changing excitation integration line direction changes power measurements in HFSS

I ran two simulations for a 3-element dipole array.

In the first simulation, the source excitation integration line for all the 3-elements was pointing in the same direction and I was getting
Pincident = 3.0 W
Paccepted = 2.5 W

I changed the direction of integration line of the middle dipole and the results became:
Pincident = 3.0 W
Paccepted = 2.95 W

So, when I alternated the direction of integration lines, Paccepted value improved denoting less impedance mismatch. How is that possible? Why alternating the direction of the source excitation integration line affects the power measurements in HFSS?

I was wondering about how to reduce impedance mismatch between dipole array elements in HFSS. If someone can shed more light on this topic, I'd appreciate it.

Thanks.

I am just guessing

but since you changed the direction of the integrtion line you changed the phase of the input voltage by 180 degree for one element

and a single element in an array is not independant, there is mutual coupling effect

considering the previous two sentences you changed the value of the problem source so the accepted power might change as well

Also, how about techniques to reduce mutual coupling between dipole array elements in HFSS? Any ideas?

If I have Pinc = 3.0W and Pacc = 2.95W, that should indicate good impedance matching and max power transmission, correct?

Thanks.