Difference between assigned lumped port impedance and Zpi impedance

I am using the simulation software HFWorks, which is a 3D electromagnetic FEM solver integrated with SolidWorks CAD.

I am using the software to simulate antennas. I am wondering what the difference is between the impedance that you assign to a lumped port feeding the antenna structure and the Zpi impedance that is calculated during the simulation.

My understanding is that when you assign a specific impedance to a lumped port antenna feed,this is the impedance that the antenna would see looking back in to the port, i.e. if you assigned the lumped port to be R = 50 Ohms, then the antenna would be modeled as if it were being fed by a 50 Ohm transmission line.

I think this is not the correct understanding however. When simulating a simple thin wire (2mm thickness) halfwave dipole, I know that the antenna impedance becomes purely real at resonance which theoretically should be around 73 Ohms. If I assign 73 Ohms to the lumped port feed (in my view this is impedance matching) then I should expect a reflection coefficient close to zero at resonance. This however is not the case. It seems that the purpose of the lumped port impedance is for calculating the normalised scattering parameters , as well as Z and Y parameters.

I know that changing the geometry of the lumped port feed changes the calculated Zpi. Does this mean that to implement impedance matching to the antenna it is necessary to adjust the geometry of the lumped port so that the Zpi impedance is 73 Ohms?

Any clarification would be appreciated.