Backfilling Against Waterproofing

Ask the wall before you ask the machine

A basement wall is not designed to be filled against on its own. It spans vertically between the footing at its base and the ground-floor diaphragm at its head, and until that deck is on and connected, filling one side turns a propped wall into a cantilever it was never sized as. The International Residential Code puts this in Section R404.1.7, Backfill placement: fill is not to go against the wall until it has sufficient strength and has been anchored to the floor above or braced well enough that the backfill cannot damage it, with an exception only for walls supporting less than four feet of unbalanced backfill. The failure it is written against is familiar — a horizontal crack near mid-height, or a wall pushed inward off its kicker, discovered later when the ground-floor setting-out will not close.

Strength is two separate questions and site collapses them into one. The strength at which formwork can come off is a formwork question, covered by ACI 347R for the removal of forms and shoring; the strength at which the finished wall may be loaded laterally comes out of the structural design to ACI 318 and is named in the specification. They are not the same number and in cold weather they can be a fortnight apart. Field-cured cylinders taken from the actual pour, kept beside the wall, are what answers it — not the calendar, and not the pour ticket. On a grouted masonry wall the same logic applies to the grout cure and to the bond beam that ties the head of the wall into the deck.

The load itself depends on whether the wall is free to move, and a propped basement box is not. IBC Section 1610.1 draws the line explicitly: walls whose horizontal movement is restricted at the top are designed for at-rest pressure, while a retaining wall free to move and rotate at its head is permitted the active case. The difference is not small. Jaky's relation gives the at-rest coefficient as one minus the sine of the effective friction angle, so a well-graded granular fill at 34 degrees sits at about 0.44, and against a fill weighing 20 kN per cubic metre that is roughly 8.8 kPa of lateral pressure for every metre of depth — near 56 pounds per square foot per foot in the other system. Over 2.7 metres of fill the triangular diagram integrates to about 32 kN on every metre run of wall, close to 2,200 pounds on every foot, acting a third of the way up from the base. Table 1610.1 tabulates minimum design lateral soil loads by soil group with separate active and at-rest columns, subject to a geotechnical investigation under Section 1803 saying otherwise, and AASHTO's LRFD Bridge Design Specifications, Section 3.11, additionally requires compaction-induced pressures to be accounted for.

The number matters less than what it is attached to: a soil group in a moist condition at its optimum density, which is how the code table states its own basis before adding that actual field conditions govern. The designer picked a value from a row, and the row assumed a material. Turn up with site-won clay because the granular fill did not arrive and you have changed the load case without changing a drawing, in the direction the wall cannot take; let the same fill saturate because the drainage layer was never connected and you have added a water pressure the row never contemplated. Fill evenly around the structure as well — two metres against one elevation and nothing against the return is a net push on a box that was designed for a balanced one, and it is the reason temporary props are designed props rather than a scaffold tube wedged across a corner.

The board is impact protection, and it is bought in sheets

Protection board exists to take the loads the membrane cannot: point pressure from angular stone bearing on a sheet a millimetre and a half thick, gouging from a bucket working close to the face, and the slow abrasion of a settling fill dragging down a bonded membrane for the next decade. That last one is the quiet killer, because settlement drag pulls at laps rather than at the field of the sheet, and a lap that has been worked loose leaks at the one line nobody photographed. The product is usually an asphaltic core board of the W. R. Meadows PROTECTION COURSE type, extruded polystyrene, or a dimpled drainage composite such as Carlisle CCW MiraDRAIN or GCP Hydroduct doing protection and drainage in one pass. If you are using XPS, check whose drawing set its thickness — an insulation thickness chosen by the thermal model is not automatically the impact protection the membrane manufacturer asked for.

Fixing is where good boards get installed badly. Below the waterproofing termination, nothing is fastened through the membrane plane: not a nail, not a washer, not a hit-and-drive anchor. The board is held by adhesive dabs, by double-sided tape, or by a mechanical fix taken into the wall above the membrane's top edge, and the membrane manufacturer's literature governs which of the three is permitted on their sheet. Take the board right down to the footing rather than stopping where the fill is expected to start, because the base corner is precisely where a bucket arrives when the driver is scraping the last of the stone off the formation.

Counting it is straightforward arithmetic with one trap in it. The sheet count is the below-grade wall area divided by the area of one sheet, rounded up, because a part sheet still costs a whole sheet. The trap is that the two standard boards are not the same board: 1.2 by 2.4 metres is 2.88 m², and 4 by 8 feet is 32 sq ft, which is 2.97 m² — about 3% apart. On a wall 2.7 m high round a 46 m perimeter, 124 m² of face takes 44 metric boards or 42 North American ones, and counting on the larger imperial sheet while metric boards arrive on the wagon leaves you a full board short at the last elevation on a Friday afternoon. Take the area off each face separately: every side of a pilaster is a face, every step in the footing changes the height, and the return at a light well is real area that a perimeter-times-height figure never sees.

Place it, do not tip it

Lift thickness is decided by the compaction plant, never by how quickly the void needs to close. A hand-guided plate will densify roughly 150 to 200 mm of loose granular material in one go; a rammer in a confined corner does less; a roller does far more but is not allowed anywhere near the wall in the first place. The lift count falls straight out of it — 2.7 metres of fill in 200 mm lifts is fourteen circuits of the whole perimeter, which is a real figure to have in front of you when someone proposes closing the dig in an afternoon. Material matters just as much: well-graded granular fill to the specified grading, with the maximum particle size capped, because one 150 mm cobble bearing on a board in the first lift is the puncture that nobody will ever find.

How the fill arrives at the wall is the difference between a protected membrane and a scarred one. Stone is placed away from the face and worked toward it, or lowered and released at low level; it is never dropped from a raised bucket, and the first lift in particular goes in by hand or with a chute where access allows. Keep tracked plant off the fill inside the standoff the designer sets, and remember that the pressure heavy compaction locks into the backfill does not relax when the machine leaves the site. If people are working in the void as it fills, the excavation is still an excavation, with everything 29 CFR 1926 Subpart P requires of it, right up until the last lift.

1. Confirm the board or composite is continuous and undamaged across the elevation about to be filled, and photograph it before anything is placed.
2. Tip the load away from the wall and push or lower it into position, so nothing lands against the protection layer from height.
3. Spread to the loose lift thickness the plant can actually compact, checking the depth at the wall face rather than out in the middle.
4. Compact from the wall outward with hand-guided equipment inside the standoff, keeping the number of passes consistent lift to lift.
5. Test to the specified frequency at the face as well as away from it, and carry the lift number and test result forward on the same record sheet.
6. Bring the fill up evenly around the whole structure rather than completing one elevation at a time.

A number, not an opinion

Percent compaction is field dry density over the laboratory maximum for that material, times a hundred, and every part of that sentence gets abused on site. The laboratory maximum comes from a Proctor test run at a stated compactive effort: ASTM D698 for standard effort, ASTM D1557 for modified. The same soil gives a higher maximum under modified effort, so 95% of one is emphatically not 95% of the other, and a specification that names a percentage without naming the effort has not specified anything. Worse, a Proctor run on the borrow that was in the ground during the site investigation tells you nothing about the material the haulier delivered last week. Classify what actually turned up, to ASTM D2487, and run the Proctor on that.

Field density comes from ASTM D6938 with a nuclear gauge, ASTM D1556 with a sand cone, or ASTM D2937 with a drive cylinder, and each reads only the depth it reaches — a gauge probed to 150 mm is telling you about one lift and not about the three beneath it. Test at the wall face, not three metres out where the plant could get a proper run, because the strip within a metre of the board is the strip that will be under-compacted and it is the strip whose settlement drags on the membrane.

Clean open-graded stone is the exception that gets fudged. Uniformly graded 20 or 40 mm drainage stone has no meaningful Proctor curve, so specifying 95% modified against it writes a requirement that cannot be complied with — and a requirement that cannot be met on site gets quietly waived, which leaves the drainage zone as the one part of the fill under no control at all. Control it by relative density instead, using the index densities from ASTM D4253 and ASTM D4254, or by method: a named machine, a named lift, a named number of passes, agreed before the stone is ordered.

The sheet is not the bottleneck, and it never is

A dimpled drainage composite works by holding an open gap against the wall so any water that reaches the face falls under gravity at atmospheric pressure instead of standing against the membrane as head. Its geotextile faces the soil, its dimples face the wall, and it only does its job if it is continuous from the top termination down into the stone at the footing. The single most common installation error is a composite that stops a foot above the drain: water runs down the plane, arrives at the bottom, meets soil, and saturates the very strip of backfill the sheet was installed to keep dry.

The rating on the data sheet comes from ASTM D4716, which measures in-plane flow rate per unit width at a stated normal stress and hydraulic gradient. Read the figure at the stress your fill actually applies, not the headline value, and be aware that transmissivity falls over time as the core creeps and the geotextile intrudes into it. One usage note on the calculator below: its rated-capacity field is expressed per metre of width and stays per metre whichever unit system the page is set to, while the width field follows the metric and imperial toggle — so take the per-metre figure straight off the manufacturer's table and let the width look after itself.

Then run the comparison, because it settles an argument that recurs on every job. A composite rated at 5 litres per second per metre, over a 46 m perimeter, has a nominal capacity of 230 litres per second — roughly 3,600 US gallons a minute. Now look at what has to accept it: a 100 mm perforated collector at a 1% fall, flowing full, with a Manning roughness of 0.010, carries something like 6.7 litres per second, about 105 gallons a minute. That is a thirty-fourth of the sheet's rating, and it is the honest constraint. And what actually arrives is smaller again: the backfill trench round that perimeter is perhaps 46 m² of loose ground open to the sky, so 25 mm of rain in an hour puts a little over a cubic metre into it, an average of about a third of a litre per second even if every drop went straight down the fill.

So the design attention belongs downstream of the sheet and at its two ends. Filtration first — a geotextile chosen against filtration criteria of the kind set out in AASHTO M288, with the apparent opening size to ASTM D4751 and permittivity to ASTM D4491 suited to the soil actually behind it, because a blinded fabric turns a drainage plane into a wall covering. Then the base connection into the stone envelope and the perforated line, whose bedding and backfill follow ASTM D2321; then the fall, the cleanouts and an outfall above the invert it discharges to. IBC Section 1805.4 and IRC Section R405.1 are where the subsoil drainage requirements themselves sit, and BS 8102:2022 makes the same point from the other direction by treating drainage and protection as parts of the design rather than as site accessories.