DU is denser and therefore heavier. That means at the same velocity in the graph, the DU rod will be shorter. That is one of the key factors why it seems that the optimum velocity for DU would be lower; at higher velocities (and a fixed 10 MJ), the DU rod becomes too short for optimal penetration.
This is also one of the reasons why the overall penetration of tungsten is displayed as being higher in the graph: the longer tungsten rod can get closer to optimum velocities.
It's a very interesting question, and it appears to have little to do with energy. I played around with Willi Odermatt's penetration calculator on his website (http://www.longrods.ch/equation.php), and it doesn't seem to have anything to do with length.
Comparing a DU rod (18000 kg/m^3) and WHA rod (17000 kg/m^3) with identical dimensions and identical striking velocity of 1800 m/s, the WHA rod penetrates a few cm more, even though the DU rod has higher bulk density and higher energy. This is for a 300 BHN target with a density of 7850 kg/m^3 at 60 degrees obliquity. Comparing the same two rods against the same target but at a striking velocity of 1600 m/s, the DU rod penetrates a few cm more.
However, this changes when the obliquity changes. Comparing the DU rod (18000 kg/m^3) and WHA rod (17000 kg/m^3) with identical dimensions and identical striking velocity of 1800 m/s on the same target but at 0 degrees obliquity, the penetration is exactly the same (difference is a hundredth of a percent). For the same 0 degree target but at 1600 m/s, the DU rod penetrates a few cm more than the WHA rod.
So it seems that WHA rods only penetrate more than DU rods on high obliquity targets and at high velocity. Otherwise, DU is better or on par with WHA. Based on this, I guess that WHA handles lateral stress better than DU, allowing it to penetrate more steel at higher obliquity without fracturing during penetration.
Edited by Hakka, 22 November 2017 - 0710 AM.