You are here:
EVALUATION OF TUPOS USING DATA FROM THE EPRI KINCAID FIELD STUDY
Citation:
Turner, D., L. Bender, J. Paumier, AND P. Boone. EVALUATION OF TUPOS USING DATA FROM THE EPRI KINCAID FIELD STUDY. U.S. Environmental Protection Agency, Washington, D.C., EPA/600/J-91/186 (NTIS PB91233494), 1991.
Description:
Data from SF6 tracer studies conducted at the Kincaid Power Plant in central Illinois by the Electric Power Research Institute in 1980 and 1981 have been used to evaluate the TUPOS air quality dispersion model. ost of the 96 hours' data are from periods representing daytime convective conditions when the impact of an elevated buoyant source would be expected to be greatest at ground level. ince on the order of 200 tracer measurement stations on four to six arcs were in operation during each hour of the study, a reasonable estimate of the maximum concentration along each arc could be made. he maximum concentration on each arc was the principal value used for purposes of comparing tracer measurements with model estimates. n addition to making comparisons between tracer and TUPOS estimates, comparisons were also made using the model MPTER. lthough the means of residuals from the hourly maxima are not statistically different at the 95% confidence level for these two models, the TUPOS shows subjective improvement. he performance results were used to suggest further changes to the TUPOS model. hese consisted primarily of determining the fluctuation statistics and wind speed and direction for a height midway between effective plume height and ground-level rather than at plume height. hese changes were implemented and shown to provide improvement when tested on the dependent data set used to evaluate the original model. he mean of the residuals (model concentration minus tracer concentration) of 7.9 ppt (parts per trillion) for the revised model was found using a t test to be significantly different at the 95% confidence level from the mean of residuals for the original model of -13.1 ppt. his was from using all data for all arcs for all hours (580 data pairs). his shifts from underestimation of the concentration for the original model to slight overestimation with the revised model. lthough the significance of the results for the revised model are not as great as they would be if an independent data set were used, these results are still useful in indicating the proper direction of changes to be made that can improve modeling.