Like Ken, I too would need more information..... But GENERALLY speaking it sounds like there isn't enough "time" or "pressure" difference between the 2 pumps.
I start my lead at 4 seconds, followed by the lag about 10 seconds later (for a total of 14 seconds from the start threshhold). But that is with a pretty aggressive P, and a calculated starting point (based on actual pressure) for the pump.
There are ALSO accel and decel times in the drives to consider. If the drives are set too sluggish, the process won't be increased as fast as the PID tries to increase it, leaving a difference that the controller thinks requires another pump to achieve.
What are your "accel" and "decel" set to?
Stationmaster