A ground loop is judged on two things a buried water main is not: it moves with every season, and its joints cannot be reached once the borehole is grouted. This page works through both, then gives the positions, the test sequence and the families that answer them.

dn20–dn1400electrofusion range
dn20–dn1600butt fusion range
dn32 / dn40typical borehole loop
SDR 11typical loop wall

Why a ground loop is not a buried water main

A borehole heat exchanger looks like a small water pipe in a hole, and almost every mistake in the fittings package comes from treating it as one. Three differences decide the design, and none of them is the pressure rating.

The first is that the loop moves. PE expands about 0.2 mm per metre per kelvin, which is roughly twenty times steel, so a 150 m U-tube sees several hundred millimetres of movement between a heating season and a cooling season. A loop that is fixed at the borehole head and weighted at the bottom has nowhere to put that movement, and the strain lands on the joints.

The second is that the joints cannot be reached. Once a borehole is grouted, a leaking fusion joint is not a repair, it is a lost borehole and a new one drilled beside it. That changes the economics of the joint completely: a coupling that costs a few renminbi is worth nothing next to the cost of the hole, so the specification should be about how the joint is made and tested, not about the unit price.

The third is the test sequence. The loop is pressure tested before grouting, and again after, and the first test is the last chance to find a bad joint while it can still be cut out. A package that arrives without the test fittings, the caps and the gauges is a package that will be tested late.

The one sentence worth putting in the specification. Every fusion joint below ground is tested, recorded and traceable to an operator, and no joint is grouted before the test passes. Everything else on this page follows from that sentence, and it costs nothing to write.

The range, from the U-bend to the heat pump

The groups follow the water: down a borehole, along a trench header, through a manifold, and into the plant room. The borehole end is small diameter and almost entirely electrofusion; the trench and plant ends are larger and mostly butt fusion.

The ground loop package, grouped by position
Family groupFamiliesSize rangeSize entriesWhat it covers
Borehole loop and head5dn20–dn1400141Couplings, elbows, reducers and end caps: dn32 and dn40 at the head, up to the largest electrofusion sizes where a header needs them.
Trench header and manifold6dn20–dn1600297The header that collects the boreholes, and the manifold tees in the chamber.
Branch on a built header4dn63–dn40048A saddle for a circuit added to a header that is already backfilled and tested.
Isolation2Ø32–Ø35519A valve per circuit at the manifold, with fusion ends.
Plant room entry2dn25–dn35553Steel-to-PE transitions to the heat pump, SDR11 and SDR17.
Flanged connections3dn40–dn160072Stub ends and flange adaptors at the heat pump, buffer vessel and plant skids.

Position by position, from the borehole to the plant room

Where each fitting sits, and why that one
PositionFitting that goes thereWhy this one
Borehole U-tube, dn32 or dn40Electrofusion coupling to the U-bend, or a factory-made loop with the bend already formedThe joint is made at the head, above ground, before the loop goes down. It is the one joint that has to be perfect, because it is the one that cannot be reached.
Borehole head, below the grout sealElectrofusion coupling, then a short riser to the trenchTwo loops per borehole usually, so the head is where the pair is dressed and identified before the trench is cut in.
Trench header, dn63–dn160Butt fusion tees, reducers and couplingsLong runs with room for the machine. Butt fusion gives the full pipe wall at the lowest cost per joint.
Branch onto an existing headerElectrofusion saddle, dn63–dn400 on the mainA trench that is already backfilled and a header already tested. A saddle does not cut the header.
Manifold in a chamberButt fusion or electrofusion tees, with ball valves on each circuitEach borehole has to be isolatable and flushable on its own, which means a valve per circuit, not one valve per manifold.
Isolation at the manifoldPE ball valve, fusion ends, Ø32–Ø355Fusion ends keep the valve in the same continuous material as the header, which matters where the trench may settle.
Plant room entrySteel-to-PE transition, then a flanged connection to the heat pumpThe heat pump is steel and factory-flanged. The transition is where the two materials meet, and it belongs outside the wall, not inside it.
Heat pump and buffer vesselFlange adaptor or stub end, plus a ball valveEvery plant item is isolated and removable, so every connection is flanged and every branch has a valve.
Flushing and pressure test pointEnd cap or a tee with a temporary capThe loop has to be flushed and tested from a point that can be closed afterwards. Plan it before the trench is backfilled.

Thermal movement, and why the bottom of the borehole is a joint

The number to design with is the coefficient of linear expansion: about 0.2 mm per metre per kelvin for PE100, against roughly 0.012 for steel. In a closed loop the water temperature swings between about 2°C in heating mode and 35°C or more in cooling mode, so a 20 K swing is a normal design cycle and a 30 K swing is not unusual.

Movement over a design cycle, and what absorbs it
Loop lengthMovement at 20 KAt 30 KWhere it has to go
50 m borehole200 mm300 mmDistributed along the loop, and taken up at the U-bend if the loop is free-hanging.
100 m borehole400 mm600 mmThe reason the loop must not be grouted tight against the pipe over its whole length.
150 m borehole600 mm900 mmHalf a metre of movement at the head, which is why the trench connection needs slack, not a straight pull.

In practice the movement is absorbed in three places, and it is worth checking each one on a drawing rather than assuming. The U-bend at the bottom is the natural expansion loop, and it works only if the loop hangs free rather than sitting on the borehole base. The connection at the head needs a slack coil or a shallow offset before it enters the trench, so the movement does not pull on the first fitting. And the trench header, which runs at a much smaller temperature swing, still needs the same treatment at each end.

The joints themselves are not the problem when the geometry is right. A fusion joint is as strong as the pipe in tension, so the loaded part is the pipe wall next to it, which is what you want. What fails is a joint that is asked to act as an anchor: a transition set in concrete, a valve clamped to a chamber wall, a loop tied to a rebar cage. Those are the details worth deleting from a drawing.

The test sequence, which is the whole quality plan

A ground loop is one of the few systems where testing is more important than material selection, because the failure is unreachable. The sequence below is the one that finds a bad joint while it can still be cut out.

Test stage, pressure, and what it catches
StageTest pressureWhat it catches
After assembly, before the loop goes into the hole1.5× design, 1 hA joint made badly at the head, which is the only joint that can still be cut out and remade at this point.
Loop in the borehole, before grouting1.5× design, hold through groutingDamage from handling and from the trip down the hole. Leaving the loop under pressure while the grout sets is the standard practice and costs nothing.
Header and manifold, before backfill1.5× design, 1 hEvery fusion on the trench run, at the stage where a trench is still open.
Whole system, after connection to the plant1.25× design, 1 hThe transitions and the plant-room joints, which are the mechanical ones and therefore the ones that can be re-made.
Commissioning, with the loop at working temperaturedesign pressureA joint that passed cold and weeps when the loop expands. Rare, but the only way to find it is to run the system.

Two practical notes. The test pressure has to be read on a gauge that stays visible for the whole grouting operation, which means a gauge at the head and not in the trench. And the test record should carry the joint numbers, because a loop with 40 fusion joints per borehole is a loop where "it passed" is not a traceable statement unless the joints were logged as they were made.

Temperature: the good news, and the one place it matters

On the heating side a ground loop is colder than a water main, not hotter, and PE gets stronger as it cools: there is no derating below 20°C, and PE100 stays ductile well below freezing, which is why a loop can be installed in winter. The place temperature does matter is the cooling side, where the loop can run at 35 to 40°C for long periods in a cooling-dominated building, and the pressure rating follows the same curve as any other PE pipe.

Indicative PE100 pressure rating by temperature, 50-year basis
Loop temperatureDerating factorSDR 11 (PN 16 at 20°C)SDR 17 (PN 10 at 20°C)
0°C, heating mode1.0016.0 bar10.0 bar
20°C1.0016.0 bar10.0 bar
30°C0.8713.9 bar8.7 bar
40°C, cooling mode0.7411.8 bar7.4 bar

In a closed loop the operating pressure is set by the static head plus the pump, so a 150 m borehole is about 15 bar of static head before anything else is added. That single number, not the temperature, is what normally decides the wall series, and it is why ground loops are usually SDR 11 rather than the SDR 17 used for the trench header. Where glycol is used for freeze protection, the mixture does not change the chemical resistance of PE100 at these concentrations, but it does change the fluid properties the pump has to be sized for.

Installation damage, which is the real failure mode

Most loop failures in the first year are not joint failures at all. They are mechanical damage from the installation, and they show up as a pressure loss that cannot be explained.

Where the damage happens, and the measure that prevents it
CauseWhat it does to the loopMeasure
Loop resting on the borehole baseThe U-bend takes the weight of the whole loop and cannot expandA centraliser or a stand-off at the bottom, and a loop length measured rather than guessed.
Grout falling directly onto the U-bendLocalised heating and impact at the one fitting that cannot be replacedTremie pipe to the bottom, grout placed from below upward.
Pipe dragged over the casing shoeA gouge that becomes a leak once the loop is under pressure and warmA guide shoe, and a second person watching the entry point. This is the cheapest control on the list.
Sharp backfill against the trench headerPoint loading, then a slow crack under settlementSand surround for the first 100 mm, which is a specification item rather than a site preference.
Loop left open between installation and connectionGroundwater, soil and animals in the loop, then a blocked circuitCaps on every open end, then flush before connection. This is what the end caps in the range are for.

The list is unglamorous, and that is the point. A ground loop package that is bought on fitting price and installed without these five controls will leak regardless of which mould it came out of.

The families, with the size each one actually reaches

Ground loops use a narrow slice of the range: dn32 and dn40 in the borehole, dn63 to dn160 in the trench, and rarely above dn200 at the plant. The table is here so the slice can be seen against what is available, and so a district-scale header is not quoted from a guess.

Butt fusion families, size range and largest moulded entry
FamilySize rangeSizesLargest entryWhere it sits in a ground loop system
Butt Fusion Reducerdn50–dn63078dn630Reducing the header where the borehole group is smaller than the main
Butt Fusion 90° Elbowdn20–dn160031dn1600Direction changes in the trench
Butt Fusion Equal Teedn40–dn160027dn1600Branch to a chamber or a second header
Butt Fusion Reducing Teedn50–dn1600102dn1600Circuit into a smaller header without a second reducer
Butt Fusion End Capdn20–dn160031dn1600Flush and test point, and future borehole group
Butt Fusion Flange Adapterdn40–dn160028dn1600The flanged end at the plant room wall

What a ground loop package has to state before it can be quoted

Ordering checklist
What to stateWhy it moves the price
Number of boreholes, and loops per boreholeSets the coupling count, which is most of the electrofusion in the package.
Loop diameter: dn32 or dn40, and single or double UDecides the fitting size and whether the head needs a tee or a pair of couplings.
Depth, and the resulting static headA 150 m borehole is about 15 bar before the pump, which is what fixes SDR 11 rather than SDR 17.
Header route and the number of manifold positionsSets the tees, the valves and the chamber count.
Circulating fluid: water or glycol, and the concentrationDoes not change the fitting, but it does change the pump and the flush volume, and it is worth stating on the same drawing.
Test pressure and who witnesses the testDecides whether test caps and gauges are in our scope or the contractor's.
Plant connection: flange standard and drillingThe heat pump is already built. A transition drilled to the wrong pattern is scrap.
Delivery sequence by borehole groupA loop field is drilled in stages; shipping the whole package at once usually means storing half of it on site.

Frequently asked questions

Which HDPE fittings are used in a borehole heat exchanger?

Almost entirely electrofusion: couplings to join the U-tube to the riser at the head, reducers where the loop meets the header, and end caps for the flush and test points. The reason is access, not preference. The joint is made above ground before the loop goes down, and it is the one joint that cannot be reached again once the borehole is grouted.

What SDR should a ground loop use?

SDR 11 for the borehole loop, because the static head of a 100 to 150 m borehole is 10 to 15 bar before the pump is added. SDR 17 is common for the trench header, where the head is a few metres. The temperature is not usually what decides it: a loop runs colder than a water main in heating mode, and only the cooling mode at 35 to 40 degrees brings any derating into play.

How much does a ground loop move thermally?

About 0.2 mm per metre per kelvin, so roughly 400 mm over a 100 m borehole for a 20 K swing between heating and cooling. The movement is absorbed at the free-hanging U-bend and in a slack coil at the head; what causes failures is a detail that anchors the loop, such as a transition cast into concrete or a loop tied to a rebar cage.

When should a ground loop be pressure tested?

Three times: after assembly before it goes into the hole, in the borehole before and during grouting, and again as a whole system after connection to the plant. The first test is the important one, because it is the last point at which a bad joint can be cut out rather than written off with the borehole.

Do you supply the U-bend at the bottom of the loop?

No. The U-bend is either factory-formed from the same pipe as the loop or made on site by bending, and it is supplied with the loop rather than as a moulded fitting. What we supply is the electrofusion coupling that joins the loop to the riser at the head, plus the reducers and caps for the trench and plant connections.

Is PE suitable for glycol or brine loops?

Yes. PE100 is resistant to the propylene and ethylene glycol concentrations used for freeze protection, and to the inhibited brines used in closed loops. What changes with glycol is the fluid viscosity and therefore the pump duty, not the fitting; the one thing to check is that the inhibitor package is compatible if a proprietary brine is used.

Can the same fittings be used for a district-scale loop field?

Yes, and the sizes are there: the butt fusion range goes to dn1600 for a large header, and the electrofusion range to dn1400. What changes at district scale is the manifold design and the number of valves, since every borehole group has to be isolatable and flushable on its own.