Clearly the 30% selection factor with the dynamic algorithm had the best performance. Of the 24 trials, 23 found satisfactory solutions, and 22 found valid solutions. The one run that did not find a satisfactory solution had only four violations, all but one under 1.05, and the maximum unity check was 1.13. That one violation was the only thing keeping it from meeting the definition of “satisfactory,” and in practice it would probably also have been usable by a designer. Weights ranged from a minimum of 10.08 kips (very close to the best solution we ever found) to 20.82 kips, with a mean of 15.27 and a median of 14.03.