Talk:Streamline/@comment-180.4.217.214-20130428093259/@comment-6782803-20130428095455

This hints toward a multiplicative stacking with Energy regenerating in integers rather than the additive stacking mentioned in the article.

Additive stacking would result in costs of 20, 28, 20 and 80 as exact whole numbers.

Multiplicative scaling, however, would predict costs of ~20.243, 28.340, 20.243 and 80.972, all of which would fit your provided data if you assume that Energy is regenerated in integers.