Abstract

This paper proposes a unified framework for evaluating the thermodynamic behavior of gas turbine components under variable operating and environmental conditions. The framework combines first- and second-law analyses, high-fidelity and reduced-order models, and rigorous uncertainty quantification to capture steady, transient, and off-design regimes. It treats the compressor, combustor, turbine, recuperator or heat-recovery elements, and secondary air systems as coupled modules linked through mass, momentum, energy, and exergy balances. Component maps, validated chemistry, and heat-transfer correlations are fused with data-driven surrogates to represent nonlinearities due to ambient temperature, humidity, altitude, load profile, fuel composition, degradation, and control schedules. State estimation using extended Kalman and particle filters enables digital-twin deployment and reconciliation of sensor noise, while Bayesian calibration aligns model parameters with test data. The methodological core is an exergy-aware assessment that decomposes losses at module and interface levels to expose where irreversibilities originate and how they shift with conditions. Global and local sensitivity measures (Sobol indices and adjoint gradients) rank drivers across envelopes, and polynomial chaos or Monte Carlo propagation quantifies prediction intervals for key metrics. Multi-objective optimization (e.g., NSGA-II) explores trade-offs among heat rate, surge margin, cooling effectiveness, lifecycle emissions, and maintenance risk, with constraints enforced by surge and metal-temperature limits. A modular API and standardized experiment design promote reproducibility and integration with existing plant historians and control simulators. Two use cases illustrate the framework. First, ambient and load variability are propagated to compressor stability, turbine cooling demand, and combined-cycle performance, revealing how control schedules mitigate off-design penalties. Second, fuel variability, including hydrogen-enriched and synthetic fuels, is assessed for flame temperature, pattern factor, and NOx formation, clarifying feasible operating windows under emissions constraints. Validation protocols compare against standardized test points, quantify measurement and model uncertainty, and establish traceable uncertainty budgets for decision support. By unifying exergy-based metrics, surrogate modeling, probabilistic analysis, and digital-twin assimilation in an architecture, the framework provides a reproducible basis for evaluating component-level thermodynamics under realistic variability. It enables faster what-if assessment, clearer attribution of losses, and more robust decisions on control scheduling, maintenance planning, fuel flexibility, and retrofit prioritization for aero-derivative and industrial gas turbine fleets.