The heat-carrier fluid is easily treated as a detail — a percentage entered once and forgotten. In reality, the fluid and its antifreeze concentration shape the flow regime, the heat transfer in the borehole, and the pumping energy for the entire service life of the system. Selecting them deliberately is among the least expensive ways to keep a design efficient; selecting them carelessly is among the easier ways to make it quietly costly.
The purpose of antifreeze, and its cost
Antifreeze allows the heat pump to continue operating in heating mode at lower loop temperatures, extending the operating window into colder conditions. That function is indispensable in heating-dominant systems, where the entering fluid temperature can approach freezing during peak extraction. But it is not free. A higher antifreeze concentration:
- raises the dynamic viscosity of the fluid, moving the flow toward the laminar regime;
- increases the convective resistance $R_f$ between the fluid and the pipe wall;
- demands more pumping energy to maintain the same flow;
- and, through the higher $R_f$, tends to enlarge the required borehole field and its capital cost.
The penalty for over-concentration therefore appears three times over: in the pump, in the field, and in the annual energy. None of these is recovered later; they are fixed at the moment the fluid is specified.
Why concentration governs the flow regime
The mechanism connecting concentration to cost runs through the Reynolds number,
$$ Re = \frac{\rho\, v\, d_i}{\mu} $$
where $\rho$ is the fluid density, $v$ its mean velocity in the pipe, $d_i$ the pipe inner diameter and $\mu$ the dynamic viscosity. It determines whether the flow is laminar or turbulent: below $Re \approx 2300$ the flow is laminar, above roughly 4000 fully turbulent. Turbulent flow mixes the fluid across the pipe section and keeps the convective resistance low; laminar flow stratifies it and lets the resistance rise sharply. Because $\mu$ sits in the denominator, every increment of antifreeze — which raises the viscosity — lowers $Re$ and pushes the circuit toward the laminar transition at a given flow rate.
The magnitudes are not small. For a 20% propylene-glycol mix, the viscosity roughly triples between the warm and cold ends of the operating range — from about 1.5 to 4.5 mPa·s — and the effect on the flow regime is decisive. In one representative design the Reynolds number falls from about 7300 at the cooling limit to about 3100 at the heating limit, dropping from firmly turbulent into the transition band even though the heating-mode flow is the higher of the two. The convective resistance $R_f$ follows, rising from roughly 0.0022 to 0.0057 m·K/W — about 2.6 times higher at the cold end, exactly where heat is hardest to draw from the ground.
The penalty at that transition is abrupt, not gradual. In turbulent flow the Nusselt number, and with it the convective heat-transfer coefficient, climbs steeply with $Re$ (Gnielinski 1976); in fully laminar flow it collapses to a small constant, independent of flow. Crossing into the laminar regime therefore raises $R_f$ as a step, which is why a circuit that slips below the threshold at peak load loses heat-transfer performance out of proportion to the flow it lost.
The consequence is a compounding penalty. To restore turbulence in a more viscous fluid, the flow rate must increase — which raises pumping energy directly — and if turbulence is not restored, the higher $R_f$ drives the entering fluid temperature lower, toward the minimum the heat pump can tolerate, which in turn calls for a longer field. A modest over-concentration thus propagates into both the operating cost and the capital cost, by two distinct paths.
Figure 1 shows this flow dependence directly: the fluid film $R_f$ swells as the flow drops toward the laminar transition — exactly the shift a thicker, over-concentrated fluid forces at a given flow rate.
That penalty, though, is only critical near peak load. Through the long stretches of the year when little heat is being moved, a high $R_f$ costs almost nothing — the entering fluid temperature stays well clear of its limit whatever the resistance — so the flow can be turned down substantially and laminar flow accepted, trading a resistance that no longer matters for a real saving in pumping energy. The requirement is therefore turbulence at peak, not at all times; and it is precisely because an over-concentrated fluid can force the circuit laminar at peak, when it does matter, that the concentration should be kept no higher than the cold demands.
A misdiagnosis in the field
This mechanism has a common and instructive practical consequence. On a cold winter night, a geothermal system shuts down. The technician called to the scene turns to the most obvious explanation: the circuit is freezing due to a lack of antifreeze. He then adds pure glycol. More often than not, this is a mistake, and it can even worsen the problem.
The lockout is real, but the cause is frequently not a frozen loop. If the circulator is undersized for the fluid it carries, a flow that was turbulent in mild weather can fall into the laminar transition as the cold thickens the fluid. As shown above, $R_f$ then rises as a step: the fluid can no longer draw heat from the ground quickly enough, the entering fluid temperature drops toward its limit, and the heat pump shuts down to protect itself. The system is not short of antifreeze — it is short of turbulence. Adding pure antifreeze at this point raises the viscosity further, lowers the Reynolds number further, and drives the flow further from the turbulence it needs; the intervention deepens the very condition that caused the fault.
The correct first step is inexpensive: measure the actual concentration with a field densimeter or refractometer before adding anything. An over-concentrated loop calls for dilution, or a stronger circulator, not more glycol; an under-concentrated one can then be corrected deliberately rather than by guesswork. A measured concentration turns a plausible but mistaken diagnosis into the right one — and it belongs in a winter service call before the first litre of fluid is poured.
A margin, not a cushion
The temptation is to add a margin for safety. It is a common and costly habit. The sound rule is the opposite of generous: the freeze point of the fluid should sit a few degrees below the lowest temperature the loop will ever experience, and no lower. A margin of 5–6 °C between the minimum entering fluid temperature and the freeze point protects the evaporator, where local temperatures can fall below the measured loop temperature (Kavanaugh and Rafferty 1997). Beyond that margin, every additional percentage point of antifreeze purchases viscosity and pumping cost that serve no purpose: over-concentration does not make the system meaningfully safer, only slower, more energy-intensive, and larger.
Reading the properties before committing
The thermo-physical properties of the common fluids — pure water, ethylene glycol, propylene glycol, methyl alcohol and ethyl alcohol — can be computed from the Melinder correlations across the full temperature and concentration range (Melinder 2010). For any fluid and concentration, the relevant quantities are the melting temperature, the fluid volume in the pipes, the thermal conductivity, viscosity, density and specific heat at the temperature limit, and the resulting convective resistance $R_f$.
These properties do not move together as concentration changes. Adding glycol depresses the freeze point — the intended effect — but also lowers the thermal conductivity and specific heat while raising the viscosity, so the same change that buys cold protection erodes heat transfer on three counts. Examining each property at the temperatures the loop will actually reach, rather than at a nominal mid-range condition, converts an assumption into a quantified trade-off, and a comparison between two candidate fluids, or two concentrations of the same fluid, becomes a matter of reading the curves (Figure 2) rather than guessing.
Selecting the fluid: thermal properties are half the decision
The thermal comparison narrows the field, but the final selection is rarely made on thermal grounds alone (ASHRAE 2023). Safety and regulatory requirements can override it outright: a fluid that is optimal on paper may be prohibited by local code, or ruled out on toxicity or flammability grounds, whatever its thermal merit. Three non-thermal factors often weigh as heavily, and sometimes decide the matter on their own.
- Toxicity. Propylene glycol is commonly chosen where any contact with potable water or food is possible, being far less toxic than ethylene glycol. Ethylene glycol offers better thermal and viscosity behaviour but carries handling and disposal constraints, and spent fluid must be managed as a regulated waste.
- Local code. Many jurisdictions restrict the fluids permitted in ground loops, particularly near drinking-water aquifers; the permissible list is sometimes shorter than the technically optimal one.
- Material compatibility. The fluid must remain compatible with the pipe, fittings and pump materials over a multi-decade service life, without promoting corrosion or attacking seals.
The alcohols can deliver attractive low-temperature performance, with lower viscosity than the glycols at deep cold, but introduce their own flammability and handling considerations. There is no universally optimal fluid — only the fluid best suited to a given climate, regulatory environment and risk tolerance.
A decision that lives with the system
The concentration chosen at construction is not necessarily the concentration the loop will carry for thirty years; fluid can be diluted during top-ups or service, shifting the freeze point and the flow regime, or it can degrade chemically and lose its inhibitor package. The target concentration is worth specifying clearly and verifying at commissioning and during maintenance — the same care given to the borehole design is owed to the fluid that runs through it, since a loop filled to the wrong concentration undoes the thermal work done upstream.
Conclusion
Antifreeze is a balance, not a safety dial to be turned up. Matching the freeze point to the coldest condition the loop will experience, holding a sensible margin and no more, and allowing the property data rather than habit to set the concentration produces a loop that is efficient across its life. Selecting the fluid on toxicity, code and compatibility as well as thermal performance turns an inherited default into a deliberate design decision — one that, because it touches pumping, heat transfer and field length at once, repays the attention several times over.
References
- ASHRAE. 2023. ASHRAE Handbook — HVAC Applications. Atlanta: ASHRAE.
- Gnielinski, V. 1976. New equations for heat and mass transfer in turbulent pipe and channel flow. International Chemical Engineering 16 (2): 359–368.
- Kavanaugh, S. P., and K. Rafferty. 1997. Ground-Source Heat Pumps: Design of Geothermal Systems for Commercial and Institutional Buildings. Atlanta: ASHRAE.
- Melinder, Å. 2010. Properties of Secondary Working Fluids for Indirect Systems (Secondary Refrigerants or Coolants, Heat Transfer Fluids). Paris: International Institute of Refrigeration.
Part of the GSHP in Practice series.
Continue: Pillar — How to design an energy-efficient GSHP system · ← Previous: Entering fluid temperature and performance · Next: Cutting borehole thermal resistance →