Why Electrifying Heat Changes the Whole System Economics

The decisive number is not the heat pump's rated COP. It is the ratio between what the heat pump draws in January and what the array produces in January. In northern China, a rooftop array in Hebei delivers roughly 4.8–5.5 kWh per kW per day in June but only 2.4–3.0 kWh per kW per day in December and January, according to long-term monitoring published by the China Photovoltaic Industry Association (CPIA). A heat pump, meanwhile, draws 60–75% of its annual electricity in the four coldest months.
That mismatch is the entire design problem. A 10 kW array that looks generous against a 4 kW summer load becomes marginal when a 6 kW heat pump runs eight hours a day at −8°C. Systems specified on annual energy balance alone routinely fail in the first cold snap, and the customer blames the panels rather than the sizing.
The prize is still large. A heat pump delivering heat at a seasonal COP of 2.8 versus a gas boiler at 90% efficiency cuts primary energy demand per delivered kilowatt-hour of heat by roughly 55%. When 70% of that electricity comes from the roof, the marginal cost of heat collapses toward the levelized cost of solar generation — around ¥0.20–0.28 per kWh for a well-built commercial rooftop in 2025, per BloombergNEF's China levelized cost of electricity tracker.

Sizing the Array: Winter Output Is the Only Number That Counts

Design the array around December, not June. Take the heat pump's design-day load — for a 140 m² well-insulated northern home, roughly 45–60 kWh of thermal energy at −10°C — divide by the cold-weather COP of 1.9–2.2 (defrost cycles included), and you get 22–30 kWh of electricity needed in 24 hours. At a winter yield of 2.6 kWh per kW per day, that requires 9–12 kW of modules before any storage.
Most installers then over-panel deliberately. Oversizing the direct current (DC) array by 130–150% relative to the inverter's alternating current (AC) rating is now standard practice, because the inverter clips only a few dozen hours per year while the extra modules deliver real winter kilowatt-hours. The economics favor this: mono-crystalline PERC and TOPCon modules landed at ¥0.70–0.85 per watt in China in 2025, down roughly 30% year-on-year, per CPIA price surveys.
Roof orientation matters more than most sales conversations admit. A 30° south-facing pitch in Beijing will out-produce a 15° east–west array by 12–18% annually, but the east–west array spreads generation across morning and evening, which better matches a heat pump's morning recovery and evening ramp. Neither is wrong. The choice depends on whether the customer has a battery and what the local tariff looks like.

Choosing the Heat Pump: COP, Defrost Losses and Refrigerant Choice

Nameplate COP is measured at 7°C ambient and 35°C flow temperature. Real installations run at −10°C and 45–50°C flow. The gap is brutal: a unit rated COP 4.2 at standard conditions typically delivers 1.8–2.2 at design conditions in North China. Any proposal that quotes the nameplate figure without derating is not a proposal, it is marketing.
Flow temperature is the single most powerful lever. Dropping from 50°C to 35°C flow — achievable with underfloor heating or oversized radiators — lifts the COP by 25–35%. A retrofit that keeps undersized cast-iron radiators and pushes 60°C water will run at COP 1.5 and destroy the payback calculation. Retrofits should budget ¥8,000–20,000 for emitter upgrades before spending that money on extra panels.
Refrigerant and compressor type matter for cold climates. Inverter-driven Enhanced Vapor Injection (EVI) scroll compressors hold capacity down to −25°C and are the default choice in the "coal-to-electricity" (煤改电) retrofit programs across Hebei, Shanxi and Beijing. Monobloc units avoid F-gas handling on site; split systems are cheaper to service. For commercial sites above 100 kW thermal, consider cascaded or two-stage units rather than a single large machine, so partial-load efficiency stays high in the shoulder seasons.

Thermal Storage and Buffer Tanks: The Battery You Already Own

A 300–500 litre buffer tank costs ¥2,500–5,000 and often displaces ¥15,000–25,000 of lithium battery. The logic: instead of storing electricity to run the heat pump at night, run the heat pump at midday on solar electricity and store the heat in water. Round-trip thermal losses from a well-insulated tank are 3–8% per day, versus 8–12% for a lithium iron phosphate (LFP) battery plus inverter conversion losses of 6–10%.
On a typical northern residential retrofit, a 400 litre tank charged to 55°C between 10:00 and 15:00 carries the house through the evening peak. This is the highest-return component in the entire system, and it is the one most often omitted because it is invisible on a sales sheet.
Choose tanks with stratified inlet diffusers, not simple dip tubes. Stratification keeps the top of the tank hot enough for domestic hot water while the bottom stays cool enough for the heat pump to run at a low condensing temperature — the same 25–35% COP gain mentioned above, achieved for free. Add a 3–6 kW electric immersion element as a backup for the coldest 5% of hours rather than sizing the heat pump for design-day peaks.

Inverter Architecture: AC Coupling, DC Coupling or Hybrid

Three architectures dominate. DC coupling ties the array and battery to one hybrid inverter — highest efficiency, single point of monitoring, but the inverter must be chosen before the battery and locks the customer into one vendor's . AC coupling keeps a conventional string inverter and adds a battery inverter later — better for staged upgrades, slightly lower round-trip efficiency. Fully separate systems with a dedicated battery inverter and a separate photovoltaic (PV) inverter are the most flexible and the most expensive.
For heat pump integration specifically, AC coupling has an underrated advantage: the heat pump is a large, surge-prone load with inrush current on compressor start. Keeping it on the AC side, behind its own breaker and soft starter, avoids perturbing the maximum power point tracking (MPPT) algorithm on a tightly coupled DC bus. Technical documentation on hybrid inverter topologies at inverter selection covers this in more detail.
Check two ratings on any hybrid inverter before purchase: continuous AC output at 45°C ambient (many datasheets derate above 40°C, which matters when the inverter sits next to a hot buffer tank), and the backup or off-grid transfer time. A 20 millisecond transfer keeps a heat pump running through a grid outage; a 4 second transfer will trip the compressor and force a restart cycle.
| Configuration | Typical capex (20 kW PV, 5 kW HP) | Self-consumption | Notes |
|---|---|---|---|
| PV only, no storage | ¥75,000 | 35–45% | Heat pump runs mostly on grid at night |
| PV + 400 L buffer tank | ¥80,000 | 55–65% | Highest return per yuan |
| PV + buffer + 10 kWh LFP | ¥98,000 | 70–80% | Enables peak shaving and backup |
| PV + 20 kWh LFP, no buffer | ¥118,000 | 68–75% | More capex, more cycling loss |

Battery Sizing and Time-of-Use Arbitrage Under Provincial Tariffs

Battery capacity should be set by the peak–valley spread, not by the desire for independence. Provincial industrial and commercial tariffs in 2025 show spreads of ¥0.55–0.95 per kWh between valley and peak periods in Guangdong, Zhejiang and Jiangsu, and ¥0.45–0.70 in Hebei and Shandong. With a 90% depth of discharge and 88% round-trip efficiency, each kilowatt-hour of LFP storage earns roughly ¥0.40–0.75 per cycle, delivering ¥1,100–2,000 per kWh of installed capacity per year on a two-cycle day.
Residential tariffs are weaker. Tiered household pricing with optional peak–valley metering gives spreads of only ¥0.25–0.40 per kWh in most northern provinces, so household batteries rarely pay back on arbitrage alone — they pay back on heat pump load shifting plus outage resilience. That is a different justification and should be presented honestly.
Storage pricing has moved fast enough to change the answer. BloombergNEF's annual battery price survey put average pack prices at USD 115 per kWh in 2024, with China-made LFP cells at the bottom of the range; system-level installed costs for residential storage landed at ¥1,300–1,700 per kWh in 2025. Compare that against a nine-year system life at 5,000 cycles and the arithmetic starts to work for commercial sites, less so for single-family homes. Details on cell chemistry trade-offs are available at battery storage technology.

A Hebei Case Study: 20 kW PV, 5 kW Heat Pump, 20 kWh Storage

A detached 180 m² farmhouse in Baoding, Hebei, was retrofitted in 2024 under a provincial coal-to-electricity programme. The engineering team installed 20 kW of TOPCon modules on a 25° south roof, a 5 kW EVI air-source heat pump with 45°C design flow into upgraded fan-coil units, a 400 litre stratified buffer tank, and 20 kWh of LFP storage behind a 10 kW hybrid inverter.
Measured results over 12 months: 26,400 kWh generated, of which 17,900 kWh was self-consumed (68%). The heat pump drew 9,150 kWh for the year, delivering an estimated 24,700 kWh of heat at a seasonal COP of 2.7. The household previously burned 2.8 tonnes of coal per winter plus 2,100 kWh of grid electricity. Total project cost was ¥142,000, offset by ¥31,000 in provincial and municipal subsidies, for a net ¥111,000. Annual savings versus the coal-plus-grid baseline came to ¥12,400, giving a simple payback of 8.9 years — and the customer's indoor winter temperature rose from 15–16°C to 21–22°C.
One design decision drove most of the value. The original proposal used 30 kWh of battery and no buffer tank, at ¥168,000. Moving battery capacity into a water tank cut capital cost by ¥26,000 and, because the tank let the heat pump run in the middle of the day at higher ambient temperature, improved the seasonal COP from a modelled 2.4 to a measured 2.7. Similar configurations are documented in the project portfolio.

Commissioning Pitfalls That Cost Real Money

The most common failure is a heat pump fighting a thermostat. Weather-compensated curve control that resets flow temperature against outdoor temperature is mandatory; fixed 55°C flow turns a COP 3.0 machine into a COP 1.9 machine and no amount of extra PV fixes it. Set the curve during commissioning with the buffer tank in the loop, not bypassed.
Second: undersized pipework. A 5 kW heat pump at a 10°C temperature differential needs roughly 0.72 m³/h of flow. Reusing 15 mm copper from a gas boiler installation typically produces a 20°C differential, doubling pumping energy and tripping high-limit cutouts. Replace the main runs to 22 mm or larger on retrofit jobs.
Third: forgetting that PV production and heat demand peak at different times on clear winter days. Without a tank or battery, the midday solar surplus exports at a feed-in tariff of ¥0.25–0.35 per kWh while evening peak imports cost ¥0.85–1.05. That arbitrage loss alone can erase two years of expected payback. High-efficiency modules and the associated balance-of-system choices are covered at solar panel selection.

Procurement Checklist: Five Decisions to Lock Down First

Fix the winter electricity balance before signing anything. Build a 24-hour December load profile for the heat pump at design outdoor temperature, derate the array for snow and low sun angle, and confirm the shortfall is covered by buffer tank plus grid. Everything downstream depends on this number.
Specify flow temperature explicitly in the contract. Aim for 45°C or lower at design conditions, and require the installer to demonstrate the weather-compensation curve at handover. If the existing emitters cannot support it, budget the emitter upgrade in the same phase rather than adding panels later.
Size the buffer tank before the battery. A 300–500 litre stratified tank delivers more usable energy per yuan than an equivalent battery in almost every northern climate, and it costs a fraction as much.
Demand a hybrid inverter with at least 15 millisecond backup transfer, 45°C continuous rating, and a documented ability to run an inductive compressor load in island mode. Ask for the surge rating, not just the continuous rating.
Insist on metered commissioning data for 30 days. Seasonal COP measured over a real December, self-consumption ratio, and export volume give the customer a verifiable performance baseline — and give the installer an early warning if the control curve is wrong. For system design reviews on larger commercial projects, talk to our engineering team.