Sizing Driven Piles in Soft Clay

The Surcharge Is Still Moving

The platform was surcharged three weeks ago — call it a metre and a half of imported granular fill, spread over the whole footprint, tracked in and rolled, with settlement plates set on the original ground before any of it went down. The rig comes on Monday. Almost everything difficult about this job lives in the gap between those two dates, because the fill started something in the clay underneath that will run for years, and the piles are going in while it is still running.

Preloading is meant to be a favour to the foundation. Squeeze the settlement out of a compressible layer before the building arrives, and the building inherits ground that has already done its moving. That logic holds for a slab on grade. It inverts for a pile. A pile driven through a layer that is still consolidating is a rigid object standing in ground that is sinking past it, and the friction along that stretch of shaft — the same adhesion that would be capacity on a settled site — reverses direction and pulls down. Nothing about the pile changed. The sign of the relative movement did.

So the number the structural engineer needs is not the column load. It is the column load plus whatever the settling ground hangs on the shaft on its way past, and that second term is often the larger surprise on a soft site. Getting it involves three separate pieces of arithmetic that have to be done in order, because each one sets an input for the next: how far the clay still has to settle, how deep the drag reaches before it turns back into support, and how much shaft is left below that depth to carry anything at all. One more thing worth fixing early — the added stress driving the consolidation comes from the fill, not from the building. The building's weight travels down the piles and is delivered below the soft layer. It is the earthworks that load the clay.

The Depth Where the Ground Changes Sides

Somewhere down the shaft the soil stops overtaking the pile and the pile starts overtaking the soil. That depth is the neutral plane, and it is the single most useful idea in this whole subject because it is the boundary between the two halves of the calculation: everything above it is load, everything below it is resistance. It can be defined two equivalent ways — as the depth where pile settlement and soil settlement are equal, or as the depth where the force in the pile reaches its maximum, which is where the load curve descending from the head meets the resistance curve rising from the toe.

Where it sits is not a property of the soil alone. Raising the sustained load on the pile pushes the neutral plane up, which shortens the drag length and lengthens the resisting length. Founding the toe on something very stiff — rock, or a dense gravel — pushes it down, which is the counterintuitive result that a better toe attracts more drag. It also means the neutral plane is frequently below the base of the soft layer, sitting some way into the stiff clay beneath, so drag develops on ground that a layer-boundary sketch would tell you is competent. Locating it properly is an iterative construction described in Bengt H. Fellenius, Basics of Foundation Design, and set out for practitioners in the Canadian Foundation Engineering Manual.

Two distinct checks come out of the neutral plane and they fail in different ways. The structural check is the pile section at that depth carrying the sustained load plus the accumulated drag — a strength question, answered against the pile's own material capacity. The serviceability check is the pile head settlement, which is the soil settlement at the neutral plane plus the elastic shortening of the shaft above it plus whatever the toe moves. On soft sites the second one governs more often than engineers expect, and no amount of extra geotechnical capacity fixes it, because the pile is being carried down by ground it is not relying on.

How the drag is combined with other actions is a code decision rather than a calculation, and the adopted documents do not all decide it the same way. EN 1997-1 Eurocode 7: Geotechnical design — Part 1: General rules treats downdrag explicitly as an action, and the AASHTO LRFD Bridge Design Specifications carry it as a load with its own factor. The recurring argument concerns transient live load: at the point a pile plunges, the relative movement along the upper shaft reverses and the drag disappears, so combining full drag with full live load in a geotechnical capacity check double-counts. Resolve that against the governing document, on the record, before you resolve it in a spreadsheet.

Choosing Alpha in the Direction That Costs You

The adhesion factor is empirical and always has been. It exists because the shear surface that forms against a driven shaft is not the same as the one a laboratory sample fails on: driving remoulds the clay, generates excess pore pressure, and then the soil reconsolidates against the pile over weeks. What the correlations capture is that the ratio of mobilised adhesion to measured undrained strength falls as the clay gets stiffer, and that it also depends on the ratio of strength to vertical effective stress rather than on strength alone. The correlation in API RP 2A-WSD is the one most often carried into onshore work, and it is a chart to be read at a site's own strength ratio, not a constant to be memorised.

The direction of caution flips depending on which side of the neutral plane you are working. For shaft capacity, adhesion is a resistance and the low end of the plausible range is the safe assumption. For drag, adhesion produces a load and the low end is the dangerous one — a conservative downdrag estimate is a high adhesion factor, not a low one. This trips up experienced engineers because the habit of reaching for the smaller number is a good habit everywhere else in geotechnics. The negative skin friction calculator here defaults to a modest factor, which suits a first look; bracket it deliberately rather than accepting it. Running the same shaft at both ends of the credible range, as the 175 kN against 77 kN example shows, tells you whether the design is sensitive to a number nobody can measure directly.

There is a deeper caveat about using an undrained parameter for this at all. Drag accumulates over the years the clay takes to consolidate, which is a drained process, and the more defensible long-term formulation is an effective stress one: a beta coefficient applied to vertical effective stress along the shaft. The alpha expression is a screening tool for the drag case, useful for establishing whether the load is 50 kN or 500 kN and whether the shaft coating conversation needs to happen. It is not the calculation to put a signature on for a critical structure — that one comes from a full effective-stress analysis of the whole pile, load curve and resistance curve together.

Whatever is chosen, write down which correlation was read, at what strength ratio, and which side of the calculation it was used on. That single line in the calculation sheet is what makes the next engineer able to check the work rather than redo it.

Getting the Drag Off the Shaft

Once the number exists there are only four honest responses, and choosing between them is largely a question of programme. Time is the cheapest of them: leave the surcharge in place longer, with vertical drains if the layer is thick, and drive the piles into ground that has already done most of its settling. That is what a preload is for, and abandoning it three weeks in because the rig is available is how a site ends up paying for the surcharge twice.

The others cost money instead. A bituminous slip coating over the drag length is genuinely effective and long-established — its performance depends on the grade and thickness of the bitumen and on ground temperature, and its weakness is mechanical: driving through granular fill can strip it, so the coated zone frequently needs a sacrificial outer sleeve through the abrasive layers. A sleeve or double casing isolating the shaft is the same idea by another route. Or accept the drag and buy section: a heavier wall, a higher steel grade, or more piles, remembering that the last of those brings its own drag with it. The coating options are described in the Canadian Foundation Engineering Manual and in the FHWA Design and Construction of Driven Pile Foundations reference manual; execution requirements for driven displacement piles sit in BS EN 12699 Execution of special geotechnical works — Displacement piles.

1. Read the preconsolidation pressure against the effective stress the fill will produce, and stop here if the clay stays inside its recompression range.
2. Estimate remaining primary settlement at the planned driving date, using field settlement plate records where they exist rather than the original prediction.
3. Locate the neutral plane for the trial pile and toe condition, and record the drag length and the resisting length separately.
4. Compute the drag load over the drag length, bracket the adhesion factor at both ends, and check the total against the weight of soil within the group plan area.
5. Compute shaft capacity below the neutral plane only, layer by layer, and add the toe resistance from the report.
6. Check the pile section at the neutral plane for sustained load plus drag, then check head settlement against the structure's tolerance — and only then decide between waiting, coating and buying section.

What the Hammer Tells You, and What It Cannot

Driving records in soft clay are quietly misleading. End-of-drive resistance is low because the clay has been remoulded and the excess pore pressures generated by driving have not yet dissipated; capacity then climbs over days or weeks as the soil reconsolidates against the shaft, a recovery large enough in sensitive clays to change the answer several times over. A restrike after a specified waiting period is the observation that matters, and the waiting period belongs in the specification rather than in a conversation on site. High-strain dynamic testing to ASTM D4945 Standard Test Method for High-Strain Dynamic Testing of Deep Foundations, with signal matching analysis, is how that restrike gets turned into a capacity estimate. Where a static verification is required, ASTM D1143/D1143M Standard Test Methods for Deep Foundations Under Static Axial Compressive Load is the reference test.

Neither test measures downdrag, and it is important to be clear about why. A compression load test pushes the pile down relative to the soil along its entire length, which reverses the relative movement in the settling layer and removes the drag for the duration of the test. What the test measures is capacity. Drag load is a long-term action computed, not proved — and if it genuinely has to be measured, that means strain gauges cast or welded into the shaft and read over months, on a programme somebody has to pay for and wait for.

The information worth capturing during driving is cheap and gets thrown away constantly: blow counts against depth for every pile, hammer energy and stroke, the toe elevation actually reached against the one designed, and any pile that ran soft or refused early. A pile that took its toe elevation in half the expected blows has probably not found the same stratum as its neighbour, and on a site with 280 mm of settlement queued up in the clay, a pile founded short is a pile whose neutral plane and drag length are not the ones anybody calculated. Cross-checking the driving log against the borehole log, pile by pile, is the last chance to find that before a cap is poured over it.

One more habit worth keeping. Record the date the fill was placed, the date each pile was driven, and the settlement plate reading on both. Downdrag is the one foundation problem where the calendar is a design input, and three years later, when somebody asks why the east end of the building has moved more than the west, that table is the only document that will answer.