The compressor is on one side of the ledger or the other
A heat pump running in heating delivers heat that is partly ground and partly electricity. The COP on the submittal is heat delivered divided by electrical input, so at a COP of 3.6 delivering 48,000 BTU/h, the electricity is 48,000 divided by 3.6, which is 13,333 BTU/h — about 3.9 kW — and the ground supplies the remaining 34,667. That is 72.2 percent of the load, and the general form is one minus one over COP.
Turn the same machine round into cooling and the sign flips. Now the building heat and the compressor work both have to go somewhere, and the somewhere is the ground. The ground-side load becomes the load times one plus one over COP, which at the same numbers is 61,333 BTU/h rather than 34,667. Nearly twice as much. This is why a cooling-dominated building on the same tonnage ends up with a longer loop than a heating-dominated one, and why the loop is normally sized on whichever season is worse rather than on the nameplate.
The mistake that shows up most often in spreadsheets is treating the delivered heat as COP times the heat extracted. It is not. Delivered heat is COP times the electricity, and extracted heat is whatever is left over once that electricity is subtracted. Getting it backwards puts the ground-side load out by a large factor, in the direction that makes the loop look shorter than it needs to be, and it is the kind of wrong nobody catches until the ground has been asked to do it for several winters.
Two resistances in series
The conductivity method on this page is the standard series model: heat leaving the pipe has to cross the grout and the pipe wall to reach the ground, then spread out into the ground itself. Those are two thermal resistances in series, and the total is what the temperature difference is divided by.
The ground half is the resistance factor divided by two pi times the conductivity. With a factor of 2.50 and a conductivity of 1.40 that is 0.284 hours foot degrees per BTU. The borehole half — grout, pipe wall, the geometry of how the two legs of the U sit in the hole — is entered separately, at 0.25 in the defaults. Total 0.534. An 18 degree difference across that gives 33.7 BTU/h per foot, and 34,667 divided by 33.7 is 1,029 feet of bore.
The resistance factor is the piece that carries time. Ground close to the pipe reaches something like steady state in days; ground fifty feet away is still warming or cooling after ten years. A factor sized for a ten-year run is larger than one sized for a peak-day pull, and the two produce very different lengths from identical ground. Which one belongs in the box is a decision the designer makes about how long the loop is expected to hold its temperature, and it is not a decision this page can make.
What the conductivity table is telling you
Run the defaults and the sensitivity table shows the length at conductivities from 0.84 to 1.96, which is forty percent either side of the 1.40 entered. The spread is real but it is nowhere near symmetric, because the borehole resistance sits in series and is unaffected by the ground. At 0.84 the loop wants 1,394 feet, 365 more than the 1,029 at the entered figure; at 1.96 it wants 872, only 157 less. Forty percent low costs you thirty-five percent more length; forty percent high buys back fifteen percent.
The asymmetry is worth staring at, and it comes from the shape of one over the conductivity rather than from anything about the borehole. Halving the conductivity doubles the ground half of the resistance; doubling the conductivity only halves it. The borehole resistance cancels out of the difference between two rows entirely — it moves the whole column up or down by the same number of feet rather than tilting it — so what a stiffer borehole changes is the proportion, not the feet. Against a bigger total, the same spread is a smaller share, and the ground conductivity looks less important than it is.
Which is the argument for spending money on the grout as well as on the test. Better grout keeps paying at any conductivity, because it comes off the total directly rather than through a reciprocal.
What this does not do
It sizes one length against one peak condition. A real ground loop design carries the whole year: the balance between how much heat comes out over a winter and how much goes back in over a summer, whether the annual imbalance drives the ground temperature off in one direction over a decade, how the bores in a grid interfere with each other, what the loop does at part load with the compressor cycling, and how all of that lands on the entering water temperature the machine actually sees in year ten rather than in week one.
None of that is here, and no single-length calculation replaces it. Use this to check an order of magnitude, to see how the ground-side load is built, and to understand which input the answer is most sensitive to before you pay for the field work. Then have the design done properly, by somebody who carries the liability for it, and permitted by the authority with jurisdiction over the ground being drilled.
Questions people ask
How many feet of ground loop does a ton need?
There is no general answer, and any number quoted as one is a number from somebody else site. On the defaults here it works out at 257 feet of bore per nominal ton, but that figure is entirely a product of the conductivity, the two resistances, the temperature difference the design allows and the COP at the design condition. Change the conductivity from 1.4 to 1.0 and the same load wants 312 feet per ton. Change the design temperature difference from 18 degrees to 25 and it wants 185. The number is an output of the site, not an input to it.
Why is the ground-side load smaller than the building load in heating?
Because part of the heat delivered indoors is the electricity the compressor consumed, turned into heat by the act of compressing. The COP counts heat delivered against electricity in, so at a COP of 3.6 every unit of electricity turns up indoors alongside 2.6 units that came out of the ground. The ground supplies one minus one over COP, which is 72.2 percent of the delivered heat. In cooling the arithmetic reverses: the ground takes the building heat plus the compressor work, so it carries 128 percent of the load rather than 72.
Is exchanger length the same as pipe length?
No, and this is the most common unit error in loop sizing. A 200 foot bore with a single U-tube is 200 feet of exchanger and 400 feet of pipe, because the pipe goes down and comes back. A trench with six pipe passes in it is one foot of exchanger and six feet of pipe per foot of trench. Every heat transfer rate quoted per foot has to be told which foot it means. This page works in exchanger feet, because that is the length of ground being asked to do the work, and reports pipe feet separately for the purchase order.
Where do I get a thermal conductivity number?
From a formation thermal conductivity test drilled on the site you are building on. It is a real test: a test bore, a known heat input, and the temperature response logged over a period of days. Numbers taken from a soil map, a textbook table or a similar project down the road are estimates, and the sensitivity table on this page shows what an estimate that is thirty percent out does to the length. For a small residential job the test may cost more than the uncertainty is worth and the designer will say so; for anything larger it is normally drilled first for exactly this reason.
What happens if the loop is short?
Not much, at first, and that is the problem. A short loop works on the day it is commissioned. What it does is let the ground around it drift further from its undisturbed temperature every season, because more heat is being pulled through less ground. Entering water temperature falls a little further each winter, capacity falls with it, the compressor runs longer to make up the difference, efficiency drops, and eventually the machine cannot hold the house on a cold night. That takes years to arrive and it cannot be fixed without drilling again. It is the specific reason ground loop design is engineered rather than estimated.