What Glycol Costs You in Heat Transfer and Pumping

Why this matters

Adding glycol to a loop is usually a defensible decision made under pressure after a freeze event, and it is almost always made without recomputing anything. The fluid then carries less heat per gallon, so the loop delivers less capacity at the same flow and the same temperature difference, and nobody connects the complaint that arrives four months later to the drum that went in. There is no fault to find because nothing is faulty. The system is doing exactly what the new fluid allows.

The penalty is two separate terms with two separate fixes, and the one that usually bites is not the one people expect.

Before you charge a system or open a pump

Read Section 8 of the glycol product's safety data sheet and use the eye protection and glove class it names. Inhibited glycol packages carry additives, not just glycol, and the glove that resists one package may be degraded by another.

Ethylene glycol is acutely toxic if swallowed and it tastes sweet, which is why it kills pets and occasionally children. Contain every spill at the point it happens, do not put it on the ground or to a storm drain, and dispose of drained fluid through the supplier or a licensed waste handler under the local authority's rules. Store it labelled and away from anything anybody drinks from.

A glycol charge point connected to a building's potable water is a cross connection, and glycol reaching a potable fixture is a public health event, not a service complaint. The backflow assembly protecting that connection is set by the local plumbing code and the water purveyor, and it is not optional because the fill is temporary.

If the answer to any of this is an impeller change, isolate and lock the circulator and relieve the loop pressure at that section before a fastener moves. That is mechanical isolation and stored energy, 29 CFR 1910.147, and the stored energy is a pressurised column of hot fluid.

The coil that stopped making its numbers

A makeup air heating coil in an outdoor air stream, served by a hydronic loop. Original design: 1,000,000 Btu/h at 100 gpm with a 20 F temperature difference across the coil, circulator selected at 100 gpm and 40 ft of head. After a hard freeze split a return line, the loop was drained and recharged with a 50% inhibited ethylene glycol mix. The pump, the piping and the controls were left alone.

The following winter the coil could not hold its leaving air temperature on design days. The complaint arrived as an airside problem, so the airside got checked first.

What came back clean. Filters were within their pressure drop range. Fan speed and airflow measured close to design. The valve stroked full open and the actuator was not stalling. Supply water temperature at the coil inlet was at setpoint. Flow was measured at the balancing valve and read 97 gpm, three percent under the 100 the system was balanced to when it was on water. Everything a tech would normally suspect was either fine or trivially off.

The 97 gpm was the misleading result. It looked like a flow problem that was too small to matter, and it sent two visits into the airside. In fact the flow was very nearly correct and the capacity was still short, because a gallon of 50% glycol is not a gallon of water.

The term that was actually wrong

Load carried by a liquid loop is flow times temperature difference times a constant, and the constant is the fluid's density times its specific heat. For water near 60 F that constant works out near 500 in the familiar Btu per hour form, because 8.33 lb per gallon times 60 minutes per hour times a specific heat of 1.0 gives about 500.

For a 50% ethylene glycol mix near the same temperature, the density is about 7% higher, roughly 8.91 lb per gallon, and the specific heat is around 0.80 Btu per lb per F instead of 1.0. Sixty times 8.91 times 0.80 is about 428.

Both constants above are evaluated near 60 F so the comparison is like for like. At heating loop temperatures water's constant falls a few percent as it expands, while the glycol mix's specific heat rises with temperature, so the gap narrows somewhat. Take the properties from the fluid manufacturer's table at your actual operating temperature rather than carrying a room temperature ratio into a 180 F loop.

Now run the loop as it was found. At 100 gpm and a 20 F difference, 428 times 100 times 20 is 856,000 Btu/h against a design of 1,000,000. That is 85.6% of design, a shortfall of about 14%, and it appears with the pump at its correct flow, the coil clean, and every reading in the log looking normal.

Splitting the penalty into its two terms

The heat capacity term. To carry the design load at the design 20 F difference you now need 1,000,000 divided by 428 times 20, which is about 117 gpm. That is 17% more flow than the water design, and it is an unavoidable consequence of the fluid, independent of the piping, the pump or the coil.

The pumping term. Push 117 gpm through piping sized for 100 and friction rises. In turbulent flow through the same pipe, head loss expressed in feet of the flowing fluid scales roughly with flow to the power 1.85, so 1.17 raised to 1.85 is about 1.34. On top of that, the mix's higher viscosity raises the friction factor; at moderate loop temperatures a correction on the order of 1.10 is a reasonable illustration for 50% ethylene glycol, and it is only an illustration. Together that is about 1.47, so 40 ft of head becomes roughly 59 ft.

Pump input power is proportional to density times flow times head in feet. That is 1.07 times 1.17 times 1.47, about 1.84. Delivering the original load through the original piping with the new fluid needs on the order of 1.8 times the pump input, and that is before the efficiency loss from moving the duty point along the curve.

Both terms have to be stated with their conditions attached. The 1.85 exponent is turbulent flow. The 1.10 friction correction is 50% ethylene glycol at moderate temperature. Change either condition and the number is not yours to use.

Why the pump was not the story but the motor still felt it

A centrifugal pump develops close to the same head in feet of fluid whatever the fluid's density, and the system curve expressed in feet is also density independent, so the operating point barely moved when the loop went from water to glycol. That is why the flow only dropped from 100 to 97 gpm. The small drop came from viscosity steepening the friction curve, not from the density.

The motor is a different story, because the power the pump absorbs is proportional to density. Even with flow essentially unchanged, the same duty in a fluid 7% denser with a friction factor perhaps 10% higher draws on the order of 18% more at the shaft. The loop that looks unchanged on a flow meter has a motor working meaningfully harder, which matters on a pump that was already selected close to its motor rating.

So the shortfall in this case is almost entirely the heat capacity term. The pumping term was quietly present in the electrical reading and would only have become the dominant problem if somebody had tried to fix the capacity by raising flow.

Three ways out, and what each one costs

Widen the temperature difference. Keep 100 gpm and let the coil work across a wider range. 1,000,000 divided by 428 times 100 is a difference of about 23.4 F, so widening from 20 to roughly 23.4 F recovers the load with no pump change and no pumping penalty at all. It is free in pumping terms and it is not free in coil terms: the coil has to be able to produce the required output at the lower return temperature, and the heat source has to accept the wider return. Check both with the equipment selections before promising it.

Raise the flow to 117 gpm. This restores the original operating temperatures and costs roughly 1.8 times the pump input, an impeller or speed change, and a rebalance. It is the right answer where the coil cannot work at a wider difference.

Reduce the glycol concentration. Every penalty above scales with concentration, so if the freeze exposure only ever justified a lower percentage, dropping it recovers capacity on both terms at once. That decision belongs with the lowest metal temperature the fluid actually sees, which a sibling article covers.

The coil's own capacity derate sits alongside all three. Glycol's lower thermal conductivity and higher viscosity reduce the tube side film coefficient, so at identical flow and identical entering temperature a coil produces less than its water rating. The correction factors for that are published by the coil manufacturer and they are not the same as the loop side arithmetic above; do not multiply the two together and call it an answer, run the coil selection at the actual fluid.

Where these coefficients stop holding

• Cold. Every viscosity related number above was taken at moderate loop temperature. At 20 F a 50% mix is several times as viscous as water, small tubes can fall out of turbulent flow into the transitional range, and both pressure drop and heat transfer change character. A cold start or a low temperature glycol loop needs its own calculation at its own temperature, not this one scaled.
• Hot. At 180 F the mix's viscosity is much closer to water's, so the friction correction shrinks toward one and the penalty is mostly density and heat capacity. The pumping term is largest exactly where people check it least.
• Propylene rather than ethylene. Propylene glycol is more viscous than ethylene at the same concentration and temperature, so its pumping penalty is larger, and the gap widens as it gets colder. Its specific heat at the same concentration is a little higher, so its heat capacity penalty is a little smaller. Two effects pointing opposite ways, which is why you take both properties from the fluid's own table rather than assuming one glycol behaves like the other.

References

• 29 CFR 1910.1200, Hazard Communication, for the safety data sheet whose Section 8 sets glove class and eye protection for the glycol package handled
• 29 CFR 1910.147, The Control of Hazardous Energy, for isolating a circulator and relieving loop pressure before an impeller change
• Local plumbing code and water purveyor cross connection requirements for any glycol charge connection to a potable supply
• Fluid manufacturer property tables for density, specific heat, viscosity and thermal conductivity at the actual operating temperature and concentration
• ASHRAE Handbook, Fundamentals, for secondary coolant properties and the flow versus pressure drop relationships used here