Mathematical model for ethane pyrolysis in an industrial furnace

Loading...
Thumbnail Image
Date
2014-09-23
Authors
Cowperthwaite, Emily
Keyword
Ethane pyrolysis , Mathematical modeling
Abstract
Ethane pyrolysis is an important industrial process that occurs by passing ethane and steam through radiant coils (tubes) in gas-fired furnaces to produce ethylene and other light olefins. Undesirable side reactions that occur during the pyrolysis of ethane lead to the formation of coke (solid carbon) on the tube walls, which has to be periodically burnt off in decoking cycles. NOVA Chemicals is interested in developing a model that can accurately predict dynamic coke formation and associated decoking times that would help to optimize run lengths, and decrease costs. A steady-state ethane pyrolysis model of the radiant section of a floor-fired furnace was developed as a first step towards development of a dynamic coke formation model. The model includes 56 pyrolysis reactions involving 28 species, and accounts for radiant heat transfer from the furnace gas to the process gas using the Roesler flux method. The process-side model includes 29 material balances (28 reacting species plus inert steam), 1 energy balance and 1 momentum balance to track the concentration of the 29 species, the process gas temperature and the process gas pressure along the length of the reactor. These model equations are implemented in PREDICI® as an initial value problem. The furnace-side model, which includes 2 radiant flux balances and 1 energy balance, resulted in numerical problems when solved as an initial value problem in PREDICI®. Instead, the model was discretized using finite differences and simplifying assumptions. The resulting system of algebraic equations was solved in PREDICI® and then radiant fluxes were imposed on the process-side model. Preliminary studies of model responses to changes in key model inputs indicate that the model performs as physically expected, rendering this model a strong starting point for future model development.
External DOI