I’ve been lurking on this site for years, and finally got my system online last fall. I thought I’d post it since it’s a little unusual in being based on radiators, and whereas it seems others have had difficulties with such systems or questioned their feasibility, ours has performed better than I expected, providing wonderfully gentle radiant heat right through the coldest days of winter without any major problems or any special high-temperature equipment. We were helped by the fact that, after multiple improvements to the insulation, our oil-fired heating system was considerably oversized, with ~360 sf of EDR for a 1400-sf 1-storey house with 7½ ft ceilings in a relatively mild climate (average January lows of 28°F). I had resigned myself to a professional install, but none of the local HVAC guys were interested in doing a hydronic system, much less one involving cast iron radiators. They all wanted to do forced air using the existing ceiling A/C ductwork. If they’d come back promptly with estimates I might have done it, but the longer I waited the more resistant I became to the idea of giving up my radiators. Meanwhile in the course of the estimates I got to know the guys who do 90% of the loops around here. To my surprise, they were willing to work with me on a DIY project, basically doing the whole source side of the equation, including all the plumbing, flow center, antifreeze, flushing and purging. Further support for a DIY system came from an unexpected source: my wife. “10-year guarantee?” she said, “That means you can’t touch it for 10 years. You’ll be miserable. Plus I don’t think I want hot air blowing in my face.” So I nabbed a scratch-n-dent Bosch 3-ton water-to-water unit on ebay, and got the twin 360-ft vertical loops drilled, plumbed and flushed. Then I got the old-timer who put in my oil boiler to do the load-side plumbing. Here's the basic setup: The distribution and control systems are extremely simple. The radiators are on a 2-branch Monoflo loop with 1½” ID steel pipe on the main trunk and 1” ID on the branches. The heat pump is spliced directly into the radiator loop, bypassing the existing boiler, with ball valves to allow a quick switch back to the boiler. The room thermostat activates the compressor, flow center and two Grundfos Alpha ECM circulator pumps simultaneously. I’ve always contended that with the huge thermal mass of a radiator system, there really should be no need for a buffer tank, and my experience with this system supports that view. There certainly is no short-cycling. The compressor typically runs for 60 minutes per cycle, tripping the room thermostat as LLT reaches 110-120°F, depending on the outdoor temperature. As far as matching flow rates, 9 gpm seems to be perfect for both the heat pump and the Monoflo loop, noiselessly keeping all the radiators within a few degrees of each other and within 10° of LLT. Thanks to the combination of old-fashioned large-diameter steel pipes and modern ECM circulators, 9 gpm can be achieved at a cost of <90 watts. I’ve got the system heavily monitored on welserver.com/WEL0961, including a running estimate of COP. Feeding the radiator return directly into the heat pump means that, in milder weather, each cycle starts with ELT between 70° and 90° for a COP between 4 and 5, gradually decreasing to between 3 and 4 by the end of a cycle, and averaging ~4. Even on the coldest days, average COP has never gone below 3. Compared to the old oil-fired system, the geo provides a more even heat distribution through the house, and a more constant temperature over time, usually within ±1°F of setpoint. I have been pleasantly surprised at how low the radiator temperatures are, exceeding 110°F only when outdoor temperatures dip into the teens. Here are some data from this week’s cold snap (green=outdoor temp; black=room temp; brown=radiator temp). Additional welserver data show that LLT rarely exceeds 115° and never exceeds 125°, well within the capability of almost any heat pump, and way below the 145°-150° ratings of the high-priced high-temp units specifically designed for use with radiators. In fact, the supply temperatures I’m seeing for a given level of heat output are basically the same as I would need if I’d done the whole house with PEX tubing under hardwood floors, based on the chart in John Manning’s video (). I’ve yet to even come close to needing backup heat. My overall impression is that, far from being a force-fit, radiator systems and especially Monoflo systems, are practically made for geo. But one thing I am beginning to strongly suspect is that the standard charts for radiator output, published in Columbia’s Cast Iron Radiator Heating Capacity Guide, and cited in e.g. Carrier’s Water-to-Water System Design Guide, seriously underestimate radiator capacity at very low temperatures. To take an extreme example, for an “average water temperature” 35° above ambient, that chart gives a capacity of 20 Btu/hr per sf EDR. So my system at 100° water temperature should dissipate 7200 Btu/hr. Yet I can see from my own data that, at 100° average supply temperature, it instead puts out 12,000 Btu/hr. (The overall average water temperature, midway between supply and return, is even lower, more like 95°.) Of course I will lose a little heat from my (well-insulated) supply lines, but certainly not 40% or more. Basic physics tells you that Columbia’s straight-line chart cannot be right, because it predicts that if the temperature difference is 25° or less, the radiator will have no output at all. Because of reduced convection, output for a 25° difference might be less than half the output for a 50° difference, but it can’t possibly be zero! I just got my December electric bill for $79, with at most $30 attributable to ~260 kWh of geo. OK, it was a very mild December, but basically I am using a little less than 1 kWh per degree-day. At that rate the whole heating season shouldn’t cost more than ~$400, less than half what I’d be paying for oil, even after the recent collapse in prices. Thanks to all the folks on this forum, without which I likely would not have gone through with the project.