14 Best Practices for Surrogate Based Uncertainty Quantification in. . .
445
Fig. 14.6 Convergence history of the drag quantile under uncertainties with direct integration, and
comparison with surrogate based approach, baseline configuration
Fig. 14.7 CDF for drag coefficient of the baseline configuration under uncertainties computed
with direct integration (17 and 800 samples) and with Surrogate Based approach
optimum configuration, as shown in Fig. 14.8. With 17 samples an error smaller
than 0.1 drag counts is obtained when compared with the converged statistics. As
a result, the surrogate is able to provide an accurate determination of the statistics
along the optimization at a reduced number of function evaluations, validating the
proposed approach.
445
Fig. 14.6 Convergence history of the drag quantile under uncertainties with direct integration, and
comparison with surrogate based approach, baseline configuration
Fig. 14.7 CDF for drag coefficient of the baseline configuration under uncertainties computed
with direct integration (17 and 800 samples) and with Surrogate Based approach
optimum configuration, as shown in Fig. 14.8. With 17 samples an error smaller
than 0.1 drag counts is obtained when compared with the converged statistics. As
a result, the surrogate is able to provide an accurate determination of the statistics
along the optimization at a reduced number of function evaluations, validating the
proposed approach.