feeding Network of phased circular antenna

Dear All, I am working to develop a feeding network for phased array of 8 antennas. Each antenna in the array is being fed with same amplitude signal but a constant phase difference of 45 degree. I have simulated this in HFSS as shown in the attachment. but result are not valid. It looks to me there are HFSS have problems in determining the exact result. I tried it with CST. can't get good results. i use em talk to generate extra lenght. which is 6.5 mm practically with HFSS for 100 ohm line. Sub height=0.5 mm of Rogers 5870. f=3.5 GHz. Please do comment.

em talk uses approximate formulas to calculate lengths that are usually required to be tuned using full wave simulator like HFSS. It is not a surprise to have unexpected results with lengths and widths calculated through approximate formulas used by microstrip line calculators. I think small tuning to lengths and widths can solve the problem.

Yes numan, I already took care of this fact and tuned the length using HFSS.

As you can see in this image the 100 ohm line is rotated 90 degree around Z- axis. and Junction A and Junction B are at similar position. still the angle(S1p4)-angle(S1p3)!= angle(S1p2)-angle(S1p1). Anyways I going through the theory of these kinds of circuitaries. If anyone experienced could give suggestion, it will be quick. Thank You