Bridge & Bedrock

Abstract: UK earthworks specifications conventionally determine the suitability of fill at source through classification and state tests — particle-size distribution, plasticity, moisture content and the moisture condition value — while the performance the earthworks must ultimately deliver is, for many end uses, governed by shear strength. Where a design limit state is strength- or stability-controlled, shear strength is properly an end-product requirement; yet it is largely absent from the source characterisation regime, which relies instead on index and state properties as proxies for as-placed strength, and which — where strength does appear — tends to use an undrained, total-stress measure that reflects state rather than the drained, effective-stress strength that governs long-term stability. This paper examines that disconnect. It argues that the relationship between index or state properties and the shear strength of compacted fill is material-specific and conditional; that treating a moisture-condition acceptance window calibrated for compaction and trafficability as a guarantee of design strength is unsafe; and that the resulting gap produces either costly rework or latent reductions in the factor of safety of restored slopes, capping systems and structural fills. A design-led, two-stage framework is proposed in which the governing strength is characterised at source, used to define material- and design-specific acceptance windows, and verified in the end product — closing the loop between what is specified, placed and accepted.

Keywords: earthworks; engineered fill; shear strength; specification; quality assurance; slope stability; quarry restoration; landfill capping

1   Introduction

Earthworks are among the most routine of geotechnical operations and, partly for that reason, among the most heavily proceduralised. In the United Kingdom the dominant framework is the Specification for Highway Works, Series 600 (Earthworks), within the Manual of Contract Documents for Highway Works, supported by BS 6031 and, increasingly, the BS EN 16907 series. These documents provide a mature, well-tested system for classifying fill materials, defining the conditions under which they may be placed, and controlling their compaction. That system is applied far beyond the highway context from which it grew — to quarry restoration, landfill engineering, flood defence and development platforms — usually with local adaptation.

The system rests on a particular logic. Material is assessed for suitability at source, principally by classification and state tests: particle-size distribution and plasticity to identify what the material is, and moisture content or the moisture condition value (MCV) to describe the state it is in when placed. Acceptability is then expressed as limits on these properties, and compaction is controlled either by method or by an end-product density or air-voids requirement. The premise is that a material of an acceptable class, placed within an acceptable moisture window and compacted to an acceptable density, will perform adequately.

For a large proportion of general earthworks that premise is sound, because the governing consideration is deformation, trafficability and the achievement of adequate compaction, for which index and state properties are reliable controls. The difficulty arises where the governing limit state of the works is instead one of shear strength or stability — as it is for restored quarry slopes and tips, for capping and cover systems, for reinforced soil, and for structural fill beneath development. In those cases shear strength is what the design depends on and what the end product must deliver; and yet shear strength is frequently absent from the source characterisation on which suitability is judged. This paper is concerned with that disconnect: its nature, its causes, its consequences, and how the source and end-product regimes might be brought back into alignment.

2   The Conventional Acceptance Framework and Its Internal Logic

The conventional framework characterises a fill material along two axes. The first is intrinsic identity — what the material is — captured by grading and, for fine-grained soils, by the Atterberg limits. These properties are essentially invariant for a given material and are used to assign it to a class. The second is state — the condition of the material at the moment of placement — captured principally by moisture content and by the MCV, a rapid field measure of how far a soil is from a condition at which further compactive effort produces no further reduction in air voids. State is variable, weather-dependent and the property that site control chiefly manages.

Acceptability limits are then set against these measures. A cohesive fill, for instance, may be deemed acceptable within a defined moisture-content range, or between MCV bounds, or — in some formulations — between limits of remoulded undrained shear strength. Compaction is verified either by prescribing plant, layer thickness and passes (method compaction), or by requiring a proportion of a reference dry density or a maximum air-voids content (end-product compaction). The MCV itself was developed precisely to give a rapid, repeatable field control on state, and its acceptability limits are calibrated to ensure that material can be trafficked, spread and compacted to the required density without becoming unstable during construction.

It is important to recognise what this framework is designed to guarantee. Its limits secure workability, adequate compaction and short-term (construction-stage) stability. Where an undrained shear strength appears among the acceptability criteria, it functions in the same register: as a total-stress, state-dependent measure of consistency, closely allied to moisture content and MCV, used to bound trafficability and short-term behaviour. It is not, and does not purport to be, a characterisation of the drained, effective-stress strength that will govern the long-term stability of the completed works. The framework, in other words, characterises identity and state, and controls the construction process; it does not characterise the governing strength of the finished product.

3   The Disconnect: State and Short-Term Strength versus Governing Long-Term Strength

For strength-critical earthworks the design limit state is, overwhelmingly, a drained (or partially drained) effective-stress condition assessed over the long term. The stability of a restored quarry slope, the veneer stability of a soil cover over a lining system, the global stability of a tip or a raised landform, and the bearing and settlement performance of structural fill are all governed by the effective-stress shear strength parameters of the placed material — the effective cohesion and angle of shearing resistance — together with the pore-water regime that will develop over the design life. These are the parameters a competent design assigns and against which it demonstrates an adequate factor of safety.

The disconnect is that these governing parameters are seldom characterised at source and seldom feature in the acceptance chain. The source suitability assessment establishes class and state; the design, meanwhile, adopts effective-strength parameters that are typically taken from published typical values for the lithology, from correlations with index properties, or from ground investigation testing of the in-situ deposit rather than of the material in its placed and compacted condition. The specification then controls placement by moisture and density. At no point is it demonstrated, by test, that the material as it will actually be placed can deliver the effective strength the design relied upon — nor is that strength verified in the end product beyond the proxy of achieving the specified density and moisture.

This is more than a pedantic distinction between total and effective stress. The in-situ strength of a natural deposit and the as-placed strength of the same material after excavation, reworking and recompaction can differ substantially, because excavation destroys the natural fabric and stress history and compaction imposes a new fabric, density and suction regime. Effective-strength parameters derived from undisturbed samples of the source deposit may therefore be unrepresentative of the compacted fill; and parameters inferred from index properties carry the material-specific scatter discussed below. Eurocode 7 requires the characteristic value of a governing parameter to be a cautious estimate derived, so far as possible, from tests representative of the limit state under consideration. A design that assigns an effective angle of shearing resistance to a compacted fill without having tested that fill in a representative condition is, in this sense, importing an assumption into the very parameter on which safety turns — and the acceptance regime provides no mechanism to confirm or refute it.

4   Why Index and State Properties Are Conditional Proxies for Strength

The framework survives in practice because index and state properties do correlate, loosely, with strength. The objection is not that the correlation is absent but that it is conditional — specific to the material and to the placement condition — and therefore unsafe to rely upon as a universal guarantee. Several mechanisms make this so.

First, the effective strength of a compacted soil depends strongly on its state relative to the optimum: material compacted dry of optimum tends to a stiffer, more brittle, higher-strength fabric, while the same material compacted wet of optimum develops a more dispersed fabric, lower strength and higher susceptibility to strain-softening and to strength loss on saturation. Two lots of a single material, both compacted to the specified density and both within the acceptable moisture window, can therefore reach the end product with materially different effective strengths. The moisture window that secures compaction is generally wider than the window that secures a given strength.

Second, plasticity is a weak predictor of effective friction. Materials of very similar plasticity index can exhibit different angles of shearing resistance according to mineralogy, particle shape and grading; the presence of platy clay minerals, in particular, depresses the residual and sometimes the peak friction angle in a way that the plasticity index only partially captures. Reliance on class alone to infer strength conflates materials that behave differently once sheared.

Third, the parameters relevant to the different limit states are not interchangeable. Peak strength governs first-time stability in intact fill; residual or fully softened strength governs where a pre-existing shear surface exists or where high-plasticity clay fill is subject to strain over the long term. A source screen based on plasticity and moisture speaks to none of these directly. Fourth, the pore-pressure regime that will develop — from infiltration, from a rising water table in a restored void, from consolidation of underlying materials, or from the loss of construction-stage suction as the fill equilibrates — is a design consideration entirely outside the source-characterisation vocabulary, yet it can dominate the long-term factor of safety.

The cumulative point is that passing the conventional source and acceptance criteria is necessary but not sufficient for delivering a strength-governed design. The criteria bound the state and secure the construction process; they do not measure the parameter that governs the finished works.

5   Consequences

The practical consequences of the gap fall into two categories. The first is rework. Where end-product strength is, belatedly, required — for instance where a validation or CQA regime for a restoration landform or a capping layer calls for confirmation of shear strength — material that satisfied every source and placement criterion may nonetheless fail to demonstrate the design strength. The remedies are all expensive and programme-critical: re-excavation and re-placement, drying-back or conditioning, the importing of replacement material, or a redesign of the profile to suit the strength actually achieved. Because the deficiency is discovered downstream, after placement, its cost is far greater than it would have been had the material been characterised for strength before it was won and placed.

The second, and more serious, consequence is the latent defect. Where strength is not verified in the end product at all — where acceptance rests entirely on density and moisture — a shortfall in the governing parameter is simply not detected. The works are signed off, but the completed slope, cover or platform carries a lower factor of safety than the design assumed. In restoration and landfill contexts this is not a transient construction risk but a long-term one: veneer instability of cover soils over low-friction interfaces is a well-documented failure mode; restored slopes and tips can deteriorate as pore pressures rise and suctions dissipate; and structural fill that under-performs in strength or stiffness transmits distress to what is built upon it. A defect embedded in the acceptance logic is, moreover, systematic rather than random — it recurs wherever the same specification is applied to the same class of works.

Both consequences share a single root: an acceptance regime whose measured quantities are not the quantity the design depends on. The remedy is correspondingly single — to characterise, and then to verify, the governing parameter.

6   A Design-Led Framework for Alignment

The principle proposed here is simple to state: source characterisation should measure the parameters that govern the design, and the acceptance chain should be expressed in terms that trace back to those parameters. Where a limit state is strength-governed, that means characterising the effective shear strength of the material in its placed condition at source, and verifying it in the end product. The following four elements give the principle practical form.

Parameter selection should be design-led. Before the source-characterisation suite is fixed, the governing limit states of the works should be identified, and the parameters that control them made explicit. For a strength-governed landform this means the relevant effective-strength parameters — peak, and where appropriate fully softened or residual — and the design pore-pressure condition. The characterisation suite then follows from the design, rather than being adopted wholesale from a generic specification.

Strength should be characterised at source on representatively compacted material. Samples of the candidate material should be recompacted to the anticipated placement condition — the specified relative compaction, across the range of the proposed moisture window — and sheared, by triaxial or direct shear testing, to establish the effective-strength parameters actually achievable. This does two things: it confirms, at feasibility or design stage, that the available arisings or borrow can deliver the design strength; and it maps the strength achieved against placement moisture and density, so that the acceptance window can be defined by strength rather than assumed to follow from compaction.

The acceptance window should be set from the strength envelope. Rather than adopting a default moisture or MCV window and hoping that adequate strength follows, the specification should adopt the window within which the source testing demonstrated that the design strength is achievable at the specified density. In general this strength-derived window will be tighter than, and lie within, the compaction-derived window; the difference is precisely the margin that the conventional approach leaves unguarded. The moisture, MCV and density limits thereby become material- and design-specific controls that are known to map onto the governing parameter, rather than generic proxies.

The end product should be verified against the same envelope. Routine production control can and should remain the fast, cheap measures — in-situ density and moisture, MCV — because, once calibrated against strength for the specific material, they are legitimate control tools. But that calibration must be established at the outset and re-validated periodically by strength testing on the placed fill, whether by recompacted specimens or by in-situ methods, at a frequency proportionate to the consequence of failure. The acceptance record then closes the loop: the design strength, the source testing that showed it was achievable, the placement controls that keep the material within the strength-verified window, and the end-product tests that confirm it — all expressed in a single, consistent currency.

None of this displaces the existing specification machinery; it disciplines it. Classification and state testing retain their role in identifying the material and controlling its placement. What changes is that, for strength-governed works, they are subordinated to, and calibrated against, a characterisation of the parameter that actually governs — so that suitability at source and acceptance in the end product become statements about the same thing.

7   Illustrative Example

The mechanism is most easily seen in a schematic example. Consider a cohesive fill, derived from on-site arisings, to be placed as a restored slope whose long-term drained stability requires an effective angle of shearing resistance of about 25° (with a small effective cohesion) to achieve the target factor of safety of 1.3. Suppose the conventional acceptance window, set for compaction and trafficability, admits the material across a moisture range corresponding to MCV values from roughly 8 to 13. The table below shows, schematically, the effect of placement moisture on the effective friction angle achieved and on the resulting slope factor of safety, all at the specified relative compaction.

Placement moisture	MCV (approx.)	Compaction / state acceptance	Achieved φ′ (°)	Slope FoS
Dry of optimum	12.5	Pass	27	1.55
Near optimum	10.5	Pass	25	1.35
Slightly wet	9.5	Pass	23	1.18
Wet of optimum	8.3	Pass	21	1.02

Every row in the table satisfies the conventional acceptance criteria: each is within the compaction- and trafficability-derived window, and each meets the density requirement. Yet only the drier two rows deliver the design strength and an adequate factor of safety; the wetter rows, though fully acceptable on the conventional criteria, leave the slope at or below limiting equilibrium. The conventional window admits material that will not perform. Had the acceptance window instead been set from strength — restricting placement to the moisture range that testing showed delivers an effective friction angle of at least 25° — it would have been tighter, excluding the wetter material, and the completed slope would carry the intended margin. The example is deliberately schematic, but the shape of the result is general: the compaction-secured window is wider than the strength-secured window, and the difference is where latent deficiency hides.

8   Discussion

Two qualifications keep the argument proportionate. The first concerns scope. The conventional framework is not deficient in general; it is deficient only where a strength or stability limit state governs and where that fact is not carried through into characterisation and acceptance. For the large body of general earthworks in which deformation, trafficability and compaction govern, index and state controls remain appropriate and strength testing would be disproportionate. The discipline proposed here should be triggered by the design, applied where the governing parameter warrants it, and not imposed indiscriminately.

The second concerns cost and practicability. Effective-strength testing is slower and more expensive than index and state testing, and cannot serve as a high-frequency production control. This is why the framework retains the fast proxies for routine control and confines strength testing to two disciplined roles: establishing the strength envelope at source, and calibrating and periodically verifying the proxies against it. The additional cost is modest against the value at risk — the cost of rework, or of a latent reduction in the factor of safety of a permanent landform — and is incurred at the stage where it does most good, before material is won and placed. There remain genuine difficulties in obtaining representative recompacted specimens and in reconciling laboratory and in-situ strengths, and these warrant careful, material-specific judgement; they are reasons to characterise strength thoughtfully, not reasons to leave it uncharacterised.

9   Conclusions and Recommendations

Earthworks acceptance in UK practice characterises the identity and state of fill and controls its placement, but for strength-governed works it does not characterise the parameter that governs the finished product. The effective shear strength on which the long-term stability of restored slopes, cover systems, tips and structural fills depends is typically assumed in design and inferred, through conditional proxies, in acceptance — rather than measured at source and verified in the end product. The proxy relationship between index or state properties and as-placed strength is real but material-specific and conditional, and the acceptance window that secures compaction is generally wider than the window that secures the design strength. The gap is discharged either as costly downstream rework or as a latent, systematic reduction in the factor of safety of permanent works.

The remedy is to align source characterisation and end-product acceptance around the governing parameter. Specifically, it is recommended that:

• the source-characterisation suite be derived from the governing limit states of the design, so that where a limit state is strength-governed, the relevant effective-strength parameters are characterised at source rather than assumed;
• shear strength be tested on representatively recompacted material across the proposed placement window, both to confirm at design stage that the available material can deliver the design strength and to map strength against placement state;
• the acceptance window for moisture, MCV and density be set from the strength envelope so derived, rather than adopted generically, so that the routine placement controls are known to map onto the governing parameter;
• end-product verification retain the fast proxy measures for routine control but calibrate and periodically re-validate them against strength testing of the placed fill, at a frequency proportionate to the consequence of failure;
• the earthworks specification and the associated construction quality assurance or validation plan state explicitly the design strength requirement, the source characterisation supporting it, the production controls and the end-product verification, so that suitability and acceptance are expressed in a single consistent currency; and
• the discipline be applied proportionately — triggered by the design where a strength or stability limit state governs, and not imposed on general earthworks where deformation and compaction govern and index controls remain appropriate.

Applied in this way, the additional characterisation is neither onerous nor novel in its individual components; it simply requires that the parameter the design depends upon be the parameter the specification measures. That the two are so often different is a weakness in the logic of the acceptance chain, not an inherent limitation of earthworks practice, and it is readily corrected.

References

British Standards Institution, 1990. BS 1377 Methods of test for soils for civil engineering purposes. BSI, London.

British Standards Institution, 2009. BS 6031:2009 Code of practice for earthworks. BSI, London.

British Standards Institution, 2004 (incorporating amendment A1:2013). BS EN 1997-1 Eurocode 7: Geotechnical design — Part 1: General rules. BSI, London.

British Standards Institution, 2018–2019. BS EN 16907 (series) Earthworks. BSI, London.

Head, K.H. & Epps, R.J., 2011–2014. Manual of Soil Laboratory Testing (3 vols). Whittles Publishing, Dunbeath.

Koerner, R.M., 2012. Designing with Geosynthetics, 6th edition. Xlibris.

National Highways. Manual of Contract Documents for Highway Works, Volume 1: Specification for Highway Works, Series 600 — Earthworks (and Volume 2, Notes for Guidance). The Stationery Office.

Parsons, A.W., 1976. The rapid measurement of the moisture condition of earthwork material. TRRL Laboratory Report 750. Transport and Road Research Laboratory, Crowthorne.

Trenter, N.A., 2001. Earthworks: A Guide. Thomas Telford, London.