The Designer’s problems

1

Three tests, and how much ground each one sees

“Percolation test” is used loosely for several procedures that look alike on site: dig a hole, fill it with water, time how long it takes to drain. They differ in how big the hole is, how deep it goes and how the result is used, and those differences decide how far the answer can be trusted.

The Building Regulations test. Approved Document H prescribes one simple percolation test, and it is precise about where the hole goes: 300 mm square, with its base 300 mm below the invert of the incoming pipe. The hole is soaked overnight, refilled and timed as the water falls from 75% to 25% full, giving a percolation value, Vp, in seconds per millimetre. Under H2 that decides whether the ground can take treated effluent through a drainage field, acceptable where Vp falls between 15 and 100 s/mm. Under H3 the same test is used for rainwater soakaways: Vp is converted to an infiltration rate to size soakaways serving small areas, with larger areas referred to BRE 365 or BS EN 752.

Followed to the letter, the test samples the ground just below the pipe, which is right for a shallow drainage field but not for a soakaway whose base may be a metre or more deeper. Some builders, sensibly, dig a trial pit to the depth of the proposed soakaway and form the 300 mm test hole in its base. That gives a reading much closer to the ground the soakaway will actually sit in, and it is good practice. Even so, the test still wets only a 300 mm cube of ground.

BRE 365. The standard method for surface water soakaways. A trial pit is dug to the depth of the proposed soakaway, filled quickly, timed from 75% to 25% full, and filled twice more, ideally on the same or consecutive days. The slowest of the three runs is used, not the average, and it is expressed as an infiltration rate in metres per second, which feeds straight into the storage calculation.

The SuDS approach. The CIRIA SuDS Manual (C753), enforced by Lead Local Flood Authorities, builds on BRE 365. It asks for the test at the location, depth and head of water of the real device, applies a factor of safety to the measured rate (higher where failure would flood buildings), and sets limits such as a minimum workable infiltration rate and a minimum depth above groundwater.

2

How the wrong test produces a soakaway that cannot cope

Ground is rarely the same all the way down, or all the way across. A sandy or silty layer just under the topsoil can drain freely while a clay or silt band a metre lower, or a lens of finer silt a couple of metres away, barely drains at all. A soakaway only works as well as the ground around its sides and base, so the test has to sample that ground.

The Building Regulations test is not wrong, and it is not meant to be the last word. It is a prescribed minimum, and its result is only as representative as the small cube of ground it is dug in. Taken as written, with the hole just below the pipe, it tests only the upper layer. If that layer is the free-draining one, the test passes comfortably while the base of a deep crate soakaway sits in a slower band the hole never reached.

Moving the hole to the base of a trial pit at soakaway depth helps, but in variable ground such as the Fens it can still miss what matters. Silts and clays change over a few metres sideways as well as downwards: a sandy pocket at the bottom of one pit, a finer-grained silt or clay lens beside the soakaway two metres away. A 300 mm hole does not see reduced permeability in the ground next to it, however carefully it is placed. A BRE 365 pit, dug to the full depth and filled over its whole face, does, because the water has to leave through every layer the soakaway will.

How the test is run matters too. Averaging results, or timing a first fill into dry ground, gives a rate the ground cannot keep up once it is wet. BRE 365 deliberately repeats the fill and takes the slowest run for exactly this reason: the third fill into saturated ground is usually the honest one.

Because the storage volume of a soakaway depends directly on the rate it can empty, an optimistic rate gives an undersized soakaway. If the rate used is several times faster than the real one, the soakaway fills during the storm, surcharges back up the pipes and spills across the garden or against the building. Worse, it may not drain down before the next rain, so it starts each storm part full.

3

Why the test should be done in late winter

Ground behaves very differently through the year. Over autumn and winter, rain soaks down and refills the groundwater, a process called recharge. By late winter, typically February and March, the soil is at its wettest and the groundwater table is at or near its highest.

That is the condition a soakaway has to work in, because long, wet winter spells are when it is most likely to be called on repeatedly with no time to dry out. A test dug in August into dry, cracked ground can drain spectacularly fast and look like an excellent result. The same pit in March may drain at a fraction of the rate, or fill from below with groundwater before any water is added.

A late-winter test gives the worst realistic case, and that is the one to design for. It also lets whoever digs the pit see and record the groundwater level directly, which a summer test cannot do. That record is one of the first things we ask for when designing from the results.

4

The metre that matters: depth to groundwater

Water can only soak away into ground that has room for it. Below the groundwater table every pore is already full, so a soakaway sitting in groundwater has nowhere to send its water. It simply fills and stays full.

For that reason SuDS guidance, and many Lead Local Flood Authorities, require at least 1 metre of unsaturated ground between the base of the soakaway and the highest seasonal groundwater level. That metre does three jobs. It gives the water somewhere to go. It stops groundwater rising into the soakaway and taking up its storage. And it filters the runoff through soil before it reaches the groundwater, which protects water quality.

This is why the depth measured in late winter is so important. A summer reading can suggest two metres of clearance when the winter peak leaves almost none. In low-lying parts of the region, including the Fens, the winter water table can sit within a metre or so of the surface, and a conventional deep soakaway is often simply not possible. A shallow, wide design or a different drainage route is then needed.

5

From 20% to 30% to 40%: why rainfall is uplifted

A warmer atmosphere holds more moisture, so the heaviest bursts of rain are becoming more intense. Drainage is designed from historical storm records, so an allowance is added on top to cover the change over the life of the development. That allowance has been revised upwards twice as the climate science has improved.

2006 to 2016: 20% or 30%. National planning guidance (PPS25, carried into the 2012 technical guidance to the NPPF) set the uplift by how long the development would stand. Peak rainfall was increased by 20% for the period to 2085 and by 30% for 2085 to 2115. Housing, expected to last a century, was normally designed to the 1 in 100 year storm plus 30%. A building with a shorter life could justify 20%.

2016: a choice between 20% and 40%. In February 2016 the Environment Agency replaced those figures with a range instead of a single number: a central allowance of 20% and an upper end allowance of 40% for the 2070 to 2115 period. Flood risk assessments had to test both, but the guidance left the design figure to be agreed with the local authority. Many Lead Local Flood Authorities required 40% for drainage. Others accepted 20% with justification, for example a shorter design life, a less vulnerable use, low consequences if the system was exceeded, or the cost of the extra storage on a marginal scheme.

2022 onwards: 40% by catchment. In May 2022 the allowances were recalculated for each river management catchment using the Met Office’s high-resolution projections. Development with a lifetime beyond 2100, which includes all housing, must now be checked against the upper end allowance for the 2070s for the 1% annual exceedance probability storm. For the Cam and Ely Ouse catchment, which covers much of our area, that is 40%. Lower central figures still apply to development with a shorter design life, which is why commercial and temporary buildings can legitimately use a smaller uplift.

There is one more reason the lower figures survived for so long. Lead Local Flood Authorities are statutory consultees only on major development. A single house or an extension is approved by Building Control against Approved Document H, which has no climate change allowance of its own, so for years many small soakaways were sized to past rainfall with no uplift at all.

The effect on storage is direct. A soakaway stores the difference between what flows in and what soaks away. On slow-draining ground almost all of the storm has to be stored, so 40% more rain needs roughly 40% more storage. A soakaway built to past rainfall holds only around 70% of today’s design storm, and one built to the 20% figure around 85%. In practice the storm that used to be the one-in-a-hundred-year event now overtops them more often. The water backs up through gullies, ponds on drives and patios, or runs onto a neighbour’s land, and owners are left wondering why a soakaway that worked for years has started to fail.

6

Is climate change raising groundwater under our soakaways?

The evidence points that way, with honest caveats. UK climate projections show milder, wetter winters and hotter, drier summers. More of the year’s rain is expected to fall in winter, when it recharges groundwater, and recharge may be squeezed into a shorter, more intense season. Both favour faster rises and higher winter peaks.

A review of UK groundwater research by Jackson, Bloomfield and Mackay (2015) found that, under a high emissions scenario, the median of the modelled projections showed higher average winter groundwater levels by the 2050s at most sites, alongside lower summer levels. The same review found no clear trend in the historical records yet, and stressed that local geology controls how each site responds. Near the coast and estuaries, projected sea level rise of 0.4 to 1 metre by 2100 adds a further push upwards on groundwater.

For a soakaway, the consequence is direct. The 1 metre of clearance measured when it was designed is not guaranteed for its lifetime. If the winter peak climbs, the unsaturated zone thins, the ground around the soakaway is wetter when the heavy rain arrives, and the infiltration rate falls just when it is needed most. A soakaway that discharged at a perfectly good rate twenty years ago can start to hold water through the winter without anything about the soakaway itself having changed.

That is why we advise testing in late winter, keep a generous margin above the groundwater, and design storage with the climate uplift in it. Where the ground or the water table will not support a soakaway, it is better to say so at the start and design an alternative, such as attenuation with a controlled discharge, than to build something that is full before the rain stops.

7

The caveats

Everything above concerns rainwater. Homes with no mains sewer also have foul water to deal with, and here what is accepted on site does not always match the rules.

A treatment plant to a soakaway. A package treatment plant serving a typical four-bedroom house discharges around 0.75 m³ of treated effluent a day, every day of the year. The Building Regulations and the Environment Agency expect that effluent to go to ground through a drainage field designed to BS 6297 and sized from the H2 percolation test. For a four-bedroom house that field typically takes up an area of around 6 m × 6 m. On a tight plot there may simply be no room for it, and some local authority building control departments will then accept the plant discharging into a soakaway instead.

It looks like a sensible compromise, but the two are not the same. Effluent from a treatment plant is cleaner, not clean: it still carries bacteria, nutrients and some organic matter. A drainage field is laid shallow, within the topsoil and upper subsoil where oxygen and soil bacteria are plentiful, and as the effluent trickles through that living layer it receives its final, tertiary stage of treatment. A soakaway is set deeper, below the organic topsoil, and releases the effluent in one place into ground with far less biological activity and much closer to the groundwater. The tertiary treatment never happens.

There is a regulatory catch too. The Environment Agency’s general binding rules for small sewage discharges to ground require a drainage field, and they exclude soakaways designed for rainwater that were installed after December 2007, as well as boreholes and similar deep structures. A building control approval does not override those rules. A system that breaks them needs an environmental permit, and the Agency will not grant one where effluent can reach groundwater directly. That is worth knowing before buying a house, as well as before building one.

When the land will not take it. Where the ground fails the percolation test, or the winter water table is too high, the water has to go somewhere other than into the ground. For treated effluent the next option is usually a watercourse that flows all year, often after further polishing through a reed bed, and it needs the Environment Agency’s rules or a permit to be met. For surface water, attenuation ponds and basins hold the storm and release it slowly to a ditch or stream at an agreed rate. Both need land, consent from the Environment Agency or the internal drainage board, and a watercourse to discharge to.

The hierarchy. A connection to the public sewer is the last resort for surface water. Defra’s National Standards for Sustainable Drainage (2025) and Anglian Water’s own surface water policy set the same order, shown alongside. Anglian Water asks for sufficient evidence that every disposal route above a sewer has been explored, which in practice means BRE 365 infiltration results and an investigation of nearby watercourses. It accepts surface water into a combined sewer only in exceptional circumstances, says it will seek to separate surface water from any new development, and limits new connections to the equivalent greenfield runoff rate, with storage on site for the 1 in 100 year storm plus climate change.

Foul water runs the other way. Anglian Water and the Environment Agency’s joint advice (2020) starts from a presumption that new development connects to the public foul sewer. A package treatment plant is considered only where a connection is shown not to be feasible on cost or practicality, and a septic tank only after that, and only on suitably permeable ground. The planning application has to justify the choice, normally on a foul drainage assessment form (FDA1). The general binding rules add that a connection is expected where a public foul sewer is within 30 m of the property boundary.

Why the order matters. In many older towns and villages foul water and rainwater share a single combined sewer. In heavy rain the rainwater fills those pipes, and storm overflows release the mix of rainwater and sewage into rivers to stop it backing up into homes and streets. Every roof and drive connected to a combined sewer adds to that storm flow. That is how untreated sewage reaches our rivers in wet weather, and it is the real reason the hierarchy exists. A soakaway that works, tested properly and sized for the climate it will face, is worth far more than a pipe to the nearest manhole.

1

Why every Fen site needs its own survey

Geological maps of the Fens show broad bands of silt and peat, but they are drawn at a scale where a whole village sits inside one colour. On the ground the layers change over a few metres. On one recent greenfield plot, two trial pits dug on the same building footprint found a pocket of peat just 300 mm below the surface in the first, and a continuous peat bed at 1.8 m in the second, with groundwater pouring out of it.

The old courses of rivers and creeks add to the patchwork. Silt-filled channels, called roddons, wind across the landscape as slightly raised ridges of firmer ground, while the land between them can be soft clay and peat. A house can sit across both.

So the starting point for any new house or extension in the Fens is a geotechnical survey on the site itself: trial pits or boreholes deep enough to find the peat and prove what lies beneath it, a record of the groundwater, and a look at the existing building’s foundations where there is one. Guessing from the house next door, or from a map, is how foundations end up in the wrong layer.

2

Young ground that has never been squeezed

Almost everything above the older glacial clays and bedrock in the Fens was laid down in the Holocene, the last few thousand years. As sea level rose after the last Ice Age, the basin flooded and drained again several times. Each time the sea pushed in it left silts and clays; each time it retreated, reeds, sedges and fen woodland grew and died where they stood, building up peat. The oldest of these deposits are only around 5,000 to 6,000 years old. Geologically, that is yesterday.

The British Geological Survey names the layers. From the bottom, the Fen Lower Peat, which in places lies 3 m or more below ground. Over it the Barroway Drove Beds: soft grey clays and silty clays with reed remains, laid down in salt marsh and brackish lagoons, up to about 12 m thick. Then the Nordelph Peat, usually less than a metre thick but locally more than 3 m, with oak trunks, roots and reed stems still in it. On top, the Terrington Beds: reddish-brown salt-marsh clays and pale silts, generally up to about 3 m thick, which form much of the surface around the Wash.

None of these layers has ever carried an ice sheet or a great thickness of overlying rock. They are normally consolidated: the only load they have ever felt is their own weight and the water in them. Put a new building on them and they compress for the first time. The upper silts are often firmer, having dried and wetted for centuries, but the grey silts and clays below are soft enough to dent with a thumb, and the peat has almost no strength at all.

3

Why peat is the problem

Peat is mostly water held in a sponge of partly rotted plants. Load it and the water is squeezed out slowly, so the ground goes on settling for years. That first stage is called primary consolidation. After it, the plant fibres themselves keep creeping and compressing, which is called secondary compression, and it can continue for decades under the weight of a house.

Peat also shrinks when it dries and swells when it rewets, and once it is above the water table and exposed to air it slowly oxidises and disappears. Because its thickness varies so much across a plot, one corner of a building can settle far more than another. That differential settlement, not the total amount, is what cracks walls.

This is why peat cannot simply be built over. Where it is shallow it has to be dug out and replaced. Where it is deeper, foundations have to pass through it to firmer ground below, or the building has to be carried on a structure designed to spread or bridge the load.

4

A landscape that has sunk by metres

The Fens were a wetland until they were drained, first by the great cuts of the 17th century, then by windmills, steam pumps and, today, the electric pumps of the internal drainage boards. Drainage made the land farmable, but it also lowered the water table below the peat. Dried peat shrinks, and exposed to air it oxidises, a process called peat wastage.

The best record is at Holme Fen, near Peterborough. When Whittlesey Mere was drained in 1851 a post was driven into the peat with its top at ground level. The ground has since fallen more than 4 metres down the post, and Holme Fen is now the lowest land in Great Britain, at about 2.75 m below sea level. Across the peat fens the surface has dropped by well over 2 m in many places, which is why the rivers now run on embankments high above the fields around them.

Buildings have paid the price. Old houses built on the peat along the drove roads commonly show subsidence damage, and many have been lost altogether, abandoned as they tilted or taken down. Houses on the firmer silts of the old river courses have generally fared better. Roads across the peat ripple and crack for the same reason.

5

Too shallow shrinks, too deep sinks

The upper silts and clays of the Fens also shrink and swell with the seasons. Their volume change potential is often medium, and trees and ditches make it worse. To get below the zone that moves, NHBC Chapter 4.2 and building control set minimum foundation depths, typically 0.75 m, rising to 1.0 m or more in shrinkable ground and much deeper near trees.

In the Fens that creates a squeeze. Taking a trenchfill foundation deeper to escape seasonal shrinkage can take it down into, or just above, the peat and the soft grey silts, where the new load causes consolidation settlement. Shallow footings avoid the soft layers but sit in the ground that shrinks. The right depth has to be found between the two, and on some sites there is no right depth at all. That is when a raft or piled foundation becomes necessary.

6

Extensions: two buildings that settle differently

An extension is where Fen ground causes the most trouble. Many older Fen houses stand on shallow brick footings, sometimes only a few courses deep. They settled long ago, and the upper silts beneath them have been compacted by a century of loading and seasonal drying. They are stable because nothing has changed.

The usual principle, followed in British Standard foundation practice and expected by building control, is that an extension should be founded in the same way and on the same stratum as the building it joins, so that the two move together. In the Fens that principle runs straight into the modern depth rules. A new foundation cannot be as shallow as the old one, because it must get below the zone of seasonal shrinkage. But taken deeper, it may bear onto the peat or the soft grey silts that the old house never touched. The new part then settles as that ground consolidates, while the old part does not, and the junction cracks.

There are ways through it, and the survey decides which. Where the ground allows, the new foundation can be taken to the agreed depth and stepped under the existing footing where they meet, so the two bear on the same layer. Where it does not, a movement joint can let the two parts move independently, or the extension can be carried on a raft or piles designed to limit its settlement. What should never happen is a deep, stiff new foundation simply butted against a shallow old one on soft ground.

Precedent counts too. The Fenland guide allows building control to take into account evidence that nearby buildings on the same type of foundation, on similar ground, have stood the test of time, for at least five years, with records of levels and inspections for cracking. An old house that has stayed level and uncracked on shallow footings is itself useful evidence for how its extension should be founded.

7

The Fenland guide

Fenland District Council’s Building Control Guide to Foundations in Fenland was written with its structural consultants after disputes over foundation design in the late 1980s, and is now in its third edition (2003). It covers single and two-storey housing, and the NHBC supports its general principles. It sets out a small number of “deemed to satisfy” solutions, each tied to what the trial holes find.

Investigation first. Unless a specialist site investigation has been done, at least two offset trial holes or pits, about 3 m deep, front and rear of the building, with the strata logged. The deemed-to-satisfy routes only apply where the ground is consistent in both.

Minimum depths. Strip foundations at least 0.75 m deep, or 1.0 m where shrinkable subsoil such as clay is found. The toe of a raft at least 0.45 m.

Four ways to deal with peat. Strip foundations can be taken down through the peat onto firm strata below, usually with a suspended ground floor. Designed reinforced strips can sit above the peat where there is at least 1.5 m of firm strata beneath them and the peat below is no more than 150 mm thick, normally for single-storey buildings. A rigid raft can be used where the peat is roughly 150 to 450 mm thick and lies below at least 1.5 m of firm strata. Where peat lies within 1 m of the underside of the foundation, it and the soil above it would normally be dug out and replaced with compacted fill.

Beyond the guide. Where none of these fits, the guide expects a full geotechnical site investigation and a specialist design, such as piled foundations or ground improvement, with calculations for ground beams, floor slabs and piles, and testing of the piles. On deep peat and soft organic silts that is where our own reports usually end up: piles through the compressible layers to firmer ground, ground beams and a suspended floor, with the piles designed for negative skin friction.

8

Hotter, drier summers since 2022

The weather of the last few years has made Fen ground move more. In 2022 England had its joint hottest summer on record and one of its driest, with a record 40.3°C in July and drought declared in August. Insurers expected to pay £219 million in subsidence claims that year, the highest for 16 years, with most claims arriving after the heatwave.

2025 went further. The Met Office recorded the UK’s warmest and sunniest year, its warmest summer, and the driest spring since 1974. East Anglia was the warmest and sunniest district in summer 2025 and received only about 62% of its normal summer rainfall. The Met Office estimates that a summer like 2025, once expected about every 340 years, now comes round roughly every five.

Longer and drier summers dry the ground to greater depths. The upper silts and clays shrink further than before, peat above the water table dries and oxidises faster, and ditch levels fall. Foundations that stayed below the moving zone for decades can find it has reached them. Since 2022 we have seen more cracking in Fen houses that had been stable for years, and it is one more reason to survey properly and design new foundations for the summers ahead, not the summers past.