Defrost: the blind spot in heat-pump ratings
In the first article, we showed that a heat pump's performance is a curve, not a straight line between two points. But there is a penalty almost no spec sheet honestly quantifies, one you won't read in the COP "at 47" or the COP "at 17": defrost.
Defrost is real, large, and different from one machine to the next, yet ratings dilute it and most calculation tools ignore it or staple a flat percentage onto it. It's the blind spot of winter performance in any cold climate. Here is why it's so hard to pin down, and why estimating it more accurately changes the savings picture you present to a client.
Why frost forms
Frost appears when the outdoor coil surface is both below freezing and below the air's dewpoint. The moisture in the air then deposits as ice. It isn't only about cold: it depends on air temperature, relative humidity, air velocity, and the cleanliness and surface condition of the coil.
The counter-intuitive part: the worst frosting isn't in dry, deep cold, but in a humid band around freezing, roughly −5 to +5 °C. That is exactly the most frequent range of a cold, humid winter: not the rare extremes, but the thousands of "ordinary" hours when the air is damp and the temperature hovers near zero. Defrost therefore strikes where the most heating hours are.
The defrost cycle: a double penalty
To melt that ice, the heat pump reverses its cycle: it switches to cooling mode to send hot refrigerant through the outdoor coil. During those few minutes, several losses stack up:
- it no longer heats the building;
- it pulls heat from indoors to thaw the outdoor coil;
- the compressor keeps drawing power;
- backup heat (resistance) often switches on to temper the cold air being blown and avoid discomfort.
That's a double, even triple penalty: heat not delivered, heat removed from the house, and energy spent, at the very moment the building needs it most. A sheet that sums all of this into a single COP simply cannot show it.
Why the impact changes from one machine to the next
This is the most misunderstood point: the defrost penalty is not a property of "heat pumps" in general, it's a property of this machine, in this climate.
The defrost strategy alone changes everything. Time-initiated defrost fires on a fixed schedule, frost or no frost, sometimes thawing for nothing, sometimes too late. Demand defrost tries to fire only when frost warrants it. Recent work shows the optimal moment to start a defrost varies from about 22 to 65 minutes depending on conditions: an "aggressive" control may defrost every 45 minutes, while another enforces 4 hours between cycles. At the same outdoor temperature, two units can therefore take very different penalties.
| Factor | Why it changes the penalty |
|---|---|
| Strategy (timed vs demand) | Sets how often, and how aptly, cycles fire |
| Coil and fin geometry | Affects frost build-up and water retention |
| Refrigerant | Alters low-temperature behaviour |
| Modulation (inverter vs single-stage) | Changes coil surface temperature and frosting frequency |
| Local site humidity | Damper air near 0 °C frosts the coil faster |
Two units with identical 17/47 ratings can therefore behave very differently in the field, purely because of defrost.
What the ratings don't tell you
AHRI 210/240, the standard behind HSPF and HSPF2, doesn't ignore defrost entirely: it includes a frosting test, the H2 point, around 35 °F (~1.7 °C), at a single humidity condition. But that one result is then folded into a composite seasonal factor computed for a reference region. Defrost ends up diluted into a single seasonal number, at one humidity, with one unit's defrost logic.
The upshot: extended sheets rarely detail defrost, manufacturers themselves struggle to quantify it, and the gap with the field is large. The literature is blunt: during frosting, COP can fall on the order of 40 % and capacity by more than 40 %; and neglecting frosting in design can "strongly overestimate performance," with total costs up to ~70 % higher once defrosting is accounted for. A rating isn't built to capture this: it's there to compare units, not to predict your bill.
The trap for savings calculations
Defrost concentrates in the most frequent temperature band of winter. A calculation that ignores it, or staples on a single flat percentage, overstates the heat actually delivered, understates the backup needed, and shifts the balance point. Every error points the same way: optimistic. You then promise savings that are too good and a payback period shorter than reality.
On the efficiency side, it's the twin of the 17/47 error on the capacity side: invisible on the sheet, very real on the bill.
What Optima does
Optima doesn't make defrost disappear, and it doesn't claim to compute it perfectly: it estimates it more accurately than approaches that ignore it or flat-rate it. In practice, the engine applies a defrost correction that depends on both temperature and the machine, calibrated against real data, on top of the capacity and COP curve, not instead of it. That correction is integrated hour by hour against the site's real weather, and it comes with an uncertainty estimate.
The effect then flows where it matters: to energy, usable capacity, the balance point, the amount of backup and, in the end, the payback period. The exact method, the shape of the correction and its parameters, is part of M-itech's know-how and is not published. What matters here is that, instead of being ignored or approximated with a single stroke, defrost is estimated with its machine dependence and its uncertainty.
Takeaways
- Frost is worst in a humid band around 0 °C, the most frequent range of many cold-climate winters.
- Defrost is a double penalty: heat not delivered, heat removed from the building, and energy consumed.
- The impact depends on the machine (defrost strategy, coil, refrigerant, modulation) and the site (humidity), not temperature alone.
- Ratings (AHRI 210/240, HSPF/HSPF2) dilute it into a single seasonal number; the gap with the field is large.
- Ignoring or flat-rating defrost makes the savings calculation optimistic and shortens the payback period you present.
- Optima estimates it more accurately: a machine-dependent correction, on top of the COP curve, integrated hour by hour, with its uncertainty (proprietary method, not shown).
This article explains defrost conceptually; Optima's defrost model (the shape of the correction and its parameters) is proprietary and not shown. References: AHRI 210/240 (H2 frosting test, HSPF/HSPF2); field studies on frosting–defrosting of air-source heat pumps (MDPI, ScienceDirect, Purdue). Figures are order-of-magnitude; the real penalty depends on the machine and the site.