The Rmax is the maximal feasible signal. Rmax is determined by the amount of binding sites (ligand concentration) and the size of the ligand and analyte molecule. Therefore, Rmax is fitted locally when different analyte molecules are used over the same surface.
The Req is the response when the interaction between ligand and analyte is at equilibrium. Req is determined by the maximal number of binding sites (RMax), kinetics (ka, kd) and the concentration of the analyte (C).
When fitting curves both Rmax and Req are fitted parameters. This means that the program tries to determine the best values for these parameters. Even is the analyte concentration range is not saturating the ligand (reaching Rmax) the fitting will calculate the theoretical Rmax. That holds also for the Req. Even if the injection time is not long enough to reach equilibrium (steady state) it will calculate the theoretical Req.
Now for the curves.
Because I can not simulate the single cycle kinetics, I did it the classical way. It makes no difference to understand.
Above two simulations. If you look carefully in the left lower corner you will see the simulation of your experiment. And the other longer curves are two hour association curves of which only the 5 nM will reach equilibrium and not Rmax because the analyte concentration is to low.
To reach Rmax the analyte concentration must be at least 40 – 50 times KD. In your case 25 nM is about 97% Rmax. You can try to simulate it yourself.
You don not tell what the difference (ligand type or density?) in ligand surface is between flow cell 3 and 4. In addition the association kinetics between the two flow cells is different.