I had a chance to actually sit down by the pumps yesterday evening as the sun was setting to witness what was actually happening with the pumps. As the sun came and went yesterday, I had episodic problems with the no-flow situation previously described. As always, unplugging the pump and "resetting" the system always resolved the problem. Unforunately, I can't babysit the pumps all day along. I need to definitively figure out what the problem is and either find a way to adjust the pump behavior or find a way of automatically resetting the system when I'm not around to reset it manually.
My observations are pretty much what I suspected. As the pump speed decreases stepwise due to the decreasing amount of thermal energy on the roof, everything seems to be fine until what appears to be the slowest speed setting. Once the pump steps down to that speed setting, the power of the pump is insufficient to maintain the column of water up to the collectors and drainback begins. What follows from here is speculation but it seems most logical to me that at a certain point, drainback occurs until the equilibrium point whereby the level of water in the pipe above the pump should be at the level of the water in the tank -- let's say 18" above the pump impeller for argument's sake. All this time, the pump is continuing to run at its slow speed thinking that is all that is required to keep the delta T at its setpoint. At this point, however, instead of circulating fluid, I suspect that the pump is frothing up the water in the pipe and agitating it enough to get air entrained into the fluid and eventual create an airlock around the impeller. Since fluid is no longer flowing up through the collectors, assuming there is still some thermal energy to collect on the roof, the collector temperature rises again and increases the delta T. The pump controller then stepwise increases the pump speed to try to bring the delta T back down to its setpoint. The lack of fluid flow ends up letting the delta T rise such that the pump controller eventually tries to run the pump at full speed. The pump stays at full speed until the sun goes down or the delta T falls spontaneously. By this time, however, the pump has gotten extremely hot as the circulating fluid acts as both the lubricant and the coolant for the pump. This is the basis of my concern that something may have gotten cooked in the pump as a result of one too many overtemperature episodes. Even if air entrainment was not an issue, I'm not sure that the pump would be able to get the fluid back up to the collectors with its stepwise increase in pump speed uintil it got to full speed anyways.
So the question becomes what are the possible remedies for this situation? Perhaps moving the pump lower in relation to the water level in the tank will allow enough water on top of the impeller even during drainback to prevent air entrainment and an air lock. A fancy system that I've thought of would be to read the flowmeter and collector/tank sensor readings and determine when the system was in its chaotic state and then have a trigger to cut the power to the pump for 30 seconds or so before restoring power to the pump and letting it automatically reset itself, much as I do manually when I'm around. I will take some pictures and send to Mr. Sweet at Taco shortly.