The power distribution sector is undergoing a structural transformation as utilities seek to balance efficiency, reliability and long-term sustainability. Prudhvitej Immadi, IAS, Chairman and Managing Director, Andhra Pradesh Eastern Power Distribution Company Limited (APEPDCL) highlights how the utility is leveraging technology and data-driven decision-making to strengthen power procurement, enhance network reliability and optimise costs, while remaining aligned with its sustainability objectives. Excerpts…
AI/ML-based demand forecasting
APEPDCL is using artificial intelligence (AI) and machine learning (ML) to optimise power purchase costs by improving demand forecasting. The utility has co-developed two tools, one for short-term forecasting and power procurement and another for long-term forecasting and resource adequacy planning up to 2036. In the short term, the energy portfolio management system (EPMS) supports real-time, day-ahead and week-ahead forecasting of up to 10-14 days. These forecasts are generated using past demand patterns, live load data and weather inputs collected from 274 weather stations across the state. Based on these forecasts, EPMS performs daily unit commitment to optimise power procurement by backing down high-cost generators when lower market prices are expected. The platform has been integrated with the Indian Energy Exchange, enabling direct bid submission in the day-ahead market. It also supports optimal bilateral trading through visibility of contracts on the Discovery of Efficient Electricity Price and Term-Ahead Market portals and power exchanges with states such as Haryana and Uttar Pradesh to balance deficits and surpluses. Over 155 days of operation, forecast errors have generally remained around 2 per cent, except during extreme weather events such as cyclones. Regulatory reporting requirements for short-term power purchases are also being built into the system.
For long-term planning, APEPDCL has developed an AI-based tool along with REInt AI, which is a custom-made alternative to PLEXOS. The model uses 10 years of historical demand, GDP, sectoral gross value added, population and weather data, among other inputs, to forecast demand and optimise capacity over the next decade. It aims to reduce reliance on costly short-term market purchases, which currently account for 10-15 per cent of demand. Projections indicate that without new capacity addition, cumulative deficits could reach 48,000 MUs by FY2036. The optimal mix over the next five years prioritises additional solar, wind and battery energy storage, followed by thermal capacity after FY2030 to meet baseload requirements. Based on this mix, power purchase costs are expected to reduce by about 14 per cent by FY2030.
Digitalisation and advanced analytics
APEPDCL has undertaken major digital initiatives to improve network visibility and operational efficiency. A key initiative is the integration of geographic information system (GIS) with systems, applications and products in data processing (SAP) to create a digital twin of the power distribution network. Earlier, project estimation and execution relied on manual sketches, tape measurements and manual SAP entries, which led to inaccurate material estimates, poor visibility of network layouts, data duplication, delays and audit challenges.
To address this, APEPDCL’s in-house IT team developed and implemented a fully integrated GIS-SAP workflow. Field officers now conduct GPS-based GIS surveys using mobile applications to capture pole locations, line routes and material requirements. This data flows automatically into SAP, where estimates are generated without manual intervention. After work completion, a mandatory post-execution GIS survey captures exact asset locations and billing is allowed only after verification. A web-based GIS platform enables officers to visually compare planned and executed layouts, improving traceability, accountability and cost control. Despite challenges related to system integration, data accuracy, connectivity in remote areas and user adoption, APEPDCL has achieved 100 per cent GIS-based infrastructure monitoring, reduced project turnaround time and strengthened governance. Apart from this, APEPDCL has established a data analytics unit to address revenue leakage and operational inefficiencies. The unit has analysed 18 months of billing and meter data to identify abnormal consumption patterns, including consumers with consistently narrow usage, repeated zero-consumption readings under incorrect meter statuses and frequent meter status changes. These cases are flagged for field inspection. With the roll-out of smart meters, APEPDCL is now using advanced analytics to monitor consumer behaviour, power quality, feeder and transformer performance, reliability indices such as SAIDI (system average interruption duration index) and SAIFI (system average interruption frequency index), remote disconnection and reconnection, and exception handling.
