Reducing Air Quality Impact From Plant Start-up Emissions by Integrating Air Quality Modeling and Plant Start-up Simulation

Chemical industry concentrated regions experience highly localized and transient air pollution events.  For instance, chemical plant flaring activities, especially intensive flaring during plant startup process, emit huge amounts of volatile organic compounds (VOC) and nitrogen oxides (NO_X), causing elevated ground-level ozone concentrations that violate the National Ambient Air Quality Standards.  Emissions are inevitable during plant startups, their air quality impacts, however, are different under different meteorological conditions.  This provides opportunities to identify cost-effective control strategies for reducing air quality impacts from plant start-up emissions.     

In this paper, detailed process system knowledge of plant startup emission and speciation is integrated with air quality modeling and simulation for the first time to quantitatively study regional air quality impact from ethylene plant startup emissions.  The general process has three major steps: emission source characterization, flaring emission speciation and quantification, and air quality simulation and decision support.  In the first step, plant-wide dynamic simulation and optimization are employed to minimize start-up emissions and identify benchmark emission dynamic profiles during a plant start-up.  In the second step, simplified flaring emission speciation and quantification are conducted to identify dynamic profiles of ozone precursors, such as various VOC and NOx.  Then, the flaring precursor information is input to the air quality model, CAMx.  In the third step, multiple air quality simulations are performed for start-up emission impact analysis and decision support.  This is the most complex step.  Firstly, a background air quality simulation is performed.  The simulation needs atmospheric reaction mechanism, geological domain information, emission inventory data, meteorology information, initial conditions and boundary conditions.  The simulation gives temporal and spatial background ozone and other air pollutants distributions.  It also gives different ozone formation mechanisms for different regions.  Secondly, the startup procedures are arranged into different time windows, and start-up emissions are integrated into the background air quality model to generate different cases.  Through multiple air quality simulations, regional air pollutant increments can be obtained.  Through comparison and analysis, the best start-up time window will be identified. 

A typical ethylene plant startup procedure has been studied.  Results show that local maximum ozone increment can be significant due to an ethylene start-up emission.  However, if appropriate start-up time window is selected, the maximum ozone increment can also be substantially reduced comparing with the worst case.