OCR Text |
Show Computational Fluid Dynamic (CFO) modelling (13) is only just beginning to be used to solve real problems and it looks set to have an excellent future eventually replacing virtually all physical modelling. At present however, its application is difficult and time consuming, especially the setting up of the field networks, and many applications are yet to be valida ted (14), furthermore a t present the technique is not capable of showing a transient unstable condition such as shown in figures 3 and 4, in the way physical modelling can (15). FCT therefore uses the technique principally where physical modelling is impractical, or too expensive, such as two phase flow, or plume dispersal. Major improvements to a flash calciner worth thousands of dollars a day to the user have followed from a combined physical and CFD modelling exercise on such a unit. The technique has also been used for predicting plume dispersal, in one case investigating the potential effect of a flare plume (16) on the stability of light aircraft on the approach to an airfield and another case looking at the risks associated with a very cold stack exhaust (17), (-390 0 F). FCT has applied these techniques to a large number of real industrial problems and in the great majority of cases, the results of the modelling work have been fully implemented on the plant. Examples of these applications are given in table 2, some are described in more detail below. 3. CASE HISTORIES 3.1 Upgrading of a Petrochemical Heater System In 1986 Conoco came to FCT with a problem in a fired heater where certain tubes were suffering internal coking. This limited the unit output and required the unit to be shut down frequently for the coke to be cleaned from the tubes, an operation costing tens of thousands of dollars. FCT identified the problem, as an aerodynamically driven heat transfer problem. Poor air distribution between the burners, figure 5, meant that several of the burners were operating sub-stoichiometrically and producing large volumes of carbon monoxide. To mi tiga te this problem required operation at high levels of excess air in the furnace. This resulted in very short flames on many of the burners. Mathematical modelling of the heat flux profile revealed a high peak heat flux where the excessive coking occurred, figure 6. Mathematical modelling of the burners operating at low excess air, figure 7 showed that the peak heat flux could be reduced and moved further up the heater and hence reduce the coking. Lower excess air would also have the benefit of increased efficiency and hence reduced energy consumption and lower emissions. The solution, conversion of the burners to run at low excess air with longer flames, required resolution of the airflow distribution problem. This was achieved with water /bead and air modelling, figure 8. The results were spectacular, an end to the problem of excessive coking in a few tubes together with an increase in output of 3-5%. The savings were worth hundreds of thousands of dollars a year, all for a small investment in modelling and minor changes to the burners and air ducting. Without the modelling, nothing would have been achieved! |