What I did not make clear or what you missed was I am not demanding the 130* the machine is able to do at EWT of 39*. I intended to impart that, although the machine can do this it is not always the cheapest way to go for operating costs. That is my bad sorry. You know how it goes, you think something trough and leave out some of the typing. Try this. What I really do is let the machine run with the call for heat and it still loads the tank according to EWTs and run times. The machine has low and high limits that I imposed on it beyond its manufacturer's provided protection limits. I have not really charted the machine for a few seasons, but when new we kept the tank at about 110* and mixed down to about 82* to the floors. This all floats by a simple thermostat. We keep the stat set at 62* air temp and can heat the 60 X 36 building for about $85.00 in January in Ohio at $0.17 per KWH before delivery and service charges. The floor is a 6" pour with R-10 on five sides. We start the floor in mid October and let it run into the spring. This is a very slow system. Once warm, the floor holds the building temps at set point and the unit cycles at -20* and a fifty MPH wind off Lake Erie. Yes this is a real job that works. The heat loss for the building calced at 93,000 BTUH. The heat pump does only 58,000 BTUH at 39* EWT, (it is supplied by a frozen over pond). Why does it work, (hold set point)? It works because of the high mass floor being able to cover the building loss due to it's heat storage capacity even when the heat pump and buffering tank can not. If we took the building 200 mile north of where it is it might not work so well. We bet on slow and steady winning the race, designed for it and won. My point was to explain how to mix floors not how to wear out equipment. I have the spread sheet data if you would like to see it. I was also trying to help with the former propane heated system. Two fails in one. Must be time for bed. ps. This is a 5 ton water to water running on R-22, so it's spec charts are more forgiving than machines running R-410A.