-based probabilistic forecasting framework that combines ensemble machine learning with Monte Carlo simulations to generate sector-wise electricity demand projections for Pakistan from 2025 to 2060. The framework decomposes demand into five sectors (residential, commercial, industrial, agricultural, and other) and evaluates six policy-relevant scenarios. Uncertainty was quantified using 2,000 Monte Carlo draws propagating parameter uncertainty from GDP growth, population dynamics, climate sensitivity, and model residuals. Under the Business-as-Usual (BAU) scenario, the P50 demand reaches 185 TWh by 2030 (P10-P90:165-210 TWh), 295 TWh by 2040, 420 TWh by 2050, and 580 TWh by 2060. The High Solar/BTM scenario reduces grid demand by 15-22% relative to BAU by 2050, whereas the High Heat scenario increases peak demand by 12-18%. These probabilistic projections provide actionable inputs for Pakistan’s Indicative Generation Capacity Expansion Plan (IGCEP), NEPRA regulatory planning, and Nationally Determined Contribution (NDC) commitments under the Paris Agreement.