System Head Curve

In generating system heads curves for an in-line booster PS I'm getting results that show a odd curve where the head actually reduces at higher flows under certain time steps in my EPS model.  The elevated storage tanks in this system are all at the same overflow, on both the suction and discharge of the PS, but we are pumping to an area further at the extreme ends of the system.  There are several elevated storage tanks on both sides of the PS.  What I have found is that the time steps when the system curve has this odd behavior is at times when the tank on the discharge side of the PS, closest to the PS, is completely (does the model automatically assume an altitude valve in this instance?).  I still cannot explain why the system curve behaves this way even with the downstream tank full.  I've attached a picture of the system head curves, I've generated several for various time steps, so some of the them look typical, but seven or so that don't are all in a period of the EPS with that tank completely full.  I appreciate any help on these system curves.

Thanks, -Brian