From figure 6 in the paper, it looks like you get the win for anything from 30 degrees angle upwards. But with 0 degrees angle, the turbines seem to interfere,

From figure 6 in the paper, it looks like you get the win for anything from 30 degrees angle upwards. But with 0 degrees angle, the turbines seem to interfere, and you get something like 50% loss. So with a westerly prevailing wind (from 270 degrees), you'd likely put the turbines in rows at 210 or 330 degrees. If you used 210 degrees, and the wind comes from 210 degrees, you lose. Almost all the rest of the time you win. With some simple analysis of historic wind directions, you ought to be able to choose the angle to maximize net output by putting the inefficient angle in a direction that is relatively uncommon.