OCR Text |
Show ~-r-----------------------------------------------, 70- 60- 40- 30- • 'l.GA'.2O'I.~ • ~~------------------~I-----------------'I -------~ sec 1000 NO" ppm Fig 4 DeNOx EHiciency vs Initial NOx Concentration and Residence Time in Reburning Zone (50 KWt) PHYS ICAL k~ D MATHL~TI CAL MODELING OF THE REBURNING PROCESS In order to better understand some of the fundamental aspects that govern the reburning process , act ivities have been conducted in the field of pnysical and mathematical modell ing . The final objective is to develop a series of codes capable of supporting appl ication of reburning technology to power station boilers . A three level approach was defined in order to achieve this important objective : 1) devel opment and validation of codes able to correc tly simulate fluid flow and mixing , 2) simulation of the impact of gas reburning on heat re l ease in the furnace and boiler heat transfer and 3) prediction of the reductions achievab le . During the code development process it was decided to maintain a strong interaction between numerical modelling and physical measurements . Velocity and mixing measurements were taken in a 0 .46 sca l e isothermal model of ABB-CE ' s Boiler Simulation Fac ili t y (BSF ). The model is located at ABB CE ' s Kresinger Development Laboratory (KDL) and here s i mu l ates the Santa Gilla Unit 2 configuration . The objectives of this activity were : 1) to get an estimate of the BSF flow field and mixing and 2) generate validation data for the numerical models . The detai led geometry of the model is shown in Figure 5, highlighting the locations of the measuring planes . Three firing configurations were examined . One test was conducted in the baseline configuration and the other two in two reburn configurations (15% and 25% reburn gas ) . A complete review of the physical model ing is available in Reference (6) . In this paper the results of baseline and 15% gas reburn tests are presented . The code utilized for the numerical simulation of the experimental data is named FLUIHP and has been described in (7). For the presented simulation a grid of only 10000 nodes was adopted , because calculations conducted with finer grids (a larger number of nodes ) have shown no substantial improvement in the predictions obtained . In Figure 6 comparison between measured and predicted velocities is reported for two planes, 1 and 3 . It is evident that the main components of the velocity field are more correctly simulated in the baseline case than 3 Fig 5 LSFATF Test Plane Locations in the reburning (staged) case . The discrepancy in the reburning case may be due to two factors : from both smoke visualizations and pitot measurements of the flow field, it was evident that in the reburning condition the flow was much more turbulent and fluctuating, with large eddy fluctuations observed involving the whole furnace . under reburn conditions it was much more difficult to achieve in the model uniform and constant partitioning of the air through all the numerous injectors . This has been verified by taking independent measurements of the velocity distribution near the main burners and the various injectors . The agreement obtained was deemed adequate, however for more detailed comparisons more rigorously determined boundary conditions will be necessary. ABB CE's Boiler Simulation Facility (BSr) was utilized to generate temperature and gas species validation data in support of reacting flow modeling activities . Ouring the testing, horizontal plane temperatures and main combustion species were measured for different test conditions to determine the effect of combustion modifications on furnace performance . All these data were initially utilized to validate the model for heat transfer predictions . A detailed description of the code used for modeling will not be presented here, but is contained in references (8) and (9) . The series of codes utilized is capable of predicting fluid flow, heat release distribution, heat fluxes and gas temperatures in the furnace . A brief description of the code section that calculates the heat release profile follows . In the previous applications, simple algorithms for heat release calculation were utilized, based on empirical formulas (10) . With the introduction of low-NOz technologies, where staging and more complex techniques are utilized, the simple heat release calculations are no longer valid, due to the strong influence of fuel and air mixing . A program for heat release calculation was therefore developed based on the equilibrium composition calculations of all the more important species (C, H2 , 02, CH4, NO, H20, CO, C02). Equilibrium compositions are calculated knowing the local temperature and the concentration of the elements C/H/O/N through the minimization of Cibb's function . These calculations |