Geotechnical Site Investigation and Ground Models

Introduction and context

A foundation is designed for the ground that actually exists, not for the ground assumed from a regional map or a single borehole. Geotechnical site investigation is therefore a process of reducing uncertainty sufficiently to support safe, economical and constructible decisions. It combines desk study, site reconnaissance, intrusive investigation, in-situ testing, laboratory testing, groundwater observations and professional interpretation.

The principal product is not a collection of borehole logs. It is a defensible ground model: a three-dimensional interpretation of geological materials, groundwater conditions, discontinuities, variability and engineering behaviour. The model is then converted into design parameters and communicated through a geotechnical report.

Investigation is necessarily iterative. Early information identifies hazards and determines where later investigation should concentrate. Construction observations then provide an opportunity to confirm or revise the model.

Learning objectives

By the end of this chapter, a student should be able to:

1. Plan a proportionate desk study, walkover survey and ground investigation.
2. Select suitable boreholes, trial pits, samples and groundwater monitoring arrangements.
3. Explain the principles and limitations of SPT, CPT/CPTu, vane, pressuremeter and geophysical methods.
4. Interpret investigation data without confusing measured values with design parameters.
5. Develop a geological and geotechnical ground model and select characteristic parameters.
6. Prepare a clear report that records uncertainty, assumptions, limitations and recommendations.
7. Recognise safety, environmental, ethical and regulatory responsibilities.

1. Investigation as a decision process

The scope of an investigation should be related to the project’s geotechnical category, scale, structural sensitivity, ground conditions and consequences of failure. A small low-rise building on uniform natural sand may require less investigation than a deep basement beside historic buildings, even if their footprints are similar.

Investigation questions should be stated before selecting methods. Typical questions include:

• What materials and geological boundaries occur beneath the site?
• Are there made ground, buried structures, voids, contamination or unexploded ordnance?
• What are the strength, stiffness, compressibility and permeability characteristics?
• Where is groundwater, and how does it vary seasonally or in response to construction?
• Are there hazards such as landslides, swelling clay, collapsible soil, dissolution or seismic liquefaction?
• Which construction methods are feasible and what effects may they have on neighbours?

A useful investigation balances three risks: under-investigation, which creates surprises; over-investigation, which wastes resources without improving decisions; and misinterpretation, which can make high-quality data unsafe to use.

2. Desk study and preliminary ground model

A desk study assembles existing evidence before fieldwork. Sources commonly include geological and soil maps, topographic maps, aerial photographs, historical maps, previous site investigation reports, borehole databases, mining records, utility plans, planning records, flood and groundwater information, satellite data and records of nearby construction.

Historical land use is especially important. Former factories, gasworks, tanneries, landfills, quarries, railway sidings and infilled ponds may produce unusual materials, contamination or voids. Changes in drainage, river courses and earthworks can explain apparent ground anomalies.

The preliminary ground model should identify:

• expected strata and their geological origin;
• likely lateral and vertical variability;
• groundwater bodies, flow directions and possible perched water;
• anthropogenic deposits and buried obstructions;
• geohazards and environmental constraints;
• confidence in each interpretation; and
• questions that fieldwork must resolve.

A desk study does not replace investigation. Maps generalise geology, historical records may be incomplete, and nearby boreholes may not represent the site.

3. Walkover survey

A walkover tests the desk-study model against physical evidence. The investigator records topography, slopes, drainage, springs, seepage, standing water, vegetation changes, cracks, settlement, retaining walls, exposed soil, outcrops, fill boundaries, former structures and access constraints. Adjacent buildings can provide evidence of foundation type, movement or groundwater effects.

Observations should be located on a plan and supported by photographs with scale and direction. The survey should also consider safe access, overhead and buried services, traffic, water bodies, contaminated materials, confined spaces and the need for permits. A walkover is interpretive: the absence of visible distress does not prove that the ground is uniform or competent.

4. Designing the intrusive investigation

Boreholes

Boreholes provide continuous or discontinuous subsurface information to depths selected according to the proposed foundations, excavation, piles, slope geometry and potential failure mechanisms. Locations should cover the building footprint, major load lines, retaining walls, changes in elevation and suspected geological boundaries. Staggered locations often provide better three-dimensional coverage than a simple grid.

Borehole methods include cable percussion, rotary drilling, augering, sonic drilling and specialised systems for rock or contaminated ground. The method affects sample quality, disturbance, groundwater observations and the ability to penetrate obstructions. Boreholes must be accurately surveyed, logged and referenced to a consistent datum.

Trial pits and trenches

Trial pits expose soil directly and permit large disturbed or relatively undisturbed samples. They are valuable for shallow foundations, pavement subgrades, buried obstructions and made ground. Their limitations are shallow reach, difficult access, collapse risk, groundwater inflow and the need for safe battering or shoring. A person must never enter an unsupported excavation.

Sampling

Samples should be taken to answer specific questions. Disturbed samples may be suitable for classification, moisture content and some chemical tests. High-quality or relatively undisturbed samples are needed for reliable strength, compressibility and permeability testing, but even these are affected by stress relief, extrusion, handling and storage.

Sample quality should be described rather than assumed. Thin-walled samplers, piston samplers and block samples generally reduce disturbance in suitable cohesive soils. In gravel, fissured clay, very soft soil or weak rock, recovery may be poor or misleading. Chain-of-custody procedures and preservation are essential for environmental and groundwater samples.

Groundwater monitoring

Groundwater readings should be treated as time-dependent observations, not as a single permanent level. Standpipes, piezometers and vibrating-wire instruments serve different purposes. A standpipe may indicate hydraulic head after equilibration; a piezometer can measure a selected stratum; an automated sensor can capture short-term variation.

A hydraulic head is commonly expressed as:

where h is hydraulic head, z is elevation of the measurement point, u is pore-water pressure, and γ_w is the unit weight of water. The equation assumes consistent elevation datum and compatible units. In layered ground, water levels in different screened zones must not be combined as though they represent one aquifer.

Monitoring should continue long enough to identify seasonal, tidal, pumping and rainfall effects where relevant. Open boreholes can allow water from different strata to mix, so installations must be designed and sealed appropriately.

5. In-situ testing and interpretation

No test directly measures every design property. Each test responds to a volume of ground and depends on drainage, stress history, anisotropy, equipment and operator practice. Correlations are empirical and should be checked against soil type and local experience.

Standard Penetration Test

The Standard Penetration Test (SPT) drives a split-spoon sampler using a standard hammer arrangement. The reported N value is normally the number of blows for the final 300 mm of penetration after an initial seating drive; the first 150 mm is commonly excluded from the penetration resistance.

Raw N values vary with hammer energy, rod length, borehole diameter, sampler configuration and overburden stress. A corrected value is often written:

where E_R is delivered hammer energy as a percentage, and C_B, C_R and C_S are corrections for borehole diameter, rod length and sampler configuration. Conventions differ, so the adopted standard and correction procedure must be reported. An overburden-normalised value may be denoted (N₁)₆₀, but the normalisation is not universally applicable and should not be used blindly in cemented, gravelly or unusual soils.

SPT is useful for indicative relative density, consistency and liquefaction screening, but it has large scatter. Refusal may indicate dense soil, gravel, cobbles, obstruction or equipment effects; it is not automatically proof of rock.

Cone Penetration Test and CPTu

A cone penetrometer is pushed continuously into the ground while measuring cone tip resistance q_c, sleeve friction f_s and, for CPTu, pore-water pressure u₂. The friction ratio is commonly defined as:

where q_c and f_s must use the same stress units. CPT produces a high-resolution profile and is excellent for identifying interfaces, loose layers and spatial variability. It does not normally recover physical samples, so selective boreholes or sampling are needed for confirmation.

CPTu interpretation depends on pore-pressure response and drainage. In soft clay, undrained behaviour may dominate during penetration; in clean sands, drainage is often rapid. Corrected cone resistance may be required when pore pressure acting on unequal cone areas is significant. Equipment calibration, saturation of the porous filter and zero readings are critical.

Field vane test

The field vane measures the torque required to rotate a cruciform vane in soft cohesive soil. An idealised undrained shear strength is often calculated from:

where T is corrected torque, D is vane diameter, H is vane height, and the expression assumes uniform undrained shear strength on the cylindrical and end surfaces. For a vane with H = D, this reduces to the familiar special-case form. The measured value should be corrected for rod friction and, where appropriate, soil anisotropy and strain-rate effects. Remoulded testing can indicate sensitivity. The method is unsuitable in gravelly, fissured or strongly layered soil and may be affected by progressive failure.

Pressuremeter

A pressuremeter expands a cylindrical probe in a borehole and records pressure against radial strain or volume change. It provides information about in-situ stiffness, pressuremeter strength and, with suitable interpretation, foundation and lateral response. Results are highly dependent on borehole quality, probe insertion, membrane behaviour and the interpretation model. A carefully formed borehole is essential; drilling disturbance can dominate the result.

Geophysical methods

Seismic refraction, multichannel analysis of surface waves, electrical resistivity, electromagnetic methods, ground-penetrating radar and downhole methods can rapidly map contrasts and support interpolation between intrusive points. For example, shear-wave velocity V_s is related to small-strain shear modulus by:

where ρ is mass density. The relation is valid for small-amplitude, approximately elastic shear waves. G_max is a small-strain stiffness, not necessarily the stiffness mobilised by a foundation at working strain.

Geophysical data are indirect and non-unique: different materials may produce similar signals. They should be calibrated with boreholes, trial pits or laboratory data.

6. From data to a ground model

A ground model integrates geology, geomorphology, investigation observations and engineering behaviour. It should distinguish three related but different descriptions:

1. Geological model: origin, age, depositional environment and structure.
2. Geotechnical model: engineering units with ranges of strength, stiffness, compressibility and permeability.
3. Groundwater model: aquifers, aquitards, perched water, heads, flow paths and seasonal variation.

An engineering unit is not necessarily a geological formation. A single formation may contain several units with different behaviour, while materials from different origins may behave similarly for a particular design problem.

Logs should record material description systematically: colour, texture, grading, plasticity, consistency, structure, weathering, discontinuities, recovery and groundwater evidence. Boundaries should be classified as observed, inferred or uncertain. Cross-sections should show both data points and interpreted boundaries, avoiding false precision.

A ground model is strengthened by conceptual redundancy: important conclusions should be supported by more than one line of evidence where practicable. For example, a suspected loose sand layer might be investigated through CPT, selected SPTs, groundwater observations and laboratory grading.

7. Characteristic values and uncertainty

Measured values are samples from a variable ground mass. A design should use a characteristic value that represents the relevant ground volume and failure mechanism, not simply the average or the lowest isolated result.

For a dataset with observations x_i, the sample mean is:

and the sample standard deviation is:

Here n is the number of observations. These statistics describe the sample only; they do not automatically describe the full stratum. Outliers must be investigated geologically and procedurally before deletion.

A lower fractile may be used for a parameter where low values are adverse, but the selected fractile, confidence, spatial averaging and consequence of failure must follow the applicable design standard and project basis. A simple statistical lower-bound expression such as x̄ − ks is not universally valid: k depends on the statistical model, sample size and required confidence. Engineering judgement remains necessary, particularly with sparse or spatially correlated data.

Characteristic stiffness, strength and permeability must be selected for the mechanism. A shallow footing may mobilise an averaged response over a broad zone, whereas a pile shaft, tunnel face or slope slip surface may be controlled by a narrow weak layer. Parameter selection should document data sources, transformations, correlations, scale effects, drainage assumptions and uncertainty.

8. Reporting and communication

A professional geotechnical report normally contains:

• project brief, scope and limitations;
• site description and proposed development;
• desk-study sources and walkover findings;
• investigation methods, locations, elevations and dates;
• borehole, trial-pit, sampling and laboratory records;
• groundwater observations and monitoring interpretation;
• geological and geotechnical ground model;
• design parameters with rationale and ranges;
• geohazards and environmental findings;
• foundation, excavation, earthworks and temporary-works implications;
• construction verification and monitoring requirements; and
• residual risks, assumptions and recommendations for further work.

Logs should separate observation from interpretation. A statement such as “grey silty clay observed from 4.2 to 6.1 m” is different from “firm London Clay.” Photographs, plans, sections and tables should be traceable to original records. Digital data should retain metadata, coordinate systems, units, calibration information and version history.

The report should state what was not investigated. Confidence can be communicated using qualitative classes or a risk register, provided the definitions are explicit. Uncertainty is not a weakness to hide; it is a design input to manage.

9. Illustrative case study: a proposed basement beside historic buildings

Background

A developer proposes a two-storey basement beneath a constrained urban site. Adjacent nineteenth-century masonry buildings have shallow foundations. Historical maps show a former industrial yard and an infilled channel nearby.

Problem

The design team must determine whether excavation will encounter variable fill, contaminated soil, groundwater or soft deposits, and whether ground movement could damage neighbouring buildings.

Analysis

The desk study identifies possible made ground, buried foundations and a shallow aquifer. The walkover records dampness at the rear boundary and cracks in a neighbouring wall, although the cause is uncertain. A preliminary model proposes fill over alluvium, underlain by dense sand and weathered rock.

A combination of methods is selected: boreholes inside and outside the proposed basement, trial pits near the boundary to expose neighbouring foundation levels, CPTs to map the alluvial layer, groundwater standpipes in separate strata, and laboratory chemical and geotechnical testing. Vibration and access constraints require low-impact equipment near the occupied buildings.

Approach and solution

The investigation finds highly variable fill, including brick fragments and old concrete, over soft silty alluvium. CPTu identifies a continuous soft layer that was missed by one borehole. Groundwater rises after prolonged rainfall and is hydraulically connected to the alluvium. Trial pits show that one neighbouring wall foundation is shallower than expected.

The ground model therefore identifies excavation-induced deformation and groundwater drawdown as controlling risks. The design adopts a stiff, low-permeability retaining system, staged excavation, internal propping, controlled dewatering or a cut-off strategy, and preconstruction condition surveys. Inclinometers, piezometers, settlement points and crack gauges are specified with trigger levels and response actions.

Outcome

The revised investigation costs more initially but avoids relying on an optimistic soil profile. Construction proceeds with observational monitoring. Movements remain within the agreed action levels, and the soft layer is locally treated in an area where the retaining response requires improvement.

Learning points

The case demonstrates that investigation locations should be driven by failure mechanisms, not merely by a regular grid. CPT can reveal continuity between boreholes; trial pits can resolve foundation interaction; groundwater must be monitored over time; and the report must link findings to construction controls.

10. Safety, ethics and contemporary practice

Investigation work involves drilling rigs, lifting operations, traffic, unstable excavations, contaminated ground, buried services, noise and possible confined spaces. A method statement, utility search, permit system, exclusion zone, competent supervision and emergency plan are essential. Sampling personnel must use appropriate chemical and biological protection.

Ethically, investigators must report adverse evidence, preserve original data and avoid selecting only favourable results. Commercial pressure must not reduce investigation below a level necessary for public safety. Local regulations, occupational safety law, environmental controls, drilling permits, waste classification and professional codes must be followed.

Digital platforms now support geographic information systems, three-dimensional geological modelling, automated logging, cloud data capture, remote sensors and machine-learning classification. AI can help identify patterns or prioritise anomalies, but it cannot replace geological reasoning or responsibility. Training data may be biased, correlations may fail outside their calibration domain, and every automated interpretation should remain auditable and checked against primary observations.

Best practice is therefore to combine appropriate investigation density, independent checks, calibrated instruments, transparent assumptions, staged decision-making and construction feedback. The most valuable investigation is not the one with the most tests, but the one that resolves the uncertainties that control risk.

Chapter Summary

Geotechnical site investigation is an evidence-based process that develops and updates a ground model for design and construction. Desk studies and walkovers establish context and hazards; boreholes, trial pits, sampling and monitoring provide direct evidence; and in-situ and geophysical tests provide measurements at different scales. SPT, CPTu, vane, pressuremeter and geophysical results require method-specific interpretation and do not automatically equal design parameters. Ground models must distinguish observed facts from inferred boundaries, represent groundwater and variability, and identify uncertainty. Characteristic parameters are selected for the relevant mechanism and ground volume, using statistics cautiously and engineering judgement transparently. A professional report links evidence to design, construction, monitoring and residual risk while meeting safety, ethical and regulatory obligations.

Key Terms

Ground model: An interpreted three-dimensional description of geology, engineering units, groundwater and uncertainty.

Desk study: Review of existing geological, historical, environmental and project information.

Walkover survey: Field reconnaissance used to test and extend the desk-study interpretation.

Made ground: Anthropogenic material placed, disturbed or reworked by human activity.

Borehole: A drilled exploratory hole used for logging, sampling, testing or monitoring.

Trial pit: An excavated exposure used to inspect shallow ground and foundations.

Disturbed sample: A sample whose natural structure or stress state has been substantially altered.

CPT: Cone Penetration Test measuring resistance during continuous cone insertion.

CPTu: CPT incorporating pore-water pressure measurement.

SPT: Standard Penetration Test using a driven sampler and recorded penetration resistance.

Field vane: In-situ test for undrained strength of suitable soft cohesive soil.

Pressuremeter: Borehole test that expands a probe and measures radial ground response.

Piezometer: Instrument used to measure pore-water pressure or hydraulic head.

Characteristic value: Selected parameter representing relevant ground behaviour for design.

Geotechnical unit: Material grouping defined by engineering behaviour for a particular purpose.

Liquefaction: Loss of effective stress and strength in susceptible saturated soil during cyclic loading.

Fractile: A value below or above which a specified proportion of a population lies.

Uncertainty: Imperfect knowledge of ground conditions, measurements, models or future states.

Observational method: Design and construction approach using predefined monitoring and response actions.

Review Questions

Short Answer Questions

1. Why is a ground model more useful than a collection of borehole logs?
2. State two limitations of trial pits.
3. What are the principal measurements made by CPTu?
4. Why should groundwater readings be monitored over time?
5. Distinguish a geological unit from a geotechnical unit.

Long Answer Questions

1. Describe a systematic process for planning an investigation from project brief to final report.
2. Compare SPT and CPTu in terms of data resolution, sampling, interpretation and limitations.
3. Explain how groundwater conditions should be investigated and represented in a ground model.
4. Discuss the principles governing selection of characteristic strength and stiffness parameters.
5. Explain how safety, ethics and uncertainty should influence professional geotechnical reporting.

Analytical / Application Questions

1. A proposed deep excavation is beside old masonry buildings on a site with suspected fill and alluvium. Design an investigation strategy and justify the location and purpose of each method.
2. A CPT profile indicates a continuous soft layer, but two nearby boreholes do not record it. Explain possible reasons and propose checks before changing the ground model.
3. A project team proposes using the average of all SPT values as the design soil resistance. Critically evaluate this proposal and recommend a defensible alternative.

Multiple Choice Questions with Answers

Question 1

Which activity primarily establishes the initial geological and hazard hypotheses before intrusive work?

A. Desk study

B. Pressuremeter testing

C. Laboratory consolidation testing

D. Pile load testing

Correct Answer: A. Desk study

Question 2

In the usual SPT procedure, the reported N-value commonly counts blows for which penetration?

A. First 150 mm only

B. Final 300 mm after seating

C. Entire first metre

D. Final 50 mm only

Correct Answer: B. Final 300 mm after seating

Question 3

Which CPTu measurement is especially useful for interpreting drainage response and soil behaviour?

A. Borehole diameter

B. Pore-water pressure

C. Sample recovery percentage

D. Hammer energy

Correct Answer: B. Pore-water pressure

Question 4

The friction ratio R_f is calculated from sleeve friction divided by:

A. Pore-water pressure

B. Unit weight of water

C. Cone tip resistance

D. Groundwater head

Correct Answer: C. Cone tip resistance

Question 5

The field vane test is most appropriate for:

A. Clean gravel

B. Soft cohesive soil

C. Massive intact granite

D. Highly cemented concrete fill

Correct Answer: B. Soft cohesive soil

Question 6

In the equation G_max = ρV_s², G_max represents:

A. Ultimate bearing capacity

B. Small-strain shear modulus

C. Hydraulic gradient

D. Plasticity index

Correct Answer: B. Small-strain shear modulus

Question 7

Why should CPT results commonly be supplemented by boreholes or sampling?

A. CPT cannot provide continuous data

B. CPT normally provides no physical soil samples

C. CPT is only usable in rock

D. CPT directly measures every design parameter

Correct Answer: B. CPT normally provides no physical soil samples

Question 8

Which statement about a characteristic parameter is most defensible?

A. It is always the arithmetic mean

B. It is always the single lowest measured value

C. It represents the relevant ground volume and mechanism

D. It can be selected without considering variability

Correct Answer: C. It represents the relevant ground volume and mechanism

Question 9

A piezometer is primarily used to measure:

A. Pore-water pressure or hydraulic head

B. Hammer energy

C. Cone sleeve friction

D. Rock joint spacing only

Correct Answer: A. Pore-water pressure or hydraulic head

Question 10

What is a major limitation of geophysical investigation?

A. It can never cover large areas

B. Its results are indirect and may be non-unique

C. It always requires destructive excavation

D. It cannot be calibrated with boreholes

Correct Answer: B. Its results are indirect and may be non-unique