PI-MVS: MATHEMATICAL MODEL FOR UNCERTAINTY EVALUATION OF VIRTUAL INSTRUMENTS IN FLARE SYSTEMS AND AUDITABLE REPORTING TO THE NATIONAL PLATFORM «ECOSYSTEMA»

Abstract

Abstract. This paper presents the PI-MVS (Physics-Informed Methane Verification System) methodology for continuous monitoring of industrial flare systems aimed at reducing the risk of unburned methane emissions and ensuring an auditable digital chain “measurement → interpretation → reporting”. The approach is motivated by two factors: (i) increasing pressure for climate verification and the practical loss of “invisibility” due to satellite and remote detection of large methane plumes; (ii) the operational “blind spot” between measured gas throughput and the actual flare plume composition under wind-driven flame lift-off, over-steaming, and unstable low-flow regimes. The method integrates multispectral optical and thermal data, meteorological inputs, and process variables, applies physically constrained combustion-regime reconstruction, and produces DRE_calc as a calculated estimate of methane destruction efficiency over a time interval. A dedicated data-quality loop assigns VALID/INVALID/ALARM states with reason codes to prevent erroneous decisions under degraded inputs. The methodology includes a Decision Engine with event tagging (e.g., CH4_risk_event, soot_event, flameout_event) and generates a machine-readable Eco-Report with qualified electronic signature and cryptographic hash references to local video fragments, enabling audit-proof integrity without transferring video outside the facility. The proposed alignment with national automated emission-monitoring logic supports PI-MVS as a “trust layer” for integration into the «Ecosystema» platform.