The trial pit was full again before the tape came out
Thirty metres by twenty, formation at 4.0 m, and the logs say fine to medium sand from about 1.5 m down to stiff clay at 11.5 m. Water was struck at 1.5 m in every pit, and by the time anyone got a tape into the first one it had filled back to within a boot's depth of the surface. The machine will reach 4.0 m in a morning. Holding it there for six weeks, with a base a gang can stand on and a subgrade that will take a blinding, is the actual job — and it is a different job from digging the hole.
There are two ways through it and they are not two techniques so much as two places to put the risk. Take the water away, and the pore pressure change you create does not stop at the hoarding; it runs out under whatever is next door, and the ground out there responds to it. Keep the water out instead — sheet piles, a bored pile wall, a slurry trench, a diaphragm wall panel held open on bentonite head — and the change stays inside your line, but the wall then has full hydrostatic pressure on the back of it for as long as the hole is open, and something has to prop that.
What settles the choice is four things, and none of them is on the trial pit log: the permeability of the layer you are digging in, what sits directly under formation, what stands inside the drawdown you are about to create, and how long the excavation stays open. The first is the one everybody guesses at and nobody should. Permeability spans nine orders of magnitude across ordinary soils, and a falling-head test on a disturbed sample measures the sample rather than the deposit.
Drainage and pressure are not the same problem
First establish which of the two you have, because they take different kit and the wrong diagnosis is expensive in both directions. If formation sits inside the water-bearing layer itself — the case above, sand from 1.5 m and a dig to 4.0 m — the task is drainage. You are lowering a phreatic surface, emptying pores in the walls and base until the water level sits below the dig, and the volume you shift is genuinely large because you are draining a body of soil.
If instead the dig is in a clay or silt blanket with a sand or gravel aquifer confined beneath it, that blanket will not drain in any programme you would recognise, and it does not need to. The problem there is pressure acting upward on the underside of the plug you are leaving. Every metre you excavate thins that plug while the pressure below stays exactly where it was, so a base that checks out comfortably on day one can be marginal by the time the last 600 mm comes out. Eurocode 7, BS EN 1997-1, treats this as the UPL ultimate limit state and treats seepage failure, internal erosion and piping as HYD; take the required margin from the specification and the geotechnical designer on your job, not from a number somebody remembers from a previous site.
The method follows the diagnosis. Depressurising a confined aquifer moves a fraction of the water that draining an unconfined sand does, because you are bleeding off head rather than emptying pore space — a handful of relief wells or deep wells screened into the aquifer, sometimes nothing more than pressure relief through the base itself where the design allows it. Reading that situation as a drainage problem puts a full wellpoint ring on a site that never needed one. Reading a drainage problem as a pressure problem gets you a base that boils.
Then look at where the permeability lands, because it decides not just how much water there is but whether pumping is even a sensible way to spend six weeks. The bands below are the working ones in specialist practice, and the ranges overlap on purpose: real deposits are layered, and a 200 mm silt band inside a sand can halve the vertical permeability while leaving the horizontal figure almost untouched.
| k (m/s) | Typical ground | What actually works |
|---|---|---|
| Above 1e-3 | Clean gravel, gravelly sand, cobbles | Very high flows and a wide cone. Deep wells at high duty, or a cut-off — pumping the whole aquifer often costs more in consents and neighbours than the wall does |
| 1e-3 to 1e-5 | Clean to slightly silty sand, sand and gravel | The classic wellpoint and deep-well range. Drawdown arrives in days, flows are manageable, and a staged ring is routine |
| 1e-5 to 1e-7 | Fine and silty sand, laminated silty deposits | Wellpoints work but slowly and at low yield; vacuum wellpoints or ejector wells are the usual answer, and the drawdown period has to be in the programme as work, not as a start-up day |
| Below 1e-7 | Silt, clay, glacial till with a clay matrix | The layer does not drain in construction time. Exclude the water, and check whether the real issue is an aquifer confined beneath it rather than the layer itself |
The cone reaches a long way past the hoarding
Before anything gets sized, find out how far the effect travels, because that number sets the size of the survey you should have done before the first pump ran. The Sichardt expression is the standard first pass in unconfined ground: radius of influence is roughly three thousand times the drawdown times the square root of the permeability, all in metres and metres per second. For the site above — formation at 4.0 m, target lowered level 4.5 m, water at 1.5 m, so 3.0 m of drawdown in a sand at 1e-4 m/s — that gives 90 m.
The square root is what makes this worth checking rather than assuming. Move the permeability one order of magnitude, to a clean sand at 1e-3, and the same 3 m of drawdown reaches about 285 m. Move it the other way to 1e-6 and the cone shrinks to 9 m — except that a wellpoint ring in ground at 1e-6 will take weeks to achieve the drawdown at all, so the ground types with a comfortable cone are precisely the ones where pumping is slowest. Drawdown itself enters linearly, so a second stage that doubles the lowering doubles the reach.
Treat the answer as a search radius, not a boundary. Drawdown decays continuously and is down to millimetres well before the nominal edge, and it is never a circle in real ground — a gravel lens, a fissure set, a river or a leaking main will pull it much further one way than another. What the radius is genuinely good for is telling you what to go and look at: shallow-founded neighbours, timber piles that only survive because they have never been above water, private and licensed abstractions, trees, contaminated land, and waterlogged archaeological deposits that are preserved by exactly the conditions you are about to remove.
Run this before the survey, not after, because the answer is the radius that survey has to cover. The drawdown box follows the metric and imperial switch and is in metres or feet; the permeability box does not and is read as metres per second on either page, so convert first — 1 cm/s is 0.01 m/s and 1 m/day is 1.16e-5 m/s.
How far you need to lower the water table at the well.
The soil's hydraulic conductivity, from a pump test or published typical values for the soil type.
Estimated radius of influence
300 ft
The Sichardt formula is a widely-used but approximate empirical estimate — it's most useful for early planning and tends to underestimate the true radius of influence compared to more rigorous pump-test-based methods. Verify with an observation well or pump test once dewatering begins, especially near property lines or sensitive structures.
They open the calculator with your figures already in it
Wellpoint Dewatering Radius of Influence Calculator (Sichardt): 300 ft — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
What this calculation does not cover
- No input describes the excavation or the wellpoint ring, so the answer is a distance out from the dewatering source, not a radius measured from a defined centre. Layering, horizontal-versus-vertical permeability, and wellpoints that penetrate only part of the aquifer all sit outside the two numbers this uses.
- Time is not in the formula. There is no pumping duration and no aquifer storage behind the figure, so it does not tell you how far the cone has spread after a week, a month, or the length of the job — only a single planning distance.
- Boundaries and recharge are excluded. A river, canal, leaking main or other recharge source near the excavation holds the water table up and keeps the real effect well short of this figure, while a cut-off wall, sheet-pile box or clay layer redirects it. Nothing about the surroundings reaches the arithmetic.
- The formula is written for unconfined, water-table conditions. Confined or artesian ground, perched water sitting on a clay layer, and pressure relief of a deeper aquifer behave differently, and this number does not describe them.
- This is not a dewatering design and not a damage assessment. It returns no well count, spacing, flow rate or pump duty, and it says nothing about settlement of compressible ground, timber piles kept sound by the water table, or services and shallow foundations inside the radius — that needs a geotechnical engineer, baseline levels and monitoring. Abstraction and discharge consents are a separate matter it does not touch.
What one well will actually give up
Yield in an unconfined aquifer comes from the Dupuit-Forchheimer expression for steady radial flow to a fully penetrating well: flow is pi times the permeability, times the difference of the squares of the saturated thickness at the outer radius and at the well, divided by the natural log of the ratio of those two radii. The squares are the part worth internalising. Flow depends on the square of saturated thickness, so a thin aquifer sitting on a hard bed gives up very little even in permeable material, which is why the depth to the impermeable layer is a bigger number in this arithmetic than most people expect.
Carrying the site through: the sand runs from 1.5 m to 11.5 m, so the undisturbed saturated thickness is 10.0 m and the 3 m of lowering leaves 7.0 m at the well. Take the outer radius as the 90 m the previous section produced and a well radius of 0.15 m, at 1e-4 m/s, and a single well delivers about 2.5 litres per second — roughly 216 cubic metres a day, or about 40 US gallons per minute. That is a useful order of magnitude and nothing more.
What it is not is a number to multiply by the count of wellpoints in the ring. Wells inside a common drawdown cone interfere with each other, and the twentieth point in a ring produces a small fraction of what the first one did. Total system flow is normally estimated by treating the whole ring as one large equivalent well — the equivalent radius being the one that gives the same plan area as the excavation, about 13.8 m for a 30 by 20 m dig — and this page cannot do that, because its well radius field stops at 1 m. Use the page for the per-point duty and the sanity check on pump selection, and take the system total from the equivalent-well calculation or, better, from a trial pumping test on site.
Two things about the form. The well radius box is displayed in centimetres on the metric page and inches on the imperial one, so the 0.15 m above is entered as 15; and the permeability box, as on the radius page, stays in metres per second whichever system is selected. A value typed in centimetres per second is the error that slips through quietly — a hundred times too large, and still inside the accepted range.
Enter the saturated thickness before drawdown and what is left at the well, then the radius of influence from the section above as the outer radius. The headline is one well's duty in litres per second on the metric page; on the imperial one it is restated as CFM, which is a ventilation unit rather than a pumping one, so take the duty from the gallons-per-minute line in the breakdown instead — that line and the cubic-metres-a-day line above it read the same on both pages.
The aquifer's hydraulic conductivity.
The undisturbed saturated aquifer thickness far from the well.
The remaining saturated thickness right at the pumping well, after drawdown.
The radius at which h1 is measured (often the radius of influence).
The radius of the pumping well itself.
Estimated well yield
44.7 gal/min
This is a steady-state estimate assuming a fully penetrating well in a homogeneous, isotropic unconfined aquifer — real dewatering systems are affected by partial penetration, aquifer boundaries, and non-steady flow, and should be verified with an actual pump test.
- Equivalent in m³/day
- 243.51 m³/day
- Equivalent in US gallons per minute
- 44.67 gpm
They open the calculator with your figures already in it
Dewatering Well Yield Calculator (Dupuit-Forchheimer): 44.67 gal/min — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 44.7 gal/min — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- The equation covers one fully penetrating well pumping on its own. It does not add the drawdown cones of neighbouring wells or wellpoints, so it cannot tell you how many wells an excavation needs or how far apart to set them — in a ring or a line each well delivers less than this figure.
- Steady state only, in a homogeneous, isotropic, unconfined aquifer. There is no time in the arithmetic, so the higher rate while the cone is still developing and the total volume pumped over the job both sit outside it, as do partial penetration, layered or anisotropic ground, confined and leaky aquifers, and any recharge or no-flow boundary such as a river, a cut-off wall or bedrock.
- This is a planning-level flow estimate, not a dewatering design and not a pump specification. Screen entrance velocity, filter pack capacity, drawdown available to the pump, suction lift limits, friction losses in the header and discharge run, and standby capacity are all separate calculations, and the yield itself should be confirmed by a pump test before plant is ordered.
- Nothing here covers the consequences of lowering the water table: settlement of shallow-founded structures and services inside the drawdown cone, migration of fines and piping into the well, or the discharge consent and abstraction licensing the pumped water needs. Where drawdown reaches a neighbour's property, an abstraction, or compressible ground, that is a geotechnical engineer's assessment, not a page of arithmetic.
- Yield is directly proportional to the hydraulic conductivity you type, and K in natural ground varies by orders of magnitude over short distances and directions. The result carries no uncertainty band around that, and the K box is read as metres per second whichever measurement system the page is set to.
Every metre of drawdown loads somebody else's clay
This is the part that decides the argument on a built-up site, and it follows from one relationship. Effective stress is total stress minus pore water pressure, so removing pore pressure without removing any soil increases the effective stress carried by the grains. Drop the water table one metre and you have added the unit weight of water — 9.81 kilonewtons per cubic metre, so 9.81 kPa — to the effective stress on everything below the new level, right out to the edge of the cone. Nothing was built, nothing was loaded, and the ground under the terrace across the road is now carrying more than it was.
In sand that increase produces a small immediate settlement and stops. In a compressible clay it produces consolidation, which is slow, which continues after your pumps come off, and which can be large. Take the 3 m of drawdown above as an added stress of 29.4 kPa on a five-metre clay layer with a compression index of 0.3, an initial void ratio of 0.8 and an initial effective stress of 100 kPa at mid-layer, and the one-dimensional consolidation result is about 93 mm. That is not a rounding error on somebody's building; it is a structural conversation.
Read that figure honestly, because the formula behind it assumes a normally consolidated clay compressing along its virgin line. A heavily overconsolidated deposit — most of the stiff clays under British and North American cities — reloaded to a stress still well below its preconsolidation pressure moves a small fraction of that, along a recompression line the calculator does not model. So the 93 mm is the shape of the risk in a soft normally consolidated deposit, not a prediction for a site in stiff clay, and the number that matters for damage is differential settlement anyway: a uniform 90 mm under a whole building does far less than 20 mm of tilt across a ten-metre frontage, and the cone guarantees the near side settles more than the far side.
A second, faster mechanism gets confused with this one and should not be. Consolidation settlement is broad, slow and roughly predictable; loss of ground from a well pumping fines out of the aquifer is sudden, local and shaped like a hole. Any point still producing sand after development has the wrong slot size or filter pack, and it is quietly excavating the ground under whatever stands nearest. The consolidation risk itself is mitigated the way it always is — recharge wells outside the line, a cut-off to keep the cone inside it, or simply less drawdown, taking the last few hundred millimetres in a sump rather than with the ring.
The added stress to enter is the drawdown multiplied by 9.81 — that is the whole of the dewatering load. Layer thickness and both stress boxes follow the metric and imperial switch, in metres and kPa or feet and psf, so on the imperial page the added stress is the drawdown in feet multiplied by 62.4.
The clay's compression index, from a laboratory oedometer (consolidation) test.
The thickness of the compressible clay layer.
The clay's initial (in-situ) void ratio.
The effective overburden stress at the mid-height of the clay layer before loading.
The additional stress applied by the new structure or fill, at the same depth.
Primary consolidation settlement
5.81 in
This gives ultimate PRIMARY consolidation settlement only, assuming the clay is normally consolidated — overconsolidated clays need a modified formula with the preconsolidation pressure, and secondary (creep) settlement is not included here.
- Equivalent in meters
- 0.15 m
They open the calculator with your figures already in it
Primary Consolidation Settlement Calculator: 5.81 in — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 5.81 in — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- Primary consolidation only. Immediate (elastic) settlement that happens as the load goes on, and secondary compression (creep) that carries on after the excess pore pressure has dissipated, are both outside this figure — and for organic and highly plastic clays the creep component can be a large share of the total. What a structure actually settles is the sum of all three.
- The formula is the normally consolidated one. There is no preconsolidation pressure input, so an overconsolidated deposit — a weathered near-surface crust, ground that has been previously loaded, desiccated or drawn down — is treated as if it had never seen a higher stress, and the result overstates its settlement. A load that starts below the preconsolidation pressure and finishes above it needs the two-part recompression-plus-virgin calculation instead, which this page does not perform.
- This is the end-of-primary magnitude, not a timescale. Coefficient of consolidation, drainage path length and degree of consolidation play no part in the arithmetic, so the answer says nothing about how much settlement will have occurred by handover, how long a surcharge has to stay in place, or whether vertical drains would change the programme.
- One layer, one stress, one point. The added stress you enter is applied uniformly through the full layer thickness, whereas the real stress under a footing or an embankment decays with depth. Enter the mid-depth value the help asks for and a layer that is thick relative to the width of the load is materially mis-estimated — under-predicted, substantially so in the worst cases — so split it into sub-layers, each with its own added stress and its own initial effective stress, and add the results, as you must anyway for a profile with several compressible strata. Compression is assumed one-dimensional, with no lateral yield or squeezing out from under the load.
- Compression index is capped at 1.0 and initial void ratio at 2.5, and a laboratory value above either is replaced with the limit rather than refused — highly plastic, organic and peaty deposits therefore come out low. This is also a settlement magnitude, not a foundation design: it applies no tolerable-settlement criterion, returns nothing about differential settlement between points, and does not substitute for a site investigation with oedometer testing interpreted by a geotechnical engineer.
Keeping the water out instead of taking it away
The exclusion options are a spectrum, and they are usually chosen on the ground they have to cut through rather than on price. Sheet piles drive quickly in sand and silt and stop dead against cobbles or obstructions. Contiguous and secant bored pile walls handle mixed and obstructed ground and give a wall you can prop hard, with the secant version being the one that actually holds water. A bentonite-cement slurry trench is a cheap perimeter curtain with no structural duty. Diaphragm wall panels, excavated under bentonite support fluid, do both jobs at once and go deepest. Jet grouting and ground freezing answer the cases where no wall can be got in — under an existing structure, around live services, in a shaft bottom.
The distinction that matters most on a groundworker's programme is whether the cut-off is complete or partial. A wall toed properly into a low-permeability stratum turns the site into a box and reduces the internal pumping to seepage through construction joints and whatever falls in as rain. A wall hanging in the sand with no impermeable layer to key into does not stop anything; it lengthens the seepage path, so you still pump inside, at a much reduced rate, and you have imported a new check — the exit hydraulic gradient at the base inside the box, where the flow that came under the toe now rises through your formation. That is the classic quick condition, and it is worth remembering that the critical gradient for ordinary soils is close to unity, so the margin between a firm base and a boiling one is not wide.
The trade-off nobody quotes at tender is lateral load. A dewatered excavation has water pressure behind the shoring largely relieved and the wall designs to earth pressure; an excluded one keeps the full hydrostatic profile there for the whole open period, so wall section and propping both get heavier. What is saved in pumps, power, consents and monitoring reappears in steel and in props that get in everybody's way. And on some sites exclusion is not a preference but the only route — where an abstraction licence will not be granted, where a contaminant plume sits inside the cone, or where the neighbours' settlement risk is unacceptable under any monitoring regime.
| Question | Pumping the table down | Excluding the water |
|---|---|---|
| Where does the pore pressure change go? | Outside the boundary, out to the radius of influence, in whatever direction the ground favours | Stays inside the box, provided the cut-off is complete and toed into a low-permeability layer |
| What does the retaining structure carry? | Earth pressure, with hydrostatic largely relieved behind the wall | Earth pressure plus the full water pressure for the whole open period, so heavier sections and heavier props |
| What is the failure mode if it stops? | Water returns, formation softens, base uplift or boiling — the pumps are load-bearing and cannot be switched off | Loss of support fluid level, or seepage under an incomplete toe raising the exit gradient at the base |
| What are you monitoring? | Piezometric levels inside and out, discharge flow, sand content, and precise levels on every neighbour in the cone | Fluid level against external water level, wall and prop movement, and base seepage inside the box |
The balance is a level, not a mud weight
Where the exclusion route runs through a slurry trench or a diaphragm wall panel, the hole is open and unsupported for hours before anything permanent goes in it, and what holds the sides is fluid pressure. Bentonite suspension does this in two ways at once: the column has weight, and the excess pressure drives fluid into the pores of the soil until a low-permeability filter cake forms on the face, against which the remaining excess pressure can act as a genuine support. In clean sand that cake forms readily. In open gravel or cobbles the fluid runs away before it can build one, and the level drops faster than the pumps can top it up — which is why loss zones get plugged, or the mix gets modified, before the grab goes in.
The arithmetic holds a surprise for anyone who assumes the mud weight does the work. Fresh bentonite suspension runs around 10.2 to 10.5 kN/m³ against water at 9.81 — a difference of less than 0.7. Check the site case: a panel at 15 m, fluid level held 0.5 m below the guide wall top, groundwater at 3 m, slurry at 10.5 kN/m³. The net stabilising pressure is 34.5 kPa, and of that, 26.3 kPa comes from the 2.5 m of column that stands above the groundwater level and only 8.3 kPa from the density difference over the 12 m the two columns share. Three quarters of your support is the level, not the mud.
That also tells you where to run the check. Because the slurry is heavier than water, the net pressure grows with depth, so the toe is never the critical section — of the depths where this comparison means anything, the tightest is the groundwater table itself, and below it the margin only improves. Above the table the page keeps returning smaller numbers, down to nothing at the fluid surface, and none of them is a stability check: there is no pore pressure up there to balance, and what holds a face above the water table is the soil and the filter cake rather than the difference between two columns. Check at the table first, then check again at any granular lens or soft layer where the soil, rather than the fluid balance, is the weak link. And watch the sensitivity in the other direction: raise the external water to ground level, as a tidal river or a wet fortnight or a leaking main will, and the same panel drops from 34.5 kPa to 5.1 kPa. That loss is exactly 9.81 kPa for every metre the water rose, and it works the same way if the fluid level falls instead — about 10.5 kPa per metre, and faster, because a fluid level can drop in minutes.
Two practical notes on the form. Slurry unit weight is entered in kN/m³ on both the metric and imperial pages while every depth follows the switch, so multiply a mud balance reading in specific gravity by 9.81 — 1.07 SG is 10.5 kN/m³, and 66.8 pcf if that is what the balance reads. The result converts, so an imperial page returns the same check as about 5.0 psi. And note what a polymer support fluid does to this check. Enter the field's floor of 9.81 kN/m³ on the same panel and the answer is 24.5 kPa rather than 34.5: the density difference has gone entirely, and all that remains is the 2.5 m of column standing above the groundwater level, now at water's own 9.81 per metre. Seventy-one per cent of the bentonite figure survives, which is the same point the paragraph above made from the other side. That is arithmetically correct and practically incomplete, because polymers stabilise through viscosity and seepage drag rather than through a filter cake, so the part of their support this page cannot see is the part that does the work. They are covered on their own terms in FHWA-NHI-10-016 and in the Federation of Piling Specialists' guidance on bentonite support fluids. BS EN 1538 requires the support fluid level to be kept above the groundwater level at all times; the margin itself belongs to the specification on your job, and a bare positive number on this page is not that margin.
Set the depth being checked to the groundwater table depth first — the shallowest depth at which the comparison means anything, and the tightest of the ones that do — then re-run it at the toe and at any weak layer, and once more with the external water at the highest level the site can credibly reach.
The depth at which you're checking the pressure balance.
The slurry's unit weight.
How far below the ground surface the slurry level is maintained.
The depth to the groundwater table below grade.
Net stabilizing pressure
5.14 psi
Positive net pressure — the slurry column is stabilizing the excavation wall at this depth. Standard practice still recommends a comfortable margin (often citing at least 1-1.5 m / 3-5 ft of extra slurry head), not just a bare positive value.
- Slurry column pressure
- 22.05 psi
- Groundwater column pressure
- 16.91 psi
They open the calculator with your figures already in it
Drilled Shaft Slurry Pressure Balance Calculator: 5.14 psi — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 5.14 psi — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- This is a hydrostatic pressure comparison, not a borehole stability design. It carries no factor of safety and takes no account of the soil's own lateral earth pressure, its strength, or surcharge from the rig, casing, spoil piles or adjacent traffic. A positive net pressure is a necessary condition for a slurry-supported wall, not proof that the wall stands.
- It checks one depth. Nothing here scans the shaft profile, and because the slurry column grows faster than the water column with depth, the margin is tightest in the upper shaft just below the water table. A positive number at the toe says nothing about the section above it.
- Groundwater is modelled as a single hydrostatic table with its surface at the depth you enter. Confined or artesian layers whose pressure head stands above that level, perched water, and any head standing above ground are outside the model, and each raises the real water pressure above the figure shown.
- Slurry quality is reduced to one unit weight. Sand content, viscosity, pH, filter-cake formation and fluid loss into coarse or gravelly ground are not modelled, and the model does not distinguish bentonite from polymer. Polymer supports the wall largely through seepage forces into the formation rather than by building a filter cake, so the same net pressure does not buy the same support.
- The result is a snapshot of the level and density at the moment they are measured. Slurry lost into permeable strata, the drop between top-ups, and the transient suction as a bucket or auger is withdrawn all push the real pressure below the calculated value. Concrete placement, where tremie concrete displaces the slurry, is a separate check.
The pumps stop being plant and become temporary works
The moment formation stability depends on drawdown, the pumping installation is a structural element and belongs in the temporary works register. In the United Kingdom that is BS 5975 procedure — a design, a design check, a permit to load and a permit to strike, with a named temporary works coordinator. In the United States, 29 CFR 1926 Subpart P puts water accumulation and the equipment controlling it under the competent person, monitored rather than assumed. Either way, standby pumps, independent power with automatic start, level alarms with an out-of-hours call-out and fuel for a bank holiday weekend are not contingency items; they are the design.
The commissioning sequence matters as much as the kit, because the whole scheme rests on a permeability figure that was, until the pumps ran, an estimate. Running the system is the first real test of that estimate, and it is far cheaper to discover a factor-of-three error before the dig goes below the original water table than after.
- Install and dip every piezometer — inside the dig, outside it, and at least one well beyond the predicted cone — and record baseline levels over enough days to see any natural or tidal variation. Once the ring runs there is no baseline to go back for. ASTM D5092 covers the installation.
- Take precise levels on every structure and buried service inside the predicted radius of influence, photograph existing cracks with a scale in frame, and agree the amber and red trigger levels in writing before anyone needs them.
- Install the wells or wellpoints and develop each one until it runs clear, recording sand content on every point; a point still producing fines is removing ground, not water.
- Start the system in stages rather than all at once, and log drawdown against time in the observation wells at every stage.
- Compare the measured flow and drawdown against the design cone. The back-figured permeability from that comparison is worth more than any laboratory value, and if it is far off, redesign before digging rather than adding pumps later.
- Only then take the excavation below the original water table, and keep dipping the observation wells daily for as long as the hole is open.
Where the water goes, and when you are allowed to stop
Pumped groundwater is abstracted water and then it is a discharge, and both ends are regulated. In England and Wales that means abstraction licensing under the Water Resources Act 1991 and a discharge permit under the Environmental Permitting (England and Wales) Regulations 2016, with SEPA and Natural Resources Wales holding the equivalent roles; in the United States, discharge to surface water runs through an NPDES permit under the Clean Water Act, issued by the EPA or a delegated state agency. Lead times run to weeks and occasionally months, so these belong at tender rather than at mobilisation — and discharge to a foul or combined sewer is a separate agreement with the undertaker, not a fallback you can assume.
Whatever the route, the water usually needs treating before it goes anywhere. Fine sand and silt come out of a new system in quantity and drop out only slowly, so settlement tanks or a lagoon sized on the actual flow, with a turbidity limit somebody is measuring, is normal rather than exceptional. Where the site or its neighbours are contaminated, the discharge question becomes the whole question: pumping draws whatever is dissolved in the groundwater toward the excavation, which can turn a modest dewatering scheme into a treatment plant with a permit to match, and is one of the commonest single reasons a job switches from pumping to a full cut-off.
Last, the pumps come off on a date the structure decides, not the programme. Once the water returns, the completed basement is a boat, and it has to resist uplift on its own dead weight, on tension piles or ground anchors, or on a permanent under-slab drainage system with a maintained pump and somebody's name against maintaining it. Eurocode 7 handles that as the same UPL limit state the base plug was checked against at the start, and the check runs on the highest credible groundwater level rather than the one that happened to be there in a dry August. Record where the water table came back to, and how long it took, because the next contractor on that street will read your record and it will be better than the one you started from.
What has to be true before the machine goes below the water table
None of these are procurement items, which is why they get left out. Each one is a thing the dewatering design is already assuming somebody did.
- Pumping test permeability, not a laboratory figure — It is the input every other number multiplies, and it varies over orders of magnitude within one site. A disturbed-sample test measures the sample, not the deposit.
- Piezometers inside, outside and beyond the predicted cone — Dipped and logged over enough days to catch tidal and seasonal movement before a pump runs. There is no way to recover a baseline afterwards.
- Precise levels and crack photographs on every structure inside the radius — With trigger levels agreed in writing. A monitoring regime introduced after the first complaint has no before to compare against.
- Standby pump, independent power with autostart, and an out-of-hours alarm — Once formation stability depends on drawdown, a stopped pump is a stability failure rather than a delay, and it will stop at three on a Sunday morning.
- Settlement tanks or lagoon sized on the actual discharge, with the consents in hand — Abstraction and discharge authorisations have lead times in weeks. Silt has to drop out somewhere before the outfall, and somebody has to be measuring turbidity.
- Sand-content checks on every point after development — Continued fines means the slot size or filter pack is wrong, and the well is excavating the ground under whatever stands nearest rather than draining it.
Opens the calculators above on one screen with the dimensions from this article already filled in. Quantities only — this site publishes no price list, because local prices vary too much to publish honestly.