DEPLOYMENT OF DIGITAL MODELLING FOR CRACKING FURNACE ONLINE YIELD OPTIMIZATION

摘要

Olefins plants are at the heart of the petrochemicals value chain.They convert gas and liquid feedstocks such as ethane and naphtha to ethylene,propylene and butylene.These products are then further processed into a wide range of valuable performance chemicals such as polymers,resins,esters,glycols and solvents.This conversion is performed at high temperature in thermal cracking furnaces.During the cracking process,coke is deposited in the furnace coils and transfer line exchanger.An olefins plant is the single most energy-consuming process in the chemical industry.Modern plants have capacities up to 1.5 million tonnes/year ethylene and at this scale,small improvements in efficiency can yield significant annual savings.Olefin plant operation is thus a prime candidate for optimisation aimed at maximising economic performance.There are several challenges that operators face regarding cracking furnaces (a) setting the desired yield for each furnace based on overall optimality,(b) understanding the extent of coking in the furnace which has a direct effect on residence time and tube metal temperatures,(c) accurately predicting ethylene yields-a key measure of plant performance and an important variable for severity control.Another over-riding challenge is to implement improvements as part of a digital strategy,where the information contained within plant data is fully exploited by a model and the calculated results can be shared with other applications.This paper describes how SABIC has implemented such a digitally-integrated system.The model calculates the key cracking furnace variables including conversion,product yields and degree of coking.The coking calculation uses historical plant data and a state-estimation technique to provide significantly better estimates for plant performance indicators than previous approaches.The data and calculated results are also available for other applications as needed.