CPT Profile Plot — Cone Penetration Test Logs with Robertson SBT

How to read the profile below

Depth increases downward, in metres below ground level, and every track shares that one axis — which is the whole point of the layout. Read across: a unit that is soft in qc, high in Rf and running well above hydrostatic in u2 is a clay, and all three statements have to agree before you believe any of them.

qc is strength, near enough. Soft clays sit under 1 MPa, stiff clays a few MPa, medium dense sands 5–15 MPa, dense sands 15–40 MPa. fs on its own is not very informative; it earns its place through Rf = fs / qc × 100 %, the friction ratio, which is roughly a property of the soil type rather than of its density. Clays run around 3–6 %, silty mixtures 1.5–3 %, clean sands under 1 %. That single ratio is what makes a classification from a cone possible at all.

u2 is the one non-obvious track and often the most useful. Below the water table a freely draining soil reads close to hydrostatic — the dashed reference line on the track — because the cone loads it no faster than it can drain. A fine-grained soil cannot drain at 20 mm/s, so pore pressure builds and the trace peels away from the line. The gap between the two IS the excess pore pressure, and its shape names the material more reliably than qc does in soft ground. A trace that dips below hydrostatic is a dense sand dilating as the cone shears it, which is worth knowing before you design a dewatering scheme around it.

The soil behaviour type strip, and what it is not

The strip beside the depth scale classifies every reading on the Robertson et al. (1986) chart — twelve numbered zones on a log-log plot of qc against Rf, running from sensitive fine grained material at zone 1 through clays, silt mixtures and sands to gravelly sand at zone 10, with two more zones for material whose cone resistance is far higher than its friction ratio suggests, i.e. overconsolidated or cemented. The zones are drawn as an ordered grey ramp with the zone number printed on every interpreted layer, and the ramp runs soft-dark to strong-light, so the strip is readable as a profile in a photocopy and in a black-and-white print.

It is a soil behaviour type chart, and Robertson has spent thirty years saying so. It reports how the ground responded to a cone, which is usually what a sieve and a set of Atterberg limits would have said, and sometimes is not — a very stiff clay and a loose cemented sand can behave alike against a cone. Treat the strip as the strongest available inference between sample points, not as a grading. Where it matters, the industry answer has not changed: put a hole down beside the sounding and sample it.

One more honesty note, and it is on the figure as well as here. The published chart is a printed figure, not a table of coordinates, so every software implementation of it — this one included — is somebody's digitisation of a curve. Readings that plot well inside a zone are classified reliably; readings sitting within a line-width of a boundary may fall either side of it here and in the original. That is exactly what the interpreted layer table is for: it absorbs runs thinner than the threshold you set, so a boundary that flickers between zones 6 and 7 forty times in a metre resolves into one stratum rather than forty.

Who this is for

Geotechnical engineers interpreting site investigation data, whether the cone was pushed by their own contractor or the file arrived with a tender pack. Engineering geologists correlating a CPT against the boreholes either side of it. Contractors and temporary-works designers who need one clear sheet to argue a founding level or a dewatering requirement from. And anybody who has been handed a `.csv` with three columns and six hundred rows and needs a figure by Thursday.

It is not an interpretation suite and does not pretend to be one. There is no liquefaction assessment, no pile capacity calculation, no CPT-to-SPT correlation, no multi-sounding fence diagram. What it does is the part everybody needs first: take the readings the cone actually made and produce one correct, printable, properly cited profile sheet, with no installation, no licence server and no upload.

Honest limits

SI units only — qc in MPa, fs and u2 in kPa, depth in metres. Feet and tsf are not supported and will not be faked; a sheet labelled in metres that carries feet is a genuinely dangerous document. One sounding per figure. The optional Ic track is the Robertson (1990) normalised index, and it needs vertical stress, which a cone does not measure: it is computed from a single assumed bulk unit weight and the water table you type in, both printed on the figure. That is the standard field approximation, and it is precisely why Ic is an optional extra track here rather than the thing the strip column is built on.

Everything runs in this browser tab. No sounding is uploaded anywhere, which for pre-acquisition site data is usually the whole question.

Common questions

The page says Pro — what can I actually do for free?

Everything except enter your own data. The profile below is the finished tool running live on a real 601-reading sounding: change the vertical scale, pin the qc axis, turn tracks on and off, add the Ic track, edit the title block, and export the figure as SVG or PNG (watermarked, like every free export on this site). What Pro buys is the right to replace the sample readings with yours. There is no trial timer and no account, and the demo does not expire.

What units does it expect?

The ones every rig in the world exports, mismatched as they are: depth in metres, qc in MPa, fs in kPa, u2 in kPa. The mismatch is real and deliberate — a cone resistance in kPa would be a five-digit number on every row. Both columns are bounds-checked, so a qc column pasted in kPa is reported on the figure rather than silently plotted as a chart of nothing.

How many readings can it take?

Fifteen hundred comfortably, which is a 15 m sounding at 10 mm intervals. The traces are decimated for drawing — reduced to about two points per vertical pixel — but by a minimum/maximum scheme rather than by keeping every nth reading, so every local peak and trough survives exactly. That distinction matters: a gravel band is one or two readings wide, and a profile that quietly loses its spikes has lost the thing you were looking for. The figure states how far it decimated, and you can turn it off.

Which soil behaviour type chart is it?

Robertson, Campanella, Gillespie & Greig (1986), the twelve-zone non-normalised chart of qc against friction ratio — the one that needs nothing but the two numbers the cone measured. It is cited in the figure footer. The normalised Robertson (1990) Ic index is available as an optional extra track, with its own citation and its assumptions stated, but it is not what the strip column is built on, because Ic needs an assumed unit weight and the 1986 chart does not.

Why does the strip flicker between two zones inside one layer?

Because the ground does not know where the boundary is. A soil sitting near a zone edge crosses it and comes back as the readings vary, which is a true picture of the classification and a poor picture of the stratigraphy. That is what the interpreted layer table underneath is for: contiguous runs thinner than the threshold you set (0.3 m by default) are absorbed into the thicker neighbour, so the table reads as strata while the strip above it still shows you exactly how much interpretation was applied.

Can it do liquefaction assessment or pile capacity?

No, and it will not pretend to. Those are design calculations with method choices, partial factors and code-specific rules behind them, and burying that inside a plotting tool would be the wrong place for it. What this gives you is the correctly drawn, correctly cited profile those calculations are argued from.