Identical at Design, Different All Season: Inverter, On/Off, Tandem

Three heat pumps selected for the same cooling capacity cannot be told apart at the 35 °C design point: all three deliver 20–21 kW. Yet when outdoor air falls to 27 °C, one reaches a compressor COP of 6.4 while the others stay at 4.9. Same unit, same day; the only difference is how compressor capacity is controlled.

In a heat pump this capacity can be controlled in three ways: a variable-speed (inverter) single compressor, a fixed-speed (on/off) single compressor, or an asymmetric tandem that stages two compressors of different size. In this article we computed all three in HVACinOne on a heat-pump heat recovery (HPHR) unit [13]. Everything except the compressor was kept strictly identical.

At the design point (35 °C outdoor, 18 °C supply) all three deliver a similar capacity. But as outdoor air changes, their behaviour separates clearly: the inverter holds the target supply temperature continuously, while the fixed-capacity systems miss the target and fall into on-off cycling. The comparison therefore has to be made across the whole cooling season, not at a single full-load point.

To make the comparison fair we selected all three systems for the same design point: 35 °C outdoor air, 100% fresh air, 8000 m³/h, 18 °C supply target. Each was sized to deliver the same cooling capacity at that point:

• Inverter: ANB66FBZMT (variable speed; 45 Hz at design)
• Single on/off: ZH15K1P-TFM (fixed 50 Hz scroll)
• Asymmetric tandem: ZH06K1P-TFM + ZH09K1P-TFM (two scrolls of different size, running together)

The function is HP-HR-R in all three cases (rotary heat recovery + single-circuit heat pump). Apart from the compressor the unit is identical: same rotor, same evaporator/condenser coils, same fans.

Three systems, identical at a single design point

At the design condition, the mechanical (compressor) cooling capacity of the three systems is virtually the same:

Table 1 — Compressor capacity, operating point and compressor COP of the three systems at the 35 °C / 18 °C supply design point.

All three provide ~20–21 kW of cooling; they were selected for the same capacity, after all. All three bring the supply air to 18 °C, operate at a similar point (~49 °C condensing), and their compressor COPs are close (~4.0–4.4). Compressor performance is published in catalogues as a function of evaporating and condensing temperature [1], [2]. In short, at the design point the three systems can be treated as equivalent on the compressor side.

What this means: Compared at a single design point, the question "which one is better" has no answer; the three are equivalent. But a unit does not operate at a single point; it operates across a cooling season of changing outdoor air. That is where the difference begins.

Process view: what happens off design?

When all three try to hold the same 18 °C supply across 27–50 °C outdoor air, they separate. Two questions are worth asking together: what supply temperature does each system produce, and at what efficiency?

What is supplied: By modulating its frequency, the inverter holds the supply flat at 18 °C across the whole range. The single on/off and the asymmetric system are pinned to a single capacity and cannot hold it. At high outdoor air (>34 °C) the load exceeds their capacity and the supply rises above 18 °C (~20 °C at 50 °C outdoor); at low outdoor air they overcool and drop below 18 °C.

For this target the single on/off and the asymmetric system almost coincide. The 18 °C load demands nearly full capacity, so the asymmetric unit must run both compressors as well.

At what efficiency: Here is the critical finding. At low and medium load the inverter is clearly more efficient (COP 6.4 at 27 °C against 4.9 for the others). Lowering the frequency also lowers the condensing temperature, so the compressor works at a milder pressure ratio. The curves cross at around 38 °C; at very high load the inverter is pushed above nominal frequency and its COP falls below the others. Yet in that region it is still the only system that holds 18 °C.

Capacity coverage: Holding 18 °C at 50 °C outdoor requires ~25 kW of cooling, and only the inverter delivers it. The single on/off saturates at ~19 kW and its supply stays at 20 °C. So in the region where the others cannot keep up, the inverter serves roughly one third (~33%) more capacity.

What this means: Three systems that are equivalent at the design point separate as outdoor air changes. The inverter's advantage comes down to two concrete gains: (1) it can hold the target, and (2) it draws less power at part load, because it trims its frequency to the load and runs at a lower condensing temperature. Both have the same root: the inverter matches its capacity to the load continuously; the others cannot.

This is also why the industry's part-load metrics weight a part-load profile instead of a single full-load point. For this equipment class the metric is IEER [3]; on the chiller side it is IPLV [4].

The compressor does only a third of the job

Stepping back to total cooling output makes the compressor's role smaller still. At the 35 °C design point, most of the cooling comes not from the compressor but from rotary heat recovery [5], [6]:

• Total cooling: 65.7 kW
• Of which 44.9 kW (68%) comes from the rotor (pre-cooling the fresh air with the return air)
• The compressor's (DX) share is only 20.8 kW (32%)

One conceptual trap must be avoided here: heat recovery and heat-pump efficiency are measured separately in the standards and are not added into one figure. The rotor's contribution is not a COP but a heat recovery effectiveness, the ratio of recovered energy to the maximum recoverable [6], [7], [8]. The compressor's efficiency is the classic COP [9]; here ~4.36.

Dividing total cooling (65.7 kW) by total electrical power (7.8 kW) to get something like a "~8.4 system COP" is common but misleading, and it is not a standard definition. It presents heat the rotor recovers for free as if it were mechanical work done by the compressor. The correct reading is this: the rotor carries most of the load through effectiveness, and the compressor closes the remaining small trim load at a COP of ~4.36.

What this means: With heat recovery in place, the compressor already sees a small and variable load. That makes capacity control, i.e. holding the compressor at the right point, even more decisive.

In the field: on/off cannot stay at the ideal point

Everything above assumes steady-state operation. In the field the picture turns further against on/off, because a fixed-speed compressor cannot stay at its ideal operating point.

Cycling and the anti-short-cycle delay → supply oscillation

An on/off compressor stops when the supply (or return) target is reached and restarts once the temperature has risen somewhat. But to prevent short cycling, the controls add an operating time delay (anti-short-cycle timer). The compressor manufacturer stresses that, especially on packaged units with large compressors controlled from supply air, short cycling creates wide swings in supply temperature and that this delay should be written into the specification [10]. So even if the intent is "restart at 19 °C", the compressor restarts at 20 °C because the delay has not expired.

The result is not a narrow band around the target but an oscillation widened by the timer. The inverter holds the target flat; since it never stops, there is neither oscillation nor delay.

Inrush current and cycling losses

At every start the compressor draws an inrush current (locked rotor amps, LRA) several times its running current; the manufacturer requires the contactor to be rated for it and recommends starting parallel compressors one at a time [14]. The energy (kWh) cost of that pulse is small, because it lasts a fraction of a second. Its real price is the electrical demand peak (kVA/demand charge) and mechanical wear on the contactor and compressor.

The actual energy penalty at part load comes from cycling itself, and it can be quantified. The standard part-load model assigns an efficiency loss to each cycle: PLF = 1 − Cd·(1 − PLR). Here PLR is the part-load ratio and Cd the cyclic degradation coefficient; it is typically 0.15–0.25, with 0.25 as the default when not measured [11].

This model and its 0.25 default are defined for small packaged units and are carried over conceptually to the larger unit examined here. AHRI 340/360, which rates part-load efficiency for this class, treats part load with four weighted load points (100/75/50/25) instead [3]. The frameworks differ, but both agree that cycling carries an efficiency penalty.

On this basis, at 50% part load (PLR = 0.5) an on/off compressor delivers the same cooling with about 8–14% more energy, and the penalty grows as the load lightens. The inverter modulates and never cycles, so it does not pay this penalty.

The asymmetric tandem stages more finely, but not at this target

Two compressors of different size offer three capacity stages: small only (ZH06 ~11 kW), large only (ZH09 ~14 kW), or both together (~21 kW). This covers the load more finely than the single on/off with its one stage. However, holding 18 °C requires nearly full capacity, so a single stage is not enough and both compressors must run together.

At this target, therefore, the asymmetric unit has no practical advantage over the single on/off. The benefit of three stages appears under lighter loads, where a single small stage is enough. This is worth stressing: on the main operating line of this article (8000 m³/h, 18 °C target) the asymmetric system runs both compressors at every outdoor condition. The rotation penalty described below belongs to a lighter load regime in which a single stage would suffice, not to that line.

The equal-ageing rotation trap

Most multi-compressor systems include a "life extension" function: equal-ageing rotation, which runs the compressors in turn to equalise their operating hours. This is a common control practice in multi-compressor arrays [12]. But this logic optimises equal ageing, not capacity, and at part load it may pick the wrong compressor.

Our data show this clearly. When a light load should be met by the correctly sized small compressor (ZH06) but rotation stages the large one (ZH09), the same cooling is delivered with 12–15% more power.

Table 2 — COP and extra power for the same cooling when rotation brings in the large ZH09 instead of the correctly sized ZH06 at part load.

The reason is again the operating point. The larger compressor loads the same coil harder, pulling evaporation down and pushing condensation up: at 27 °C, ZH06 runs at TEvap 13.8 / TCond 35.6 °C while ZH09 runs at TEvap 12.5 / TCond 37.8 °C. That drives the compressor to a worse pressure ratio and a lower-COP point. The irony is plain: a rotation scheme meant to extend compressor life burns extra power at part load and puts the large compressor through unnecessary cycling.

What this means: The field losses of on/off and multi-compressor systems, even when they are equivalent at the design point, come from their inability to stay at the intended operating point. Cycling, timer oscillation, inrush current, coarse capacity staging and rotation all push the compressor away from its efficient point.

Why the inverter is exempt

Examined one by one, the inverter's advantage always comes back to the same cause: it can hold its operating point.

• It modulates continuously, so it holds the target flat; there is no oscillation.
• It starts once, softly, and keeps running; there is no inrush current pulse.
• It sets its frequency in proportion to the load, so at part load it cools exactly as much as needed; no coarse stages and no overcooling.
• It is a single compressor, so there is no rotation/equal-ageing penalty.
• At part load it lowers frequency and therefore condensing temperature, so it runs at a mild pressure ratio and a high COP.

None of this is "the inverter compressor being more efficient". All of it follows from the inverter's ability to keep the compressor at the right operating point.

Conclusion: you select a control strategy, not a compressor

Equalising the three systems at the same 35 °C design capacity and following them across the cooling season gives a clear picture:

• At the design point all three are equivalent: similar capacity, similar operating point, close COP.
• What creates the difference is not which compressor is used but how the load is met: capacity control.
• Heat recovery reduces the compressor to a small trim load; the real leverage is holding that load at the intended point.
• The inverter wins, not because it is an "efficient compressor", but because it holds its operating point across a wide load range and avoids the operational penalties seen in the field.

Practical conclusion: an inverter should be evaluated not as "automatic efficiency" but as a control strategy that provides operating-point management (modulation range + continuous operation). For a fixed-speed or multi-compressor system, the outcome is not decided by the compressor itself. It is decided by how far sizing, cycling/timer settings and rotation logic degrade capacity control.

Note (method): The steady-state values (capacity, COP, operating point, extra power under rotation) were computed in HVACinOne on the same unit, with the HP-HR-R function, using three compressor configurations matched to the 35 °C summer design point [13]. Transient and control effects such as cycling oscillation, anti-short-cycle delay and inrush current are outside the scope of this steady-state engine; they are addressed through engineering judgement and manufacturer application data.

References

1. CEN, EN 12900 – Refrigerant compressors: rating conditions, tolerances and presentation of manufacturer's performance data, 2013, https://www.en-standard.eu/bs-en-12900-2013-refrigerant-compressors-rating-conditions-tolerances-and-presentation-of-manufacturer-s-performance-data/
2. AHRI, AHRI Standard 540 – Performance Rating of Positive Displacement Refrigerant Compressors and Compressor Units, 2020, https://www.ahrinet.org/system/files/2023-06/AHRI_Standard_540_(I-P_and_SI)_2020_Standard_for_Performance_Rating_of_Positive_Displacement_Refrigerant_Compressors_and_Compressor_Units.pdf
3. AHRI, AHRI Standard 340/360 – Performance Rating of Commercial and Industrial Unitary Air-conditioning and Heat Pump Equipment, 2022, https://www.ahrinet.org/system/files/2023-06/AHRI%20Standard%20340-360-2022%20(I-P).pdf
4. AHRI, AHRI Standard 550/590 – Performance Rating of Water-chilling and Heat Pump Water-heating Packages, 2023, https://www.ahrinet.org/system/files/2023-06/ANSI%20AHRI%20Standard%20550%20590-2023%20(I-P).pdf
5. ASHRAE, ASHRAE Handbook – HVAC Systems and Equipment, chapter "Air-to-Air Energy Recovery Equipment", n.d., https://www.ashrae.org/technical-resources/ashrae-handbook
6. Eurovent, Eurovent 17/11 – Guidelines for Heat Recovery, 2015, https://www.eurovent.eu/wp-content/uploads/publications-files/eurovent-rec-17-11-eurovent-guidelines-for-heat-recovery-2015-en.pdf
7. CEN, EN 308 – Heat exchangers: test procedures for establishing performance of air-to-air heat recovery components, 2022, https://www.en-standard.eu/bs-en-308-2022-heat-exchangers-test-procedures-for-establishing-performance-of-air-to-air-heat-recovery-components/
8. European Commission, Commission Regulation (EU) No 1253/2014 – Ecodesign requirements for ventilation units, 2014, https://eur-lex.europa.eu/eli/reg/2014/1253/oj/eng
9. CEN, EN 14511-1 – Air conditioners, liquid chilling packages and heat pumps: terms and definitions, 2022, https://standards.iteh.ai/catalog/standards/cen/93a37503-c91b-49dc-ad23-82ec9715acf1/en-14511-1-2022
10. Copeland, AE17-1262 R2 – Compressor Short Cycling: An Unrecognized Problem, 2024, https://webapps.copeland.com/online-product-information/Publication/LaunchPDF?Index=aeb&PDF=AE17-1262_R2.pdf
11. AHRI, AHRI Standard 210/240 – Performance Rating of Unitary Air-conditioning and Air-source Heat Pump Equipment, 2024, https://www.ahrinet.org/system/files/2024-05/AHRI%20Standard%20210%20240-2024%20(I-P)_0.pdf
12. ASHRAE, ASHRAE Handbook – HVAC Applications, n.d., https://www.ashrae.org/technical-resources/ashrae-handbook
13. HVACinOne, HVAC selection and calculation software – unit technical outputs (8000 m³/h HP-HR-R; inverter / on-off / asymmetric tandem summer process and rotation calculations), 2026
14. Copeland, AE4-1303 R18 – 7 to 15 Ton ZR-KC, ZH-KC and ZP-KC Copeland Compressors (Application Engineering Bulletin), 2025, https://webapps.copeland.com/online-product-information/Publication/LaunchPDF?Index=AEB&PDF=1303

← All posts