I am seeing that result as well. I also reproduce the same error in the calculated y-direction force for the problem obtained by rotating Greg's geometry 90 degrees clockwise (program calculates a positive upward force when it should be negative downwards...)