Linking SPYRO Rigorous Stream Cracking Model

Furnace simulations are currently used in all layers of process control from planning and scheduling down to the APC layer. However in the DCS regression models are still present for process optimization. Typically the purpose of these regressions is to predict the severity on a continuous basis to allow immediate feedback to the temperature control to maintain a constant severity. Most of the time these regressions are developed when the furnace was taken in operation and are never updated afterwards. Any changes in the furnace geometry, operating condition or feed stock are not taken into account.

In this paper the regression model in DCS level was replaced with SPYRO^® furnace simulation. The furnace simulation is connected to DCS and receives real-time operating condition and feed composition. This configuration eliminates the need of having series of regression models for each feed and operating condition, as the furnace simulation can incorporate those changes and deliver more accurate advisory temperature setpoint. 

This paper summarizes the furnace optimization solution at DCS level and how it is implemented, and discusses the details of flexibility of the optimization. This is supported by field data of a cracking furnace cracking.