(203e) A Generic Approach of Using Dynamic Simulation for Emission Reduction Under Abnormal Operations: Scenario Study of An Ethylene Plant Startup With Sulfur Recovery Unit

Within the last two decades, the Chemical Process Industry has been spending efforts in recognizing and improving the three impacts of industrial sustainability: environment, economy, and society. In this light, flare minimization (FM) is such a necessary approach. Flare emissions release volatile organic compounds (VOCs) and pollutants into the atmosphere, which react with nitrous oxide in sunlight to generate ozone and smog.  Therefore, it is essential to limit this amount to improve local and regional air quality, especially around chemical plant locations. Besides, environmental performance also takes into account material usage, energy consumptions, non-product outputs and pollutant releases. From the economic point of view, reducing burned-off gases saves material, helps avoiding penalty and also creates a more effective way to distribute energy, thus lowering utility cost. Ethylene production process is one of the biggest and most complicated processes in petrochemical industry. A typical startup of an ethylene plant sends approximately 400,000 - 1,000,000 lbs of materials to flare and generates 200,000 lbs of CO and VOCs. Hence, managing flare emission during startup, shutdown and malfunction of an ethylene plant are very important. Since flaring occurs in both normal and abnormal conditions, FM practices are done in either end-of-pipe approach, where flare gas is processed downstream to avoid going to burners, or flare source reduction approach, in which more upstream modification and optimization contribute in limiting unwanted products. Recently, with the gaining maturity of FGR, gasification and GTL technologies, the focus of FM is shifting from end-of-pipe to flare source reduction approach. This method, which heavily employs simulation and optimization, emphasizes the upstream section of the flare system and the whole plant integration.

Literature suggests that there lacks a proactive approach in FM under abnormal situations. In this study, rigorous steady-state and dynamic models of a front-end De-Ethanizer ethylene plant are constructed to serve as both the generic foundation to further explore causes roots of process abnormal behaviors and subsequently an optimization tool to provide guidance to both design and operational strategies. There are four major sections containing their relative unit operations: Desulphurization (Acid Gas Recovery Unit, Sulfur Recovery Unit, and Tail Gas Treating Unit), Cracking (Furnaces and Quench sections), Compression (Charged Gas Compressors), and Recovery systems (Front-end De-Ethanizer, Chilling Train and De-Methanizer, C2 Splitter, and De-Propanizer columns). The steady-state model is constructed in details in order to offer more realistic insights for dynamic simulation of process upsets. In the chilling train section, specifications of equipment sizes and heat capacities are considered for simulating pre-cooling process.  Rigorous distillation model and tray sizing calculation are also employed.  Both steady-state and dynamic models are tuned and validated before case studies are carried out. Based on the models, possible flaring sources and incidents are identified, which are demonstrated extensively in this paper throughout three case studies of CGC, chilling train, and De-Ethanizer/C2 Splitter sections during an approach of flareless startup on total recycle.

Because charged gas is continuously produced from furnaces and sent to flare to retain compressors’ inlet pressure, CGC startup is reportedly the most critical step and has largest amount of vent gas.  Several scenario cases of charged gas at different flowrates and methane and nitrogen contents are examined to find the least flaring one.  Another case study focuses in the C2-supplier columns recycling off-specs to CGC as substitute partially for furnace feed. In dynamic simulation environment, all of the major steps and options can be simulated and evaluated while flare emissions can also be calculated. The best results in these case studies are incorporated and validated in the integrated model of the whole plant. It is proven that this procedure contributes a great deal in evaluating new strategies and predicting process dynamic behaviors. It also helps achieving more precise timing to control the recycle loop; hence, flaring is potentially minimized while the start-up time is shortened.