Tanking a Basement Wall Against Standing Water

Nobody tanks against the level on the day

The design water level is a decision, not an observation, and it is the number every other number on the job hangs off. Water in an open excavation tells you where the table is this week, under this rainfall, with the site drainage in whatever state it is in. It tells you nothing about February, about the river two hundred metres away at its hundred-year stage, about the tidal cycle if the site is estuarine, or about what happens when the neighbouring development stops its own dewatering. Standpipe piezometers installed and developed to ASTM D5092, read over a period long enough to catch a seasonal swing, are the only honest input. One trial pit dug at nine in the morning is not.

Work the level to the top of the structure where the ground can credibly deliver it, then design for that. It matters twice over. Laterally, a 3 m submerged wall height carries a triangular water pressure diagram with a resultant near 44 kN per metre run — roughly 3,000 lb on every foot of wall — acting in addition to the effective soil pressure, and a wall detailed for damp ground is not a wall detailed for that. Vertically, the same head pushes up on the underside of the slab across the whole footprint, and a light basement box below the table wants to float. Eurocode 7, BS EN 1997-1, treats that as its own class of limit state alongside heave and internal erosion, and it is checked against the buoyant weight of the structure and whatever the design chooses to mobilise against it, not against the weight of the walls alone. Get the design level wrong and you have not misjudged a waterproofing detail; you have misjudged the structure.

What the ground can actually deliver to the wall

Two sites at the same head behave nothing alike. A basement in stiff glacial clay two metres below the table sits in ground that will not pass measurable water in a working lifetime; the same basement in a sand and gravel terrace is a hole in an aquifer, and the difference is permeability, which spans about ten orders of magnitude across ordinary soils. That range is why nobody should be estimating a below-grade job from head alone, and why the first arithmetic on the page is Darcy's law rather than an area takeoff.

Q = k i A. The hydraulic conductivity k is the soil property; the gradient i is the head lost across the flow path divided by the length of that path; A is the cross-sectional area the flow crosses. At a basement wall the useful interpretation of A is the vertical plane of submerged wall face the water has to cross to reach the excavation or the drainage layer, and the useful interpretation of i is the head difference between the far-field table and whatever level is being held at the wall, divided by the distance over which that drop occurs. Where a drainage composite holds the face at atmospheric pressure the drop is steep and local, which raises i sharply — a fact worth noticing, because relieving pressure at the face is exactly the same thing as increasing the seepage the relief system then has to carry.

Get k from a test, not from a table, and be honest about which test. Laboratory values come from a flexible-wall permeameter to ASTM D5084 for fine-grained material, or the constant-head method of ASTM D2434 for granular soils; field values come from a slug test to ASTM D4044 or, better on any job worth the money, a pumping test. Field values are usually higher and always more relevant, because a laboratory specimen has lost the fabric — the sand partings, the fissures, the relic root channels — that actually conveys water. Layered ground is anisotropic by an order of magnitude or more, horizontal conductivity over vertical, and a single k entered into a calculator is a horizontal-flow number whether you meant it to be or not.

Run the figure and then read it for what it is. A 3 m by 8 m submerged wall face in silty sand at 1 × 10⁻⁵ m/s under a gradient of 0.5 delivers about 0.12 litres per second — roughly 10 m³ a day, near 2,700 US gallons — which is a trivial flow for a pump and a serious flow for a barrier that is supposed to admit none of it. The number is not what leaks through a tanked wall; it is what the surrounding ground can hand to the excavation while it is open, and what a relief layer would have to carry forever if you chose to give the water somewhere to go. Darcy assumes laminar flow through saturated, homogeneous, isotropic soil, so treat a result in open gravel or fissured rock as an order of magnitude rather than a quantity.

A panel that only works once it is squeezed

Bentonite waterproofing is a granular sodium bentonite core held between a woven and a non-woven geotextile, or bonded to an HDPE carrier sheet, and it is unusual among barriers in that it is not waterproof when you install it. It is a dry clay in a bag. It becomes waterproof by getting wet: the sodium montmorillonite takes water into its interlayers, the clay expands many times its dry volume, and the resulting gel has a hydraulic conductivity in the region a manufacturer will quote against a flexible-wall index flux test to ASTM D5887. The material's whole appeal below the table follows from that mechanism — a gel can flow into a form-tie hole, close a shrinkage crack and re-seal itself around a puncture, which is exactly the kind of small local defect that sinks a sheet membrane.

The mechanism has a condition attached, and it is the condition that gets skipped. Swelling clay does one of two things depending on what is around it. Confined, it develops swelling pressure against the restraint and consolidates into a dense, low-permeability seal at the interface. Unconfined, it simply expands, loses density, disperses and washes away, and the panel that looked correct on Tuesday is a stained geotextile by the following spring. Every bentonite manufacturer states a minimum confining condition in its product literature — CETCO's Voltex data sheets express it in those terms — and it is expressed either as a depth of backfill over the panel or as a confining pressure. That figure is a performance requirement, not a recommendation, and it is why a bentonite panel on a blindside application against a shoring wall behaves differently from the same panel on a positive-side application backfilled with 200 mm lifts of well-graded fill.

Specify by test method, because the trade name tells you nothing. The relevant ones are all indices of the clay rather than of the finished panel: swell index by ASTM D5890, reported in millilitres per two grams and commonly specified at 24 mL/2 g or better for a sodium grade; fluid loss by ASTM D5891; mass per unit area by ASTM D5993, which is the number the swell arithmetic below actually consumes; and index flux by ASTM D5887. ASTM D6102 covers installing the material. A calcium bentonite, or a sodium bentonite that has undergone cation exchange in calcium-rich or brackish groundwater, gives up most of its swell — the exchange is progressive and irreversible, and it is the standard failure mode on coastal sites, on ground contaminated with hydrocarbons, and anywhere the groundwater chemistry was never sampled. Polymer-modified grades exist for exactly those conditions and cost what they cost.

The arithmetic is a two-step and worth doing on paper before the panels are on a pallet. Divide the bentonite content per unit area by the dry bulk density to get the dry clay thickness spread across the panel, then multiply by the volumetric swell factor for a confined condition to get the gel thickness that will actually stand between the water and the concrete. A panel carrying 5 kg/m² of clay at a dry density of 800 kg/m³ holds a layer only 6.25 mm thick before it sees water, and at a tenfold swell that becomes about 62 mm of gel. Both halves of that are informative: 6 mm is how little material is doing the work, and 62 mm is how much room the swell needs to find, which is a real question at a tight blindside gap or against a rigid protection layer that cannot yield.

Backfill is the last trade to touch the waterproofing

Everything above the panel exists to get it to its confined working state without damaging it on the way. Protection board is the first item and it is not optional on a job with standing water: it takes the point loading from angular fill, the gouging from a bucket working close to the wall, and the abrasion of settlement dragging down the face for the next decade. Board it as a plane, butted and taped or adhered per the membrane manufacturer, and take it below the level of the finished fill rather than stopping it where the backfill starts, because the corner at the base is where a bucket tooth arrives.

Then the fill itself, placed in controlled lifts and compacted with plant appropriate to the wall, which for a basement wall usually means hand-guided equipment in the first metre of the face and nothing tracked closer than the designer allows. Two constraints run at once. The wall is not designed for one-sided backfill until it is propped by the ground floor slab or by temporary props, and a permanent basement wall pushed inward at first lift is a demolition, not a defect. And the confinement the bentonite needs builds with the fill, so premature hydration — a panel left exposed through a wet week before any fill arrives — swells the clay with nothing to swell against and produces a low-density gel that will never recover its density when the fill finally lands. Sheeting exposed panels, and sequencing the fill to follow the waterproofing rather than the other way round, is what stops that.

Where a drainage composite goes in against the tanking, it goes over the protection board, not directly against the panel, and its geotextile face turns toward the soil. The rule is uncomfortable but consistent: nothing that channels water is allowed to sit between the barrier and the concrete, because a drainage layer inside the barrier is a distribution system for whatever gets past it.

Water finds the joints, so the joints get designed twice

A tanked wall almost never fails through the middle of a panel or the body of a pour. It fails at the kicker joint, at the day joint on a long wall, at the tie holes, at the pile heads breaking through the slab, and at every service passing through the box. Those are not incidental details; on a typical basement they are the majority of the risk, and they are the reason BS 8102 pushes toward combined protection rather than trusting a single barrier over a structure that is full of discontinuities.

The concrete side of the answer is Type B: a low-permeability mix, adequate cover, and crack widths controlled by reinforcement and by joint layout rather than by hope. ACI 224R, Control of Cracking in Concrete Structures, is the standing reference for the crack-width thinking, and ACI 350, the code for environmental engineering concrete structures, is where the liquid-tightness discipline is written down properly — it is a code for tanks, and a basement below the table is a tank with the pressure on the outside. Waterstop is the joint-by-joint part of it: PVC or rubber profiles cast continuously into every construction and expansion joint that has to hold water, spliced by heat welding rather than by overlap and taped, with proprietary fittings at corners and T-intersections. The US Army Corps of Engineers specification CRD-C 572 covers the PVC material. Hydrophilic strips are the other common tool and behave like the bentonite above — they swell to seal, which means they must be positioned where they are confined, and they must not be allowed to hydrate before the pour arrives.

Take the joint length off the drawing rather than off the wall length. On a rectangular basement the kicker joint runs the full internal perimeter, then every vertical day joint in the wall adds its height, then the slab joints and every construction joint at a return. It adds up faster than crews expect, and waterstop is the item that stops a pour at four in the afternoon because a corner fitting is missing. Carry a real waste allowance: splices, corners and cut-and-fit around reinforcement all consume material that a straight-run measurement does not see.

Resist it, relieve it, or both, and know which you have chosen

A drainage composite against a tanked wall is not decoration and it is not a second membrane. It is a decision to cut the head at the face, and it changes the whole problem: the barrier now sees a much lower pressure, the seepage the ground delivers goes up because the gradient at the wall went up, and something downstream has to accept that flow in perpetuity. Above the water table this is nearly free. Below it, the composite discharges to a drain that is itself below the standing level, which means the discharge has to be pumped, which means the building has committed to a permanently powered, permanently maintainable system for the life of the structure — the Type C bargain, and a legitimate one when the sump, the duty and standby pumps, the alarm and the access are all designed as a maintainable installation rather than added at handover.

Check the arithmetic against the seepage before this becomes an argument. A composite rated at 5 litres per second per metre of width, over a 40 m perimeter, has a nominal capacity around 200 litres per second. Put the seepage on the same 40 m of wall before comparing: the silty-sand example delivered 0.12 litres per second across an 8 m face, so the same ground across the full perimeter hands over about 0.6 litres per second — still more than two orders of magnitude below what the sheet is rated to carry. The drainage plane is essentially never the constraint. The collector pipe, the sump, the pump curve at real total head and the outfall are the constraint, every time, and the rated capacity of the sheet is quoted at a stated compressive load that the backfill will exceed if nobody checks. Design the chain, not the sheet.