Özet:
Hydrogen demand in refineries is constantly increasing due to effort on processing heavier crudes and producing higher grade transportation fuels in order to meet increasing demand. In this respect, refineries are investing in hydrotreating and hydroprocessing units, and are continuously investigating ways to maximize profit margins through cheaper but reliable sources of hydrogen. Currently, the most economically feasible hydrogen production route is steam methane reforming (SMR). SMR is conventionally carried out in steam reformers in which natural gas is catalytically reacted with steam in tubular reactors placed into a furnace where heat required for endothermic reaction is supplied by combustion of fuels. Among several types of reformer furnaces, Foster Wheeler’s Terrace WallTM type reformer is currently under operation in TÜPRAŞ İzmit Refinery. Due to metallurgical features of reformer tubes wall temperatures should not exceed design values as it strongly reduces expected life span of the tubes. In this respect, prediction of maximum wall temperatures, along with other process outcomes such as product composition and pressure drop under various operating conditions become important issues. For this purpose, a mathematical model, combining the combustion in the furnace and reforming within the reactor tubes, is developed and tested with field data. Simulation results show that, composition of the reformer effluent, especially those of hyrogen and methane can be well predicted with max. 4.3% and 7.2% deviation from plant data respectively. However, other components may show considerable variations most likely due to industrial kinetic parameters. Pressure drop occured in packed bed is succesfully predicted with max. 0.37% deviation. Outer tube wall temperature occuring in the system can also be closely predicted respective to location along the tube with max. 1% deviation. Due to model assumptions and literature parameters, reactor outlet temperature also shows appreciable deviation in the range of 1.9-3.3% although product composition is well predicted. Simulation results show the importance of proper parameter usage for the accuracy of the model results.
