Module 8 · Lesson 8.2
CPT and plate tests
Continuous data, and a test at the wrong scale.
Why this matters
The SPT gives a number every 1.5 m from a test that hammers a tube into the ground. The CPT gives a reading every 20 mm from a controlled, repeatable push. Where both are available the CPT is the better instrument — but it still hands you one judgement that decides the answer.
For clays, undrained shear strength comes from the net cone resistance divided by a cone factor:
su = (qt − σv0) / Nkt
Everything on the right except Nkt is measured. Nkt is chosen. It commonly falls between about 10 and 20, it depends on the clay and on how the correlation was calibrated, and the answer is directly proportional to it.
A CPT giving qt = 1.8 MPa at a depth where σv0 = 150 kPa yields su = 110 kPa with Nkt = 15. Choose 10 instead and it becomes 165 kPa; choose 20 and it becomes 83 kPa. Same data, same soil, a factor of two between the extremes — decided entirely by a number the engineer supplied.
Plate loading tests measure the ground's response directly, which is appealing — no correlation, no cone factor, just load and settlement. The difficulty is scale.
A 300 mm plate stresses roughly 300–600 mm of ground. A 3 m footing stresses 4.5 to 6 m. They are interrogating different soil, and on a layered site they can give entirely different answers: the plate may sit happily on a firm crust that the real footing will punch straight through.
A plate test tells you about the ground near the surface. It becomes a foundation prediction only with a scaling rule, and the scaling rules are themselves empirical.
Check yourself
A 300 mm plate test on a site with 1 m of firm crust over soft clay gives an excellent result. What should be concluded about a proposed 3 m wide footing?
Practice
A CPT gives a corrected tip resistance of 1.8 MPa at a depth where the total vertical stress is 150 kPa. Using Nkt = 15, what is the undrained shear strength, in kPa?
Try it
From a blow count to a bearing factor
Four corrections, one correlation and an exponential. The band at the end is the honest output.
What each correction removes
- N₆₀ removes the rig — the energy the hammer actually delivered.
- (N₁)₆₀ removes the depth, so a uniform sand does not appear to stiffen with depth.
- φ′ from (N₁)₆₀ is a fitted correlation carrying roughly ±3°.
- 1MeasuredN = 18
- 2Energy correctedN₆₀ = 22.5
- 3Effective stress at depthσ′v = 57.0 kPa
- 4Overburden corrected(N₁)₆₀ = 29.8
- 5Density statemedium dense
- 6Friction angleφ′ = 41.4°
Nq across the ±3.0° band
- Nq at φ′ − band
- 51.8
- Nq at φ′
- 78.5
- Nq at φ′ + band
- 123.1
- Ratio across the band
- 2.38×
The same blow count supports Nq anywhere from 51.8 to 123.1 — a factor of 2.38. A design that only works at the top of that range is not a design.
Summary
- N₆₀ removes the rig; (N₁)₆₀ removes the depth
- φ′ from SPT carries about ±3°, which moved Nq by a factor of 2.4 in the worked example
- su from CPT is proportional to Nkt, a number chosen rather than measured
- Plate tests measure the ground near the surface, not the ground a footing will load
- With correlated parameters, a range is a more honest output than a single value
This is educational material. It uses simplified examples to teach principles, and must not be relied on for real design or safety-critical decisions. Module overview and checkpoint