Start with a question. Read the basics, prepare site data, or find an engineering reference.从一个问题开始：看懂原理、准备现场资料，或查找工程参考。
Choose the question you have now
For study and teaching. Paid training stays with Professional Support; this sequence is open reading.
Heat pump fundamentals
Choose a chapter below. Open only what you need, starting with the basic principle.从原理开始，选择感兴趣的一章展开阅读。
Refrigerants Compressors Heat exchangers Valves Back to Home
Part 2 · Cognition
A boiler “makes heat”. A heat pump “moves heat”. The moment you must lift heat across a temperature gap, the design space explodes—cycles, fluids, compressors, and integration details.
Small ΔT: single-stage vapor compression may work well. Large ΔT: you may need economizers, two-stage, cascade, absorption, steam compression, or other cycles to stay within safe and efficient boundaries.
Air, water, brine, flue gas, process waste heat, solvent streams… each brings different fouling, corrosion, freezing risk, and temperature stability—so the “best” cycle and heat exchanger choices change.
Hot water, hot air, steam, thermal oil—each has different temperature requirements and curve shapes. Matching those curves often matters more than the nameplate COP.
Not “pick any refrigerant”. Critical temperature, operating pressures, glide, material compatibility, flammability/toxicity class, PFAS/F-gas rules—these constraints push designs into different families.
Scroll/screw/reciprocating/centrifugal, vapor vs steam compression, oil-free vs oil-injected—each has a different “safe & efficient” map, affecting cycle choice and staging.
Plants cannot stop. Redundancy, defrosting, part-load control, bypass, thermal storage, and commissioning measurement turn a heat pump into a system engineering problem—not just a box.
Boilers are “energy → heat”. Heat pumps are “heat → higher heat”, so they must obey more boundaries—and boundaries create variants.
Want the cycle families drawn out (reverse Carnot, Brayton, VCC upgrades, transcritical CO₂, absorption, steam/MVR)? Continue in the cycle guide.
The core idea is related, but industrial heat pumps face much larger ΔT, harsher sources, stricter reliability, and often steam/hot-water targets—so the required cycles and equipment scale are different.
COP depends on your source/sink temperatures and part-load behavior. The best project is usually about process coupling and stable operating hours, not a single catalog COP number.
Part 1 · Extreme cold
— Conventional fluorocarbon cascade vs CO₂ cascade, head-to-head
North of about −20 °C, heat pumps hit brutal physics. Conventional fluorocarbon cascades (e.g. R410A/R32 + R134a) pushed into polar markets often collapse on efficiency and reliability.
As an industrial heat-pump R&D team we standardize on CO₂ (R744) + R515B full cascade: not “more metal for show”, but a consequence of thermodynamics and lifecycle economics.
| Dimension | Fluorocarbon cascade (R410A/R32 + R134a) | CO₂ cascade (R744 + R515B) | Field takeaway |
|---|---|---|---|
| Survival / trip risk at −40 °C | High trip risk Very low suction density, pressure ratio often 7–8+, discharge >130 °C risks oil carbonization and compressor damage. |
Stable envelope Near −40 °C ambient, CO₂ still holds useful evaporation-side pressure (~10 bar class), system pressure ratio ~3–4, volumetric efficiency >75%, discharge temperature stays controllable. |
Pressure ratio and discharge temperature set compressor life in industrial duty. |
| Heating fade (−25 to −30 °C) | Cliff-edge loss Capacity drops sharply; COP approaches 1.0 and the plant effectively buys resistance backup heat. |
Strong winter curve At −25 °C COP stays above 2.0; at −30 °C extremes COP still lands ~1.6–1.9 depending on design point. |
CO₂’s high suction density drives volumetric heating power in deep cold. |
| Maximum leaving-water temperature | ~50–55 °C Hard to lift in deep cold; mismatches cast-iron radiator grids in northern China. |
~80–90 °C Low-pressure stage scavenges heat; high-pressure stage rejects at useful glides—80 °C+ hot water is realistic. |
Often avoids wholesale terminal retrofit; couples to legacy distribution temperatures. |
| Environment & compliance | High-GWP HFC exposure e.g. R410A GWP ≈2088; tightening international F-gas style pressure. |
Cleaner stack CO₂ low stage (GWP = 1, ODP 0) plus upper-stage R515B class lower-GWP fluids versus legacy high-GWP-only cascades—better aligned with long-run decarbonization and F-gas trajectories (confirm charge and jurisdiction). |
Align long procurement with where refrigerant law is heading. |
Selling conventional fluorocarbon cascade units “as extreme-cold machines” often under-sizes the physics customers pay for. CO₂ cascade pays differently:
Expert line: respect the physics redline
We never tell clients “one air-source box solves everything.” In Inner Mongolia, northeast China, Xinjiang and similar climates, exceptional years can reach about −50 °C—near CO₂’s dry-ice triple-point territory. Pure air-source alone cannot be the only safety narrative.
The deliverable is both lifecycle margin and a heating supply that cannot ethically fail: finance plus an engineered availability floor.
Part 3 · Pricing & standards
People often ask what one kilowatt of heat-pump capacity “costs”. The question is fine—the missing part is which kilowatt, at which outdoor and water conditions, and whether the denominator is thermal output, electrical input, or a trade “horsepower” habit. Here is the shortest checklist.
Comparing “¥ per kW” using a severe-climate nameplate (e.g. very low outdoor temperature) against a mild-climate nameplate (e.g. +7 °C air source) is like measuring the same table with two rulers. The numbers look comparable; the products are not always in the same class.
Delivering usable heat at -40 °C outdoor versus +7 °C can imply different compressor families, reinjection/cascade choices, defrost strategy, antifreeze and exchanger sizing—and a different bill of materials. Treat that as a tier change, not a gentle discount. Use the rating table agreed for your project.
Fair comparison needs aligned source/sink temperatures, flow strategy (constant vs variable), whether pumps and auxiliaries are in the input power, how defrost is defined, and whether you compare a single rating point or seasonal/part-load curves. Delivery scope—controls, warranty, installation interface—also belongs outside a naked “per kW” sticker.
Takeaway for buyers: ask for three things first—outdoor/source temperature, supply/return water (or sink) temperature, and whether you need nominal catalog capacity or guaranteed output at a cold hour—then talk unit price.
Without defined rating conditions and a defined denominator, “money per kilowatt” is a map with no coordinates.
The unit conversion is fixed; the engineering problem is not. Heat duty at -40 °C source and +7 °C source is not the same curve—so the “same” kW never meant the same machine duty.
Labels can share similar words while differing in test tolerances, auxiliary inclusion, defrost credit, hydraulic assumptions, and what is inside the supplier’s boundary. Treat price as a bundled offer, not a scalar divided by one line on a brochure.
AI can organize comparisons; it should not replace hazard analysis, utility limits, refrigerant policy, site access, and lifecycle assumptions signed by qualified engineers.
Part 4 · AI & human roles
No—not for the parts that matter most. AI can speed up literature and table work, but refrigeration and heat-pump R&D still hinge on human judgment over rating assumptions, safety boundaries, confidential process data, and who carries legal responsibility. The same line appears in Part 3’s “Pricing & rating” FAQ (“Can AI pick the cheapest heat pump?”): AI may organize comparisons, but it should not replace hazard analysis or engineer-backed, signed lifecycle assumptions. The sections below explain why.
Models can state catalog numbers, code clauses, or cycle details that were never verified against your plant data. In R&D that misroutes compressor maps, safety margins, and test matrices—people must cross-check datasheets, experiments, and field logs.
If the premise is wrong—an impossible ΔT, a mismatched rating point, or an unsafe shortcut—the model may still sound agreeable. Engineering needs independent challenge and red-team review, not flattery.
Useful answers often want process curves, contracts, and SCADA traces that cannot leave the company unchanged. AI does not replace your rules on data custody, anonymization, and who may see which slice of a heat balance.
Hazard studies, drawings, warranties, and incident liability still attach to people and organizations. A chat log is not a signed engineering basis—and must not be treated as one.
Skipping first-principles checks blunts intuition for thermodynamic limits and system integration. The engineers who thrive are still those who can sanity-check a line of reasoning when the tool is wrong or silent.
Use AI to organize and draft; use qualified humans to judge, verify, and sign—especially wherever safety, data custody, and lifecycle risk meet the heat pump.
Part 5 · Refrigerants & policy
Comfort HVAC and industrial heat pumps share refrigerant names, but industrial plants often mean higher temperatures, larger charges, and tougher on-site rules. Policy mixes HFC phasedowns, quota-style fluorinated-gas laws, chemical controls (including PFAS debates), and national standards—and the clocks are not synchronized.
This chapter is a brief orientation only. For illustrated chapters on fluid families, safety classes, and China / EU / US policy timelines—and the interactive IIR Table 2 + ASHRAE 34 property table—see the dedicated Refrigerants guide in the Knowledge base.
Part 6 · AI industrial heat pump expert
AI industrial heat pump expert · Q&A
The contradiction shows up in real project spreadsheets—and it is usually not because heat pumps are physically doomed to be dirtier than gas. More often, teams compare different kinds of carbon, different system boundaries, or a brochure COP against field gas use.
Same operating-carbon shortcut as below: heat pump ≈ EFgrid ÷ COPsystem; gas boiler ≈ EFgas ÷ ηboiler. Illustrative factors: EFgrid ≈ 0.55 kgCO₂/kWhelectricity, EFgas ≈ 0.20 kgCO₂/kWhgas energy, ηboiler ≈ 90%. Replace with your audit values.
Quick check (illustrative factors only): breakeven system COP ≈ EFgrid ÷ (EFgas ÷ η) ≈ 0.55 ÷ 0.22 ≈ 2.5. Above that, heat pump operating carbon tends to beat efficient gas; below it, gas can win—exact crossover depends on your real EF and η.
Human version: boiler carbon is in the chimney; heat-pump carbon is on the meter and in the grid factor—if you do not line up the ledger, you are not comparing the same thing.
Engineering version: Scope 1 (site direct): gas boilers emit combustion CO₂; heat pumps have no stack—chimney-only accounting flatters heat pumps. Operating carbon: electricity × grid emission factor—where most heat-pump footprints live. LCA / corporate inventory: may add manufacturing, refrigerant leakage (GWP), logistics; “gas fuel only” is a different boundary.
Human version: a heat pump is only as clean as the power behind the plug—and the spreadsheet factor you pick.
Engineering version: Average vs marginal factors, regional vs national defaults, and outdated yearbooks all move results. Rough operating carbon intensity per kWh of useful heat: heat pump ≈ EFgrid / COPsystem; gas boiler ≈ EFgas / ηboiler (pumps, fans, defrost, standby belong in COPsystem). Low COPsystem or high EFgrid—common with large industrial ΔT—can make heat pumps lose to efficient gas. Use the factor required by your carbon audit or contract, not a generic internet number.
Human version: the COP on the brochure is not the COP your carbon spreadsheet should use.
Engineering version: Large temperature lift (steam / high-temperature water), part load, cycling, defrost, and antifreeze loops cut real performance—see Part 2 · Cognition and Part 3 · Pricing & standards. Auxiliary power (source pumps, towers, controls, trace heat) is often excluded from “compressor COP”. Matching a +7 °C comfort nameplate to a 120 °C steam heat pump’s annual curve will distort carbon results.
Boiler cases often stop at furnace efficiency. Heat-pump cases sometimes count only the compressor while omitting source-side exchange, distribution pumps, storage, backup electric or gas peaking, and tower/fan power—good equipment COP, poor system carbon.
Grey grid without green power or PPA caps electrified heat. Long hours on electric backup or gas peaking in cold windows load high-carbon energy into the heat-pump total. Condensing gas at high load factor vs a heat pump at low load compares a best boiler hour to a worst heat-pump window. For policy context on assessment, see industrial heat pump policies.
Before you declare a winner, align period, boundary, and factors:
A heat pump is not born green—greenness ≈ how clean the grid is × how high system COP is × whether the boundary is complete.
For project numbers, use the Heat Pump Benefit Analyzer with aligned boundaries; cross-check steam metrics on the industrial heat pump standards page where relevant.
Indirect emissions from electricity—and often omitted auxiliaries—are in the total. Stack zero ≠ carbon zero.
Not always. It depends on EFgrid, boiler η, boundary, and hours. In coal-heavy grids or low system COP, gas can win on operating carbon intensity.
Improve weighted system COP and process coupling, cut auxiliaries and peaking, add green power or storage, and compare with audited factors—not AI-invented grid or COP values.
Emission factors and carbon accounting rules vary by region and reporting regime; this chapter is engineering orientation, not a carbon audit or legal opinion.
Part 7 · Refrigeration & heat pump expert
Refrigeration & heat pump expert · Q&A
In the trade, “60P” sounds like a fixed size. In engineering, it is often three different things at once—and the rating temperature you never asked about can move real delivered kW by a large margin. This chapter untangles 匹 vs kW, refrigerant and operating conditions, what national standards actually say, and why two “same horsepower” quotes are not the same machine.
A common field story: two suppliers both offer 60P air-source heat pumps. One price looks “expensive”; the other looks like a bargain. After installation, the owner finds the cheaper box was rated at +7 °C outdoor air and moderate leaving-water temperature, while the premium unit was sized for −12 °C ambient and higher supply water—different hardware tiers, not the same product mis-priced. Numbers below are illustrative; always use each vendor’s rating sheet.
Same “60P” label ≠ same heating kW at your site. Higher price is not automatically margin; lower price is not automatically a steal—often it is a different rating tier. See also Part 3 · Pricing & standards (“two rulers, one table”).
Motor / frame class: OEM catalogs (Bitzer, Dorin, etc.) often use HP for motor power or product series—not guaranteed system heating kW.
Cooling-capacity shorthand: in Chinese HVAC trade, 1 匹 ≈ 2.5 kW cooling at a historical comfort rating point—already refrigerant- and condition-dependent.
Marketing label: “60P” on a brochure may summarize none of the above consistently. Before comparing ¥/匹, ask which definition the seller uses.
For the same compressor displacement, changing evaporation and condensation temperatures shifts mass flow, volumetric efficiency, and power—so heating/cooling kW moves. There is no “匹” detached from source temperature, sink temperature, and part-load point. −12 °C vs +7 °C outdoor air is not a small correction; it can change compressor family, reinjection/cascade, defrost, and heat-exchanger area.
Switching fluid—R134a, R410A, R32, R515B, R744, ammonia, etc.—changes suction density, pressure ratio, and discharge temperature. The same frame “HP class” therefore maps to different kW on the performance map. Cross-brand 匹 comparisons without aligned refrigerant and rating point are unreliable.
Chinese product and test standards express nominal capacity in kW at specified rating conditions—not “匹” as the legal unit. Examples on this site’s standards quick reference: GB/T 18430.1 (industrial/commercial water chilling & heat pump packages), GB/T 25127.1 (low-ambient air-source, cold-climate oriented), GB/T 10870 (performance test methods referenced by those families). Procurement contracts should cite kW + outdoor/source temp + supply/return water (or sink) temp, not 匹 alone.
Before ¥/匹 or ¥/kW, align:
“匹” is a spoken label; kW @ stated conditions is engineering language. Comparing 匹 without aligned rating points is ordering without units.
For ¥/kW pitfalls and boundary issues, read Part 3 · Pricing & standards; for GB/T families and test points, use the industrial heat pump standards quick reference. For machine families, maps, and oil return, see the compressor knowledge guide.
Trade habit often uses ~2.5 kW cooling at a comfort rating point—but that is not a universal heating kW. For procurement, demand the vendor’s kW table at your source/sink temperatures, not a mental conversion factor.
Standards recognize kW at defined test conditions (see GB/T 18430.1, GB/T 25127.1, GB/T 10870). “匹” remains market shorthand. Contracts should state kW, ambient/source temperature, and water temperatures—not 匹 alone.
Only as a coarse filter. Final selection needs aligned refrigerant, evaporation/condensation (or air/water) temperatures, heating vs cooling duty, and whether auxiliaries are inside the supplier boundary.
Part 3 explains why denominators misalign. This chapter explains why “匹” is one of the most confusing denominators—and why two identical 匹 numbers can imply different machines and prices.
OEM maps, standard editions, and project warranties vary; this chapter is engineering orientation for buyers and integrators, not a substitute for type-test reports or signed specifications.
Part 8 · Coupling & operating economics
Coupling & operating economics · Q&A
Human version: when gas is cheap relative to electricity, a heat pump must work harder (higher COP) to beat the boiler on the bill—and in cold or defrost hours it often cannot, so a coupled system makes sense.
Engineering version: define the gas–electric price ratio r = Pgas (¥/m³) ÷ Pelec (¥/kWh), align unit heat costs, and dispatch to whichever source wins at each hour’s system COP. This chapter is about operating money, not carbon—see Part 6 · Carbon for the parallel carbon breakeven.
Use consistent units. Natural gas lower heating value LHV ≈ 9.87 kWh/m³ (≈ 8600 kcal/m³)—confirm with your gas contract. This worked example is not a contract calorific value; compare published ranges on the energy-factor reference. Boiler efficiency ηboiler is the useful-heat fraction (include condensing recovery if present).
Intuition: higher r (gas expensive vs electricity) → lower COP* → heat pump wins more hours. COPsystem moves with outdoor temperature, defrost, and part load—coupling is about picking the cheaper source hour by hour.
Part 6 uses grid and gas emission factors (kgCO₂/kWh)—breakeven COPcarbon ≈ EFgrid ÷ (EFgas ÷ η). With illustrative factors that is often ~2.5. This chapter’s COP* uses tariffs. Example: Pelec = 0.6 ¥/kWh, Pgas = 2.2 ¥/m³, η = 90% → r ≈ 3.67, COP* ≈ 2.4—similar number, different ledger. A scheme can be cheap but carbon-heavy, or clean but expensive.
Numbers below are illustrative only—replace with contract gas price, your tariff structure, measured η, and field COP curves.
Pelec = 0.6 ¥/kWh, Pgas = 2.5 ¥/m³, η = 90%, winter average COPsystem = 3.2 → r = 4.17, COP* ≈ 2.13. 3.2 > 2.13 → prefer heat pump in those hours.
Pelec = 0.8 ¥/kWh, Pgas = 2.0 ¥/m³, COPsystem = 1.8 after defrost → r = 2.5, COP* ≈ 3.55. 1.8 < 3.55 → let gas carry this window.
Same season: ~70% of hours COPsystem > COP* (heat pump), ~30% below (boiler for defrost, peak load, or cold ambient). Hybrid dispatch is not a compromise—it is marginal-cost routing per hour.
Existing gas boiler + new air-source heat pump on the same heating main. When COPsystem > COP* and the heat pump is not maxed out → heat pump first; boiler fills the gap.
Typical in cold climates: heat pump covers normal hours; boiler covers extreme ambient, heavy defrost, and peak load—same engineering logic as Part 1 · CO₂ cascade backup, without requiring CO₂ hardware.
With valley electricity, run the heat pump hard in cheap hours (buffer tank or mild oversupply). In peak-price hours, run heat pump only if COPsystem still beats COP* at the peak tariff; otherwise shift to gas.
Industrial boundary note: high-temperature or steam duty pushes COPsystem down and COP* up—coupling and thermal storage matter more; this chapter focuses on building / district hot-water heating.
Tiered gas contracts, seasonal heating tariffs, commercial peak/valley electricity, and demand charges all move <em>r</em> and COP* through the year. Size dispatch logic for <strong>your bill structure</strong>, not a single national average.
Outdoor temperature, supply water setpoint, defrost, part load, and auxiliary power (fans, pumps, controls) set the real denominator—see Part 2 · Cognition, Part 3 · Pricing, and Part 7 · Compressor HP. Dispatch on measured or simulated hourly COP, not a +7 °C brochure line.
Even when COP<sub>system</sub> > COP*, a undersized heat pump must call the boiler for peak load. COP* tells you <strong>which fuel is cheaper</strong>; capacity tells you <strong>who can deliver the kW</strong>.
Boiler hours in extreme cold may be poor economics yet good engineering—backup is insurance, not “wasted gas”. Separate <strong>minimum reliability</strong> rules from <strong>economic dispatch</strong> in the control spec.
Include circulation pumps, flue losses, and condensing recovery in η<sub>boiler</sub>; include source-side fans, defrost, and standby in COP<sub>system</sub>. Otherwise COP* looks favorable to the side with the narrower boundary.
Before writing dispatch logic or selling “always on heat pump”:
Coupling is hourly marginal cost: compute COP* from r and η; run the heat pump when COPsystem beats COP* and capacity allows—otherwise let the boiler work.
For project economics, use the Heat Pump Benefit Analyzer with aligned tariffs and COP curves—not a single brochure point. Carbon crossover is in Part 6; extreme-cold peaking context in Part 1.
No. COP* moves with local <em>r</em>, boiler η, and sometimes hourly tariffs. Recompute per site and season.
Not always—it depends on r. Low gas price or high electricity can push COP* above 2; that is money, not the same as Part 6’s carbon breakeven.
AI can organize hourly COP and tariff tables; contract gas price, tariff riders, and signed COP test data still need qualified human review before control logic goes live.
In many regions the majority of hours still favor the heat pump once COP* is computed—hybrid captures those hours while keeping boiler capacity for peaks and cold snaps.
Gas and electricity prices, tariff rules, and measured efficiencies vary by contract and site; illustrative LHV and η values here are not audit or feasibility-study substitutes.