Stress-Aware Maintenance Intelligence System (SAMIS) of Synchronous Generators

To address these challenges, this paper presents the Stress-Aware Maintenance Intelligence System

(SAMIS), an analytical framework implemented as an overlay on existing utility-wide SCADA and centralized data platforms. SAMIS integrates physics-based electro-thermal digital twins with operational stress tracking, fuzzy-logic-based consistency assessment, and structured human observations. The objective is to transform raw operational data into stress-aware, decision-relevant indicators that support both maintenance prioritization and informed operation without introducing additional sensors or interfering with protection and control systems.

The proposed methodology is based on a parallel, pillar-oriented architecture. The electro-thermal reference pillar provides a physically consistent baseline for stator and rotor thermal behavior using

SCADA-derived electrical and cooling data, including virtual sensing of rotor excitation current and temperature where direct measurements are unavailable. The operational stress pillar quantifies cumulative exposure to ageing mechanisms such as cycling, excitation forcing, and operation near capability limits. A consistency and plausibility pillar employs fuzzy logic to detect physically implausible combinations of electrical and thermal variables, enabling early identification of latent defects or measurement inconsistencies. In parallel, a structured human observation pillar incorporates qualitative maintenance insights in a standardized and reproducible manner.

A key feature of SAMIS is the strict separation between maintenance-oriented decision support and adaptive operational constraints. Stress-aware corrections are applied to advisory operating envelopes within the generator P–Q capability diagram, reflecting the actual thermal margin of the machine rather than conservative static limits, while preserving existing protection and control hierarchies.

Engineering validation is performed through scenario-based analysis using real operational data from nominally identical hydrogenerators. The results demonstrate that SAMIS detects latent stress conditions well before classical alarms are triggered, distinguishes true physical degradation from measurement bias, and identifies reduced thermal margins even when all monitored quantities remain within IEC limits. Scenarios involving excitation current deviation, inductive capability narrowing, measurement inconsistency correction, and localized cooling degradation confirm that stress accumulation, rather than absolute limit exceedance, is the dominant ageing driver under modern operating regimes.

The presented results show that SAMIS provides a transparent, physically grounded, and scalable approach to stress-aware asset management, enabling earlier maintenance prioritization and more sustainable exploitation of synchronous generator flexibility in power systems with high renewable penetration.