OCR Text |
Show thes~ Dunlerical procedures to simulate observed performance in real systems and to quant ify the error introduced by uncertainty in model parameters. Of particular note, two recent studies have been completed on such evaluations. The first (10], is a concurrent analysis of data reliability and code performance which reviewed comparisons of model predictions to local experimental data. This evaluation inc1uded seventeen cases encompassing simpJe gas mixing studies through to reacting coal fired furnace studies. The second [11], is a global non-linear sensitivity analysis aimed at quantification of the effect of uncertaint) in input parameters on output functions. Such analyses demonstrate that these numerical tools are capable of qualitative a priori predictions of pc furnace performance even though there remain unanswered questions pertaining to each of the modeled subprocesses. In this section we use the conditions and measurements from the' \\·ell controlled laboratory furnace to examine the global and local effects of two aspects of heat transfer on pulverized coal combustion processes: 1) the effect of local energy loss from the system on flame behavior, and 2) the impact of turbulent fluctuations on the mean flame properties. These two aspects are demonstrated by showing results from three different predictions. One calculation is performed assuming that the gas-phase flame is locally adiabatic. That is, the local gas phase enthalpy. and thus the temperature, is calculated by knowing the local stoichiometry even though the partic1e phase heat transfer is fully accounted for. This assumption simplifies the computation and historically has been thought to be adequate. In this calculation all turbulent fluctuations in local conlposition and energy are fully included. A second computation includes all heat transfer aspects in both the gas and solid phases and includes local fluctuations in energy and stoichiometry due to fluid turbulence. The third conlputation examines the inlpact of local fluid turbulence on the calculation by ignoring these fluet uations and using only mean \'alues for all variables. Figure 4 shows the local gas mean temperature predicted in the BY1T furnace for each of these three calculations. Although the adiabatic calculations shows higher temperatures throughout the furnace, as would be expected, the magnitude of the difference in temperature between the adiabatic calculation and the full energy equation simulation is dramatic. For exan1ple a difference of SOOK is obser\'ed bet ween the centerline outlet values in the two cases. The major mode of heat loss for the gases arises from the predorrunately (99+<;(,) radiative heat transfer from the particles to the reactor walls and by convective/ conductive heat transfer between the particles and the gases. From Figure 4 it is also clear that the turbulent fluct uations cause significant snloothing of ten1perature peaks and produce overall mean temperatures of lOO-200E higher than when these fluctuations are ignored. This effect results from the coupling of chemical reaction kinetics that vary exponentially with increasing temperat ure~ with other combustion processes, causing n1ean values to shift higher when a full ten1poral distribution of properties is accounted for at each spatial location. Although tenlperature data ha\'e not been collected in this furnace. local sanl- 9 |