Centralized AI-Enabled Energy Performance Management Over Enterprise IIoT Platform

Abstract

Abstract Energy is one of the largest controllable operating costs and a primary source of greenhouse-gas emissions in upstream oil and gas operations. Yet in most large assets, energy information remains fragmented across electrical meters, supervisory control systems, historians, spreadsheets, and monthly reports. This fragmentation delays optimization actions, obscures emissions performance, and prevents systematic energy governance at scale. This paper presents a large-scale, centralized, Artificial-Intelligence (AI) enabled Energy Performance Management and Reporting System built on an enterprise Industrial Internet of Things (IIoT) platform for upstream field operations. The system integrates field-to-enterprise energy data, monitors electrical and fuel-gas consumption, computes Energy Performance Indicators (EnPIs), generates AI-assisted targets, detects anomalies, and delivers role-based dashboards and automated reporting, all designed for scalable deployment across assets and major energy users. The architecture contextualizes distributed energy-meter, supervisory-control, historian, operator-round, and control-system data into asset, system, and equipment hierarchies, and converts it into EnPIs, baselines, targets, and alerts. A resilient connectivity backbone, combining a private 5G operational-technology network with a hybrid low-power wide-area network (LPWAN) sensor layer, provides the field-to-enterprise data path, enabling high-bandwidth telemetry from major facilities alongside low-power monitoring of previously unmetered and remote energy users. Machine-learning models support target generation, deviation detection, forecasting, and abnormal-consumption analysis, while configurable dashboards serve management, operations, energy teams, and functional specialists to enable timely corrective action and governance. At asset scale, the approach supports monitoring of major electrical users, fuel-gas consumers, water-injection systems, pumps, compressors, fired heaters, gas turbines, production facilities, and accommodation loads. It enables real-time EnPIs such as energy per barrel of oil, power per barrel, fuel gas per barrel, energy per unit of injected water, equipment and thermal efficiency, greenhouse-gas indicators, and daily operating cost. AI-assisted targets and anomaly detection allow users to compare actual against expected performance, identify inefficient operating patterns, and prioritize corrective actions. The paper concludes that a centralized AI-enabled energy performance management platform can improve energy-efficiency governance, reduce reporting delays, enhance emissions visibility, support cost avoidance, and provide a scalable model for enterprise-wide energy optimization. The contribution is a practical field-to-enterprise architecture that converts fragmented energy and process data into actionable intelligence for efficiency, emissions reduction, and operational excellence.