Water expands, and it will not compress
Water expands by roughly four per cent as it goes from cold fill to normal heating temperature. In an open system that expansion simply raises the level in a tank. In a SEALED system there is nowhere for it to go, and because water is effectively incompressible, the pressure rises until something relieves it.
An expansion vessel gives it somewhere: a flexible diaphragm with a gas cushion behind it, pre-charged so that at the system's cold fill pressure the vessel is empty and at its maximum working pressure it is full.
The sizing trap is that the vessel is not the size of the expansion. Only part of a vessel's volume is usable, because the gas cushion has to be compressed from the pre-charge to the maximum pressure and that compression is what accepts the water. The usable fraction — the ACCEPTANCE FACTOR — is typically a third to a half, so the vessel is two to three times the expansion volume it has to absorb.
That means the vessel's size depends on the system's pressures as much as on its water content. Raising the cold fill pressure without repricing the vessel reduces the acceptance factor and can leave a correctly-specified vessel too small — which is why a system that has been re-pressurised after a repair sometimes starts discharging through its relief valve on every heat-up.
- V_e
- expansion volume: system content times the fractional expansion over the temperature range
- P_pre
- vessel pre-charge, matched to the cold fill pressure
- P_max
- maximum working pressure, below the relief valve setting
- V_t
- vessel volume — always larger than V_e, usually by two to three times
Makeup water is a fault report
A sealed heating circuit is a closed loop. Once filled and vented it should consume NO water at all, for years.
So a system that needs topping up regularly is telling you something: it is leaking, and the leak is being replaced by fresh water. That matters far beyond the water bill, because fresh water is not inert. Every litre brings dissolved OXYGEN, which is what corrodes steel, and hardness, which is what scales heat exchangers.
A circuit that has been filled once reaches an equilibrium — its oxygen is consumed in the first weeks and corrosion then effectively stops. A circuit continuously topped up never reaches that state and corrodes for its whole life, which is why systems with a history of makeup are the ones full of black magnetite sludge, with blocked radiators and failed pumps.
That inverts what a makeup water calculation is for. It is not sizing a supply; it is quantifying a defect. A figure of any size is a reason to find the leak, and the common causes — a weeping relief valve, a failed expansion vessel discharging on every cycle, an automatic filling loop left open — are all things that hide because the automatic top-up conceals the symptom.
Galvanic corrosion is decided by area as much as by metal
Two dissimilar metals in contact with a shared electrolyte form a cell, and the less noble one corrodes. The galvanic series ranks which is which, and that ranking is the part everyone knows.
The part that decides the severity is the AREA RATIO. The corrosion current is shared over the anodic surface, so a small anode connected to a large cathode concentrates the whole reaction into a small area and destroys it quickly. The same pairing reversed — a large anode and a small cathode — spreads the same current thinly and is often harmless.
The practical consequence is asymmetric in a way people do not expect. A steel fitting in a copper system is a small anode against a large cathode and will fail; a copper fitting in a steel system is a small cathode against a large anode and is comparatively benign. The materials are identical in both cases and the outcomes are not.
The electrolyte matters too, and the published series is quoted for a specific one. Rankings shift between fresh water, seawater and soil, and some pairs reverse. Fittings that isolate the metals electrically, or a sacrificial anode deliberately placed as the small-noble-metal victim, are the two standard answers, and both are design decisions rather than installation details.
Refrigerant lines: the charge belongs to the length
A split system leaves the factory charged for a stated line length. Beyond it, refrigerant has to be ADDED at a published rate per unit length, because the extra pipe contains liquid that the factory charge did not allow for.
Both errors are damaging and they are not symmetric. Undercharge starves the evaporator, reduces capacity and returns superheated gas that gives the compressor no cooling; overcharge floods the evaporator and can return LIQUID to the compressor, which is a mechanical failure rather than an efficiency one. Neither shows on a thermostat, and a system running a third below capacity looks like a system that was undersized.
Length also derates capacity independently of charge. Pressure drop along the lines costs performance, and vertical separation costs more still because the compressor has to lift liquid or return oil against gravity. Manufacturers publish derating tables against equivalent length and against lift, and a long run to a convenient outdoor position can cost a meaningful share of the unit's rated output.
The equivalent length is what matters rather than the measured one. Every bend, trap and service valve contributes, exactly as it does in the head-loss paper — so a run with many bends is longer than the tape says, and the derating is read against the equivalent figure.
A suction line can be too large, which reverses the usual rule
Elsewhere on this site the advice is consistent: a larger pipe has less friction and is better. Refrigerant suction lines are the exception, and the reason is OIL.
A compressor's lubricating oil circulates with the refrigerant. It has to be carried back to the compressor by the gas, and it is carried by VELOCITY — the gas dragging it along the pipe wall. Oversize the line and the velocity falls below what is needed to move it, particularly up a vertical riser, and the oil stays where it is.
The failure is delayed and total. Oil accumulates in the evaporator and the lines, which both reduces heat transfer and starves the compressor of lubrication, and the compressor eventually seizes. Nothing about the symptom points at the pipe size that caused it.
So refrigerant line sizing is bounded on both sides: large enough that pressure drop does not cost capacity, small enough that velocity returns the oil at the LOWEST load the system will run at — which is the demanding case, because a modulating system spends most of its life well below full output. That is also why double risers and traps exist, and why a line size taken from a table for the nominal capacity can be wrong for a variable-speed machine.
Heat pumps: the capacity falls as the demand rises
A heat pump moves heat from outside to inside, and the colder it is outside the less heat there is to move and the harder the machine works to move it. Its output falls as the outdoor temperature falls — and the building's demand rises at exactly the same time.
The two curves cross at the BALANCE POINT: the outdoor temperature at which the heat pump exactly meets the load. Above it the machine modulates or cycles; below it something else has to make up the difference, which is what supplementary heating is for and why it is sized from the gap rather than from the whole load.
This changes the sizing rule compared with a boiler. A boiler's output does not vary with the weather, so oversizing it costs efficiency through cycling and nothing else. A heat pump sized on peak load is heavily oversized for the mild weather where it does most of its annual work, and an oversized inverter-driven machine that cannot modulate low enough cycles — which costs efficiency and compressor life together.
So the sizing answer is not the peak. It is a selection against the load curve, with a deliberate decision about the balance point and what covers the tail below it, and a design flow temperature as low as the emitters permit — because a heat pump's efficiency depends more on the temperature it has to reach than on almost anything else.
The outdoor unit needs its air back, and it needs it cold
An air-source outdoor unit rejects heat to the air around it. If the air it discharges finds its way back to its own inlet, it is working against its own exhaust, and its performance collapses — in heating, the reverse: it recirculates its own chilled discharge.
That is why clearances are published for each face separately and are not symmetrical. Inlet faces need free air; the discharge face needs somewhere for the air to GO. Two units facing each other across a narrow gap, or a unit in a recess with a decorative screen, are the standard ways of building the problem in, and they are usually done for appearance after the equipment was selected.
The pad has two jobs beyond carrying the weight. It holds the unit above the local snow line and above standing water, because a coil buried in drifted snow cannot breathe; and it accepts the DEFROST meltwater, which in heating mode is continuous in cold damp weather and which will build into an ice mound under an inadequately drained unit.
Service access is the third dimension and the one that only becomes visible years later. A unit installed with its service face against a wall can be maintained only by being moved, which turns a routine job into a lifting operation. The footprint required is the equipment plus its airflow clearances plus the space a technician needs, and the pages here return that combined figure rather than the machine's own dimensions.
Calculators that use this method
Basis
- ASHRAE Handbook, HVAC Systems and Equipment — hydronic system expansion tank sizing, including the acceptance factor and pre-charge relationship.
- BS 7593 and equivalent guidance on the preparation and maintenance of sealed heating systems, including oxygen ingress as the driver of magnetite formation.
- ASTM G82 and the galvanic series in seawater and fresh water, and the area-ratio effect on galvanic corrosion rate.
- ASHRAE Handbook, Refrigeration — refrigerant line sizing for pressure drop and for oil return velocity, double risers, and equivalent length of fittings.
- Manufacturers' published line-length derating tables, per-metre charge adjustment rates and maximum lift for split systems.
- AHRI 210/240 and EN 14511 rating conditions, and published capacity-versus-outdoor-temperature curves — the basis of the balance point discussion.
- ACCA Manual J for the design heating load and Manual S for equipment selection against it rather than against the peak alone.
- Manufacturers' installation clearances for outdoor units, and guidance on mounting height above snow line with drainage for defrost water.