Subgrade Investigation and Geotechnical Design Inputs

Introduction and Context

The subgrade is the engineered soil foundation that supports the pavement structure. Its condition influences pavement thickness, construction quality, drainage performance, frost behaviour, permanent deformation and fatigue life. A pavement designed using an unrealistic subgrade strength may be either uneconomically thick or prematurely damaged.

Subgrade design inputs are not single laboratory numbers. They are interpreted estimates of the material state expected beneath the completed pavement, including density, moisture, stress history, drainage, seasonal variation and spatial variability. Postgraduate practice therefore requires integration of field investigation, laboratory testing, statistical analysis and mechanistic interpretation.

Learning Objectives

On completing this chapter, the reader should be able to:

1. classify subgrade soils using engineering classification systems;

2. interpret compaction and moisture–density relationships;

3. select and assess CBR and resilient modulus inputs;

4. evaluate frost susceptibility and expansive-soil risk quantitatively;

5. develop preliminary stabilization mix designs and verification plans;

6. design and verify geosynthetic functions relevant to subgrades; and

7. account for spatial variability and reliability in pavement design.

1. Investigation Framework

A defensible investigation proceeds from reconnaissance to progressively more detailed testing:

• Desk study: geology, groundwater, topography, climate, historical pavement performance, borrow sources and construction records.
• Reconnaissance: drainage, cut and fill transitions, soft spots, cracking, rutting, vegetation and evidence of seasonal water levels.
• Exploration: trial pits, auger holes, boreholes, dynamic probing or cone penetration testing, supplemented by groundwater observations.
• Sampling: disturbed samples for classification and compaction; undisturbed or carefully preserved samples for strength, compressibility and swelling tests.
• Laboratory testing: particle-size distribution, Atterberg limits, compaction, CBR, resilient modulus, swelling, frost and chemical compatibility tests.
• Data integration: zoning, outlier review, characteristic-value selection and design verification during construction.

Investigation spacing should reflect geological variability rather than a fixed geometric rule. Additional points are warranted at soil boundaries, drainage low points, embankment transitions, cut–fill interfaces and locations of anomalous test results.

2. Soil Classification and Engineering Interpretation

Particle-size distribution distinguishes gravel, sand, silt and clay fractions. Atterberg limits describe the water contents separating consistency states. The plasticity index is

\[

PI=w_L-w_P

\]

where \(w_L\) is liquid limit and \(w_P\) is plastic limit, both expressed as percentages. A high PI generally indicates greater plasticity and potential volume change, but mineralogy and structure must also be considered.

The Unified Soil Classification System (USCS) uses grain size, gradation and plasticity. AASHTO classification additionally relates soil groups to highway subgrade quality. Classification is not a strength test: two soils in the same group may have substantially different CBR or resilient modulus because of density, moisture, fabric and stress history.

Useful interpretive indicators include:

• coefficient of uniformity \(C_u=D_{60}/D_{10}\);
• coefficient of curvature \(C_c=D_{30}^2/(D_{10}D_{60})\);
• percentage passing 0.075 mm; and
• plasticity and linear shrinkage for fine soils.

These indicators should be combined with mineralogical, drainage and field-condition evidence. For example, a low-plasticity silt may have modest plasticity but remain highly frost susceptible because of its pore-size distribution.

3. Phase Relationships and Compaction

For a soil mass, the basic relationships are

\[

e=\frac{G_s\gamma_w}{\gamma_d}-1,

\qquad S_r=\frac{wG_s}{e},

\]

where \(e\) is void ratio, \(G_s\) is specific gravity of solids, \(\gamma_w\) is the unit weight of water, \(\gamma_d\) is dry unit weight, \(w\) is water content as a decimal and \(S_r\) is degree of saturation. Rearrangement gives

\[

\gamma_d=\frac{G_s\gamma_w}{1+wG_s/S_r}.

\]

The zero-air-voids, or theoretical saturation, line is the special case \(S_r=1\):

\[

\gamma_{d,zav}=\frac{G_s\gamma_w}{1+wG_s}.

\]

It is a physical upper boundary for a soil at a specified water content, not a normally attainable compaction curve.

Moisture–Density Relationship

In a standard or modified Proctor test, several specimens are compacted at different moisture contents. The resulting curve identifies optimum moisture content (OMC) and maximum dry density (MDD) for the specified effort. Increasing compactive effort commonly increases MDD and decreases OMC, although the response depends on soil type.

Field relative compaction is commonly expressed as

\[

R_c=\frac{\gamma_{d,field}}{\gamma_{d,max,lab}}\times100\%.

\]

The laboratory reference must be stated: standard and modified Proctor results are not interchangeable. Acceptance should also specify moisture limits, lift thickness, equipment, test method and treatment of oversize particles.

Worked Example 1: Phase and Compaction Calculation

A soil has \(G_s=2.70\), water content \(w=0.10\), and degree of saturation \(S_r=0.80\). Taking \(\gamma_w=9.81\,kN/m^3\),

\[

e=\frac{0.10(2.70)}{0.80}=0.3375.

\]

Thus

\[

\gamma_d=\frac{2.70(9.81)}{1+0.3375}=19.80\,kN/m^3.

\]

If the laboratory MDD is \(18.60\,kN/m^3\), the calculated value exceeds it, indicating that the assumed state and laboratory reference are inconsistent. This is an important quality-control finding, not a result to be accepted uncritically. Possible causes include incompatible test data, incorrect water content, oversize correction or an inappropriate laboratory compaction reference.

4. Strength and Stiffness Inputs

California Bearing Ratio

CBR is an empirical penetration resistance ratio:

\[

CBR=\frac{P_{soil}}{P_{standard}}\times100\%.

\]

The standard loads are conventionally associated with penetrations of 2.5 and 5.0 mm; the higher valid ratio is normally reported, subject to the applicable standard. Test preparation must specify compaction effort, soaking, surcharge, drainage and penetration rate. A soaked CBR may represent a conservative wet-season condition, but it does not automatically represent every climate or pavement drainage condition.

CBR is useful for empirical pavement design but has limited transferability to mechanistic analyses. It is strongly affected by moisture, density, particle size, confinement and swelling during soaking.

Resilient Modulus

Resilient modulus is a repeated-load stiffness parameter:

\[

M_r=\frac{\sigma_d}{\varepsilon_r},

\]

where \(\sigma_d\) is the cyclic deviator stress and \(\varepsilon_r\) is recoverable axial strain. It is not identical to Young’s modulus. Because soil response is stress-dependent, a single value should be tied to the anticipated stress state, moisture and conditioning procedure.

Constitutive models may express modulus as a function of bulk and deviator stress, for example a calibrated nonlinear relation of the form

\[

M_r=k_1\left(\frac{\theta}{p_a}\right)^{k_2}\left(\frac{\tau_{oct}}{p_a}+1\right)^{k_3}p_a,

\]

where \(\theta\) is bulk stress, \(\tau_{oct}\) is octahedral shear stress, \(p_a\) is atmospheric pressure and the \(k\)-values are fitted constants. Such equations are model forms, not universal constants; calibration and unit consistency are essential.

Correlation and Selection

Correlations between CBR and \(M_r\) may be useful for preliminary screening but should not replace direct resilient-modulus testing where mechanistic design is important. The selected design value should represent the relevant moisture and density state, not simply the arithmetic mean of all measurements.

5. Frost Susceptibility and Seasonal Design

Frost damage requires three ingredients: frost-susceptible soil, freezing temperatures and a continuing water supply. Frost-susceptible soils commonly contain capillary pore sizes associated with silts and silty sands, although grading, permeability and stratification control actual behaviour.

A quantitative assessment should include:

1. Thermal regime: design freezing index, freezing duration and pavement temperature profile.

2. Water supply: groundwater elevation, drainage path, permeability and capillary continuity.

3. Soil response: laboratory frost-heave rate or heave pressure under controlled temperature gradients.

4. Structural consequence: differential heave, thaw weakening and loss of bearing capacity.

The air-freezing index is commonly approximated as the accumulated sum of daily mean temperatures below 0°C, expressed in degree-days. It is a climatic screening index, not a direct prediction of heave. A simplified heat-flow estimate is

\[

q=k_T\frac{\Delta T}{L},

\]

where \(q\) is heat flux, \(k_T\) is thermal conductivity, \(\Delta T\) is temperature difference and \(L\) is thermal path length. A full frost analysis must consider latent heat, phase change, variable thermal properties and transient boundary conditions.

A practical design criterion may limit predicted heave, for example

\[

H_{pred}=\sum_i h_i\le H_{allow},

\]

where \(h_i\) is the contribution of each frost-active layer. The values of \(h_i\) should come from calibrated frost-heave testing or a validated frost model, not from an unsupported soil-class multiplier. Mitigation options include removing frost-susceptible soil, increasing non-frost-susceptible cover, improving drainage, lowering the groundwater level and using insulation where life-cycle analysis justifies it.

Thaw weakening must be checked separately. A pavement may survive upward heave but fail during spring thaw because saturated subgrade stiffness decreases sharply.

6. Expansive Soils

Expansive soils undergo volume change as suction and water content vary. Risk is associated with clay mineralogy, plasticity, dry density, initial suction, stress state and water-path changes. Atterberg limits alone are screening indicators.

Investigation should include swell pressure, free swell or one-dimensional swell under representative surcharge, suction or soil-water characteristic behaviour, and seasonal moisture monitoring. The design question is whether predicted movement exceeds the pavement’s tolerance, not merely whether the soil is labelled “expansive.”

Mitigation may involve moisture control, replacement, lime treatment, granular cover, chemical barriers or structural separation. Treatments must address both initial expansion and long-term wetting and drying. Construction should avoid placing expansive soil at an uncontrolled moisture state or density that increases future heave.

7. Stabilization and Mix-Design Verification

Stabilization changes soil properties through mechanical blending, cementation, ion exchange, flocculation or reduction of plasticity. Common agents include lime, cement, emulsions, fly ash or other pozzolanic materials, selected only after compatibility testing.

Lime Stabilization

Lime is often effective for plastic clays. Initial tests should establish lime demand, commonly by a standardized pH or equivalent reactivity procedure. Trial dosages are then evaluated for plasticity reduction, workability, swell, durability and strength. A dosage is not justified solely because it reduces PI.

Cementitious Stabilization

Cement-treated soil requires proportioning by dry soil mass, water content and target performance. A trial matrix should vary binder content and curing condition. Typical acceptance properties may include unconfined compressive strength (UCS), wet–dry or freeze–thaw durability, stiffness and permeability. The specified UCS must be tied to the pavement layer function and test age; an arbitrary high strength can increase brittleness and shrinkage cracking.

For a soil mass \(M_s\) and binder dosage \(p\) expressed as a fraction of dry soil mass,

\[

M_b=pM_s.

\]

For 2,400 tonnes of dry soil at 4% binder, \(M_b=0.04(2400)=96\) tonnes. Field dosage must correct for actual dry soil mass and account for binder purity, moisture and mixing uniformity.

Verification Plan

A defensible stabilization specification includes laboratory screening, pilot sections, production-control tests and performance verification. It should define sampling frequency, curing temperature, moisture range, density, mixing depth, pulverization, delay before compaction, binder content and rejection criteria. Chemical compatibility and environmental leachability must be considered where industrial by-products are used.

8. Geosynthetics for Subgrade Problems

Geotextiles, geogrids, geocells and geomembranes perform different functions. A material should not be selected by product name alone.

• Separation: prevents subgrade fines from contaminating aggregate.
• Filtration and drainage: permits water flow while retaining soil.
• Reinforcement: develops tensile resistance and interaction with aggregate.
• Stabilization: improves construction-platform performance over weak soil.
• Containment: limits fluid migration.

Design Verification

A separation or filtration design checks apparent opening size, permittivity, permeability, clogging and survivability. Hydraulic compatibility requires the geosynthetic transmissivity or permittivity to remain adequate under normal stress and anticipated contamination.

A reinforcement design should verify, at minimum:

1. tensile strength after installation damage and environmental reduction;

2. connection and seam strength where applicable;

3. pullout resistance;

4. interface shear or aggregate interlock;

5. creep and durability reduction factors; and

6. required factor of safety.

A simplified pullout resistance check is

\[

R_p=2L\left(\sigma'_v\tan\delta\right)\quad\text{per unit width},

\]

for a simplified two-sided interface model, where \(L\) is embedded length, \(\sigma'_v\) is effective normal stress and \(\delta\) is interface friction angle. This expression excludes anchorage effects, progressive pullout, nonuniform stress and geosynthetic aperture interaction; project-specific test data are preferred.

The design tensile resistance may be expressed as

\[

T_d=\frac{T_{ult}}{RF_{ID}RF_{CR}RF_DRF_{J}},

\]

where the reduction factors represent installation damage, creep, durability and joints/connections. The required resistance must satisfy both ultimate and serviceability criteria. A geosynthetic that has adequate ultimate tensile strength may still fail by excessive rutting, inadequate filtration or poor construction survivability.

9. Spatial Variability and Reliability

Subgrade support varies laterally and vertically. Sources include geological changes, variable compaction, moisture gradients, borrowed fill and construction disturbance. Averaging all results can conceal weak zones.

A practical workflow is:

1. divide the alignment into geotechnically consistent segments;

2. inspect distributions and identify measurement or transcription errors;

3. estimate mean, standard deviation and spatial correlation where data permit;

4. select a characteristic or reliability-based design value; and

5. validate the zoning through additional investigation and construction testing.

For an approximately normal property, a lower fractile can be estimated as

\[

x_k=\bar{x}-z s,

\]

where \(\bar{x}\) is the mean, \(s\) is sample standard deviation and \(z\) corresponds to the selected one-sided reliability level. This approximation is inappropriate for strongly skewed data, censored results or small samples without suitable statistical treatment.

Spatial correlation matters: ten tests clustered within one homogeneous area do not provide the same information as ten well-distributed tests. Geostatistical tools such as variograms and kriging can support mapping, but they do not replace engineering judgement or independent verification.

10. Illustrative Case: Variable Subgrade beneath a Rural Highway

Background

An illustrative 2.4-km highway widening project crosses residual clay, alluvial silt and weathered shale. Preliminary design used one soaked CBR value for the entire alignment.

Problem

Construction records showed pumping in two low-lying sections, while laboratory tests ranged from CBR 2% to 18%. A single average would mask the weak zones.

Analysis

Additional pits and moisture measurements identified three subgrade zones. The alluvial silt had moderate plasticity but high capillary susceptibility and a shallow seasonal water table. The residual clay showed high PI and significant swell under low surcharge. The shale-derived material had high CBR when compacted near OMC but degraded when wet.

For each zone, the design team selected moisture-conditioned CBR and resilient-modulus inputs, checked thaw weakening, and calculated a lower characteristic value rather than using the overall mean. A reliability review showed that the weak alluvial segment controlled thickness.

Approach and Solution

The alluvial segment received improved drainage, a separator, and a controlled granular capping layer. The expansive clay was lime-treated after laboratory dosage and durability trials. The shale-derived fill was placed in thin lifts with moisture control and density verification. Geosynthetic survivability and filtration were verified against the aggregate and subgrade gradations.

Outcome

Post-construction proof rolling and repeat testing confirmed improved uniformity. The revised design required additional treatment only where justified, avoiding unnecessary stabilization across the full project.

Learning Points

The case demonstrates that classification, strength, moisture, frost, swelling, stabilization and spatial reliability must be interpreted together. The controlling design segment is not necessarily the segment with the lowest isolated test result, but neither can weak results be hidden by averaging.

11. Professional Practice, Governance and Emerging Methods

Reports should state sample locations, test standards, specimen preparation, moisture state, compaction effort, curing, units, uncertainty and deviations. Traceability is essential when design values affect public safety and construction cost. Environmental controls are required for lime, cement and industrial by-products, including dust, worker exposure, runoff and potential leaching.

Digital field density systems, automated moisture sensors, remote sensing and machine-learning prediction can improve spatial coverage. However, AI models require representative training data, independent validation and uncertainty reporting. A model should support—not replace—soil identification, laboratory confirmation and professional accountability.

Best-Practice Design Sequence

1. Establish geological and climatic context.

2. Investigate soil boundaries, groundwater and drainage.

3. Classify materials and measure relevant index properties.

4. Determine compaction relationships using the specified effort.

5. Condition strength and stiffness tests to expected field states.

6. Quantify frost, thaw and expansive-soil risks.

7. Develop and verify stabilization or geosynthetic solutions.

8. Divide the alignment into statistically and geotechnically consistent zones.

9. Select characteristic inputs with stated reliability and limitations.

10. Confirm assumptions through construction quality assurance and performance monitoring.

Chapter Summary

Subgrade investigation is an integrated process rather than a search for one representative strength value. Soil classification provides a framework for interpretation but cannot substitute for state-dependent strength and stiffness testing. Phase relationships and compaction theory establish the links among water content, density, void ratio and saturation. CBR is useful for empirical design, whereas resilient modulus supports mechanistic analysis and must be tied to stress and moisture conditions.

Frost assessment requires climate, thermal conditions, water supply and measured or calibrated heave behaviour. Expansive soils require evaluation of wetting, suction, surcharge and mineralogy. Stabilization demands dosage trials, durability testing and field verification. Geosynthetic selection must verify the actual function—separation, filtration, drainage, reinforcement or containment—and account for reduction factors and installation conditions. Finally, spatial variability should be represented through zoning and reliability-based characteristic values rather than concealed by simple averaging.

Key Terms

Subgrade: Soil foundation supporting the pavement structure.

Atterberg limits: Water-content boundaries describing fine-soil consistency.

Plasticity index: Difference between liquid and plastic limits.

Void ratio: Volume of voids divided by volume of solids.

Degree of saturation: Proportion of void space occupied by water.

Optimum moisture content: Water content corresponding to maximum dry density for a specified compaction effort.

Maximum dry density: Highest dry unit weight obtained in a specified compaction test.

Relative compaction: Field dry density expressed as a percentage of laboratory reference density.

CBR: Empirical penetration resistance ratio used in pavement design.

Resilient modulus: Recoverable cyclic stress divided by recoverable strain.

Frost susceptibility: Tendency of soil to develop frost heave and thaw weakening.

Freezing index: Accumulated climatic measure of freezing severity.

Expansive soil: Soil capable of significant volume change with moisture variation.

Lime stabilization: Chemical treatment that modifies and may cement plastic soil.

UCS: Unconfined compressive strength of a cohesive or treated specimen.

Geosynthetic: Polymer product used for separation, filtration, drainage, reinforcement or containment.

Pullout resistance: Resistance mobilized against extraction of a reinforcement.

Characteristic value: Statistical design value representing a specified reliability level.

Spatial variability: Change in engineering properties across location or depth.

Quality assurance: Planned activities confirming that constructed work meets requirements.

Review Questions

Short Answer Questions

1. Why is soil classification insufficient by itself for selecting a subgrade design value?

2. Define optimum moisture content and maximum dry density.

3. Distinguish CBR from resilient modulus.

4. Identify the three essential conditions for frost damage.

5. State two functions that geosynthetics may perform beneath pavements.

Long Answer Questions

1. Explain the investigation and testing sequence for a variable highway subgrade.

2. Discuss the assumptions and limitations of using CBR in pavement design.

3. Describe a quantitative frost and thaw assessment procedure.

4. Develop a laboratory and field verification programme for lime- or cement-stabilized soil.

5. Explain how spatial variability and reliability should influence subgrade zoning and design inputs.

Analytical / Application Questions

1. A soil has \(G_s=2.70\), \(w=0.10\) and \(S_r=0.80\). Calculate its void ratio and dry unit weight using \(\gamma_w=9.81\,kN/m^3\), then explain how you would investigate inconsistency if the result exceeded the laboratory MDD.

2. A project has mean soaked CBR 8% and standard deviation 2.5% in a geologically consistent zone. Estimate the one-sided 5% lower normal fractile using \(z=1.645\), and identify reasons why this estimate might be unsuitable.

3. A weak subgrade is to receive a geogrid. Prepare a verification checklist covering tensile reduction factors, pullout, interface interaction, filtration, installation damage and serviceability rutting.

Multiple Choice Questions with Answers

Question 1

A soil has Gs = 2.70, w = 0.10, and Sr = 0.80. Using γw = 9.81 kN/m³, what is its dry unit weight?

A. 16.40 kN/m³

B. 18.60 kN/m³

C. 19.80 kN/m³

D. 20.90 kN/m³

Correct Answer: C. 19.80 kN/m³

Question 2

Which interpretation is most technically defensible when a calculated in-situ dry unit weight exceeds the laboratory MDD?

A. The field result must be accepted because field compaction always exceeds laboratory compaction.

B. The laboratory MDD should automatically be increased by 10%.

C. The inconsistency should be investigated for incompatible assumptions, oversize correction, moisture data or test reference.

D. The degree of saturation should be assumed to be zero.

Correct Answer: C. The inconsistency should be investigated for incompatible assumptions, oversize correction, moisture data or test reference.

Question 3

A resilient test applies a cyclic deviator stress of 100 kPa and produces recoverable axial strain of 0.0005. What is the resilient modulus?

A. 0.05 MPa

B. 20 MPa

C. 200 MPa

D. 500 MPa

Correct Answer: C. 200 MPa

Question 4

Which condition most directly distinguishes a frost-susceptibility risk from merely having freezing air temperatures?

A. Presence of a continuing water supply and a frost-active pore structure

B. High maximum dry density alone

C. A high cement content in the pavement surface

D. A low traffic volume

Correct Answer: A. Presence of a continuing water supply and a frost-active pore structure

Question 5

A stabilization project treats 2,400 tonnes of dry soil with binder at 4% of dry soil mass. What binder mass is required before corrections for purity or wastage?

A. 24 tonnes

B. 48 tonnes

C. 96 tonnes

D. 600 tonnes

Correct Answer: C. 96 tonnes

Question 6

Which test programme best supports selection of a cement-treated subgrade mix?

A. One UCS test at an unspecified age on an unconditioned specimen

B. Trial binder contents with controlled moisture, density, curing and durability verification

C. Particle-size analysis alone

D. A visual inspection after mixing

Correct Answer: B. Trial binder contents with controlled moisture, density, curing and durability verification

Question 7

For a geogrid reinforcement design, which item is a serviceability verification rather than only an ultimate-strength check?

A. Ultimate tensile strength of the polymer before installation

B. Predicted pavement rutting under design traffic

C. Initial roll width

D. Colour stability in storage

Correct Answer: B. Predicted pavement rutting under design traffic

Question 8

A subgrade property has mean 80 units and standard deviation 12 units. Using a normal one-sided lower fractile with z = 1.645, what value is obtained?

A. 60.3 units

B. 67.8 units

C. 79.9 units

D. 99.7 units

Correct Answer: A. 60.3 units

Question 9

Why can ten closely clustered subgrade tests provide less design information than ten well-distributed tests?

A. Clustered tests always have higher laboratory error.

B. Spatial correlation may make clustered observations represent only one local soil condition.

C. Well-distributed tests require no quality assurance.

D. Clustered tests cannot measure water content.

Correct Answer: B. Spatial correlation may make clustered observations represent only one local soil condition.

Question 10

Which statement about the zero-air-voids line is correct?

A. It is the usual field compaction curve for every soil.

B. It is obtained by setting degree of saturation equal to zero.

C. It is a theoretical upper boundary for a specified water content when Sr = 1.

D. It applies only to coarse aggregate and not soil.

Correct Answer: C. It is a theoretical upper boundary for a specified water content when Sr = 1.