Reduction of environmental risks during open storage of oil and condensate through automatic evaporation control systems and optimization of ventilation modes in tank farms

Abstract

The article substantiates and empirically confirms the effectiveness of a comprehensive automated system for reducing hydrocarbon losses and environmental risks in the open storage of oil and stabilized condensate at tank farms, shifting vapor management from static protective means to dynamic, real-time predictive control; the aim is to reduce VOC emissions and economic losses through the synergy of continuous monitoring, physico-mathematical modeling of large/small breathing and optimization of ventilation regimes and VRU operation. Materials comprise 24-month field trials at a tank farm of ten RVS-20000 equipped with a wireless sensor network (optical gas analyzers, pressure, product and gas temperature profiles, shell temperature, a meteorological station with insolation) at a 5-minute interval; methodologically, an evaporation model on instantaneous parameters was developed (convergence 92-95%) alongside an MPC module that controls ventilation/VRU throughput and, where necessary, the fill-withdrawal sequence, with analytics in STATISTICA and Python. Compared with traditional operation, the system reduces the mean annual vapor concentration in the gas volume by 38,70%, peak maxima by 54,00%, and process variability by 64.00%, thereby suppressing breakthrough emission episodes; adaptive effectiveness increases as weather conditions worsen: under low/medium/high insolation, losses decrease by 28,50%/41.00%/58.00%, respectively. By operating mode, key effects were achieved: small breathing -35,50%, large breathing during filling -61,80%, during withdrawal near-complete suppression of crossflows from adjacent tanks (-98,20%), confirming preemptive VRU operation based on forecasts. The economic assessment for a single RVS-20000 indicates average annual avoided losses of 62,35 t and a net benefit of $121,55 thousand/year with a 1,85-year payback; additional gains are expected from reduced environmental fees and penalty risks. It is concluded that dynamic intelligence fundamentally outperforms static measures, is practically ready for integration with MES/ERP, and merits algorithmic expansion (self-learning, end-to-end optimization) to scale the technology and standardize industry requirements.