Every number in every other chapter — c', φ', γ, Cc, cv — came from somewhere. This chapter is that somewhere. It is the only part of geotechnical engineering where you are allowed to find out what the ground actually is, instead of assuming.
A site investigation typically costs well under 1% of project value. Foundation problems — discovered after construction starts — routinely cost many times that, and they are the single most common source of construction claims and delays.
The purpose of subsoil exploration is to determine the stratigraphy (what layers, how thick), the engineering properties of each layer, and the groundwater regime — with enough coverage to be confident the ground between boreholes behaves like the ground in them.
Desk study and site walkover: geological maps, aerial photographs, records of nearby construction, groundwater data, and evidence you can see — cracked adjacent buildings, seepage, old fill, slope movement.
A small number of boreholes to establish the broad picture: how many layers, roughly how thick, where the water table is. Enough to plan the detailed programme intelligently.
Full borehole grid with in-situ testing and undisturbed sampling for laboratory work, targeted at the layers that will actually govern the design.
Bore deep enough that the load no longer matters. Two common rules, take the shallower:
Below that depth the added stress is a rounding error, so deeper drilling buys nothing.
S = number of storeys; C = 3 m for light steel / narrow concrete buildings, 6 m for heavy steel / wide concrete. A quick first estimate before you have any stress calculation.
Simple and cheap, for shallow depths in cohesive soil above the water table. Hollow-stem augers are the workhorse: the hollow centre keeps the hole open so you can sample straight down the middle without casing.
Water pumped down a drill rod jets soil loose and carries cuttings up. Cheap and can go deep, but the returned material is thoroughly disturbed — useful for advancing the hole, not for identifying soil precisely.
The method of choice for deep holes and for rock coring. Drilling mud stabilises the borehole wall, which matters below the water table and in granular soil.
For hard ground, cobbles, boulders and rock where other methods stall. Very disturbing to the soil — sampling quality is poor.
Every sample is disturbed to some degree; the question is how much, and whether what you plan to measure survives it.
Structure is destroyed, but grain sizes and moisture survive. Fine for classification, Atterberg limits, grain-size distribution and compaction tests. Useless for strength or compressibility, which depend on the fabric you just wrecked.
A thin steel tube pushed (never driven) into clay, preserving structure well enough for triaxial and consolidation testing. This is where Cc, cv and cu come from — the numbers driving the settlement chapter.
The tube has to shove aside a volume of soil equal to its own wall. A thick wall displaces more soil and disturbs more of what it captures. A Shelby tube achieves roughly 10%; a split-spoon sampler is above 100%, which is exactly why it cannot give undisturbed samples.
Drive a standard split-spoon sampler into the base of the borehole with a 63.5 kg hammer falling 760 mm. Count blows for three 150 mm increments; discard the first (disturbed by drilling) and N = the sum of the second and third. It is crude, it is the most widely used in-situ test in the world, and almost every empirical correlation in foundation engineering is built on it.
Different rigs deliver wildly different fractions of the theoretical hammer energy — a donut hammer might deliver 45%, a modern automatic trip hammer 80%. Results are normalised to 60% efficiency.
ηH hammer efficiency, ηB borehole diameter, ηS sampler, ηR rod length — all as percentages.
The same sand gives a higher blow count when it is deeper, simply because it is more confined — not because it is denser. To compare density between depths, normalise to a reference stress of 100 kPa.
Liao & Whitman. Apply only in granular soil.
Once you have (N1)60, correlations give you design parameters: relative density \( D_r \approx \sqrt{(N_1)_{60}/60} \), friction angle \( \phi' \approx 27.1 + 0.3(N_1)_{60} - 0.00054(N_1)_{60}^2 \), and for clay a rough \( c_u \approx 6N_{60} \) kPa — though for clay you should be sampling and testing, not correlating.
Take a field blow count all the way through to design parameters, one correction at a time.
A 60° cone of 10 cm² area is pushed in at 20 mm/s while load cells continuously record cone resistance qc and sleeve friction fs. The friction ratio \( F_r = f_s/q_c \times 100\% \) identifies soil type: low ratio means sand, high ratio means clay.
Advantages: continuous profile rather than a reading every 1.5 m, repeatable, no operator effect. Disadvantage: brings up no sample, and cannot penetrate gravel.
A four-bladed vane is pushed into soft clay and rotated until the soil shears on a cylindrical surface. The peak torque gives undrained shear strength directly, with no sampling disturbance at all.
Apply Bjerrum's correction factor λ (which falls with plasticity index) before design use — the raw vane value is generally unconservative.
The borehole log is the deliverable that every later chapter consumes. It must record depth and description of each stratum, sample type and depth, SPT N values, groundwater level (both on encountering it and after standing), and the drilling method.
Record what you observed, not what you concluded. Interpretation belongs in the report; the log is evidence, and it will be read years later by people who were not on site.
Problem: A field SPT in sand at 6 m depth gives N = 18. A safety hammer with rope and pulley was used (ηH = 60%), borehole 150 mm (ηB = 105%), rods 5 m (ηR = 85%), standard sampler (ηS = 100%). Effective overburden at that depth is 95 kPa. Find N60, (N1)60, φ' and Dr.
Problem: A 6-storey reinforced concrete building with a relatively wide footprint is planned. Estimate the required borehole depth using Sowers' rule.
Problem: A thin-walled tube has outside diameter 76.2 mm and inside diameter 73.0 mm. Compute the area ratio and state whether it is suitable for consolidation testing.
1. Why is the blow count from the first 150 mm of SPT penetration discarded?
2. Two identical sands are tested, one at 3 m and one at 15 m. The deeper gives a much higher N. What does the overburden correction do?
3. You need Cc and cv for a settlement analysis. Which sample will do?
4. Drilling meets refusal at 8 m. Is it safe to conclude bedrock has been reached?