Marine Clay SHANSEP, OCR and Undrained Strength Explained

What a SHANSEP chart is really telling you

A SHANSEP chart looks like a collection of laboratory strength results, but it is really a story about stress history. SHANSEP stands for Stress History And Normalized Soil Engineering Properties. The name is worth unfolding, because a reader can otherwise mistake the graph for a single ready-made design formula. The central question is not merely how strongly a marine clay resisted one test. It is how its measured undrained resistance changes when the same kind of soil has experienced different prior maximum effective stresses, is returned to different current effective stresses, and is then sheared in a specified way. That is the difference between reading an isolated number and translating an experiment into a defensible interpretation.

Imagine two core samples taken from similar-looking marine clay. Both presently sit under an estimated effective vertical stress of 80 kPa. One was never substantially loaded beyond that level. The other has evidence of an earlier 160 kPa maximum effective vertical stress. A laboratory result of 20 kPa undrained shear strength for the first and 31 kPa for the second does not necessarily mean the second has a completely different mineral composition. Its earlier loading can help explain the difference. Equally, a careful reader should not immediately attribute every difference to stress history: sample disturbance, soil structure, testing direction and consolidation procedure can change the reported resistance.

A practical translation starts with the labels. Undrained strength, often written s_u, is a stress, reported in kPa. Effective vertical stress, sigma-prime-v, is also a stress, reported in kPa. Overconsolidation ratio, OCR, has no units because it divides a prior maximum effective stress by the current effective stress. Normalised strength s_u divided by current effective vertical stress is likewise dimensionless. Keeping those three quantities separate is the quickest way to avoid errors when reading a SHANSEP table or trying to explain it in another language.

Why the original research matters

Charles Ladd and Roger Foott introduced the SHANSEP design approach in their 1974 paper, New Design Procedure for Stability of Soft Clays. Their contribution was not a magic constant. It was a method of considering measured stress history and normalised laboratory behavior together, while recognizing that laboratory shearing should represent the relevant mode of deformation. Their paper described how to reconsolidate specimens and obtain relationships that could be combined with measured overconsolidation ratios. The approach was a response to the weaknesses of selecting strengths from scattered conventional tests without a coherent account of stress history.

A later offshore example by Koutsoftas and Ladd in 1985 concerned soft marine clay associated with a breakwater project. Its value for the reader is methodological: it shows why a project may compare conventional undrained test interpretation with a program that characterises in-situ preconsolidation pressure and strength under compression, extension and direct simple shear. The point is not that a value from that project should be copied into an unrelated marine clay deposit. The point is that strength is conditional on its testing and geological context, so a strength profile needs its provenance.

Modern offshore geotechnical guidance, including DNV-RP-C212, places such interpretations within a broader investigation and foundation assessment process. Laboratory relationships are evidence, not an automatic substitute for sound sampling, geological description, field testing, groundwater conditions or a project-specific design framework. A reader interpreting SHANSEP should be able to state what was tested, what was inferred, which assumptions are provisional and where engineering judgment remains necessary. This article teaches how to interpret those claims, not how to certify a particular foundation.

The three measurements to identify first

First identify present vertical effective stress at the depth being assessed. Effective stress is total vertical stress minus pore-water pressure, with careful attention to datum, submerged unit weight and the groundwater or seawater boundary. A table may provide effective vertical stress directly; if it does, do not subtract pore pressure again. If you calculate it, state the profile and whether loading conditions are static. Marine sediments may be subject to changing overburden, water level or construction history. A simple single number is an approximation to a defined state, not necessarily a timeless constant.

Next identify preconsolidation pressure or the best-supported estimate of past maximum vertical effective stress. Often this is written sigma-prime-p. The usual OCR calculation is sigma-prime-p divided by sigma-prime-v0, where the latter is the present effective vertical stress at the same specimen depth. If a report says OCR equals two, ask whether the value came from an oedometer curve, another laboratory procedure, a field correlation or an interpreted stratigraphic profile. Each route has measurement uncertainty, and some natural clays are influenced by aging, bonding and chemistry, not only removal of former overburden.

Third identify the undrained shearing mode. A triaxial compression result, triaxial extension result and direct simple shear result are not interchangeable simply because they share units of kPa. Principal stresses evolve differently and some clays show pronounced anisotropy. The meaning of S and m in a fitted normalisation changes with the specimen preparation, consolidation path, failure criterion and shearing mode. When a report offers three SHANSEP envelopes, preserve their labels and do not average them into a supposedly universal strength law.

How to read the familiar power relationship

One convenient representation is s_u divided by sigma-prime-v0 equals S multiplied by OCR raised to power m. S is the normalized undrained strength at OCR equal to one for the specific material and testing mode. The exponent m describes the fitted dependence on OCR over the experimental range. There is no unit on S, OCR or m, so the right side remains dimensionless. Multiplying by sigma-prime-v0 returns strength in kPa. A common reading mistake is treating S as a stress in kPa, which would make the equation dimensionally inconsistent.

Suppose an illustrative marine clay has S equal to 0.22 and m equal to 0.80 in direct simple shear. If current effective vertical stress is 80 kPa and the estimated preconsolidation pressure is 160 kPa, OCR is 160 divided by 80, or 2. The normalized strength is 0.22 times 2 to the power 0.80, approximately 0.383. The calculated undrained strength is then about 30.6 kPa. Those arithmetic steps can be checked independently: 2 to the power 0.80 is approximately 1.741, multiplied by 0.22 is 0.383, and multiplied by 80 is 30.6.

These numbers are intentionally illustrative, not recommended parameter values. The S and m pair should be inferred from suitable tests and assessed for its actual stress range. An exponent fitted between OCR one and three cannot be presumed reliable at OCR fifteen. If the clay changes composition with depth, or its structure and cementation depart from the tested population, one smooth curve may disguise separate behaviors. A transparent interpretation says where the curve is an interpolation supported by data and where it would become an extrapolation requiring fresh evidence.

A worked example with the same clay at two depths

Consider a teaching table with two depths and one laboratory-derived relation: S equals 0.22, m equals 0.80, and direct simple shear is the chosen mode. At depth A the present effective stress is 80 kPa and the estimated prior maximum is 160 kPa. At depth B the present effective stress is 120 kPa and the prior maximum is 168 kPa. The OCR values are 2.00 and 1.40 respectively. We want both normalized strengths and the final predicted undrained strengths, without assuming that a larger OCR always produces a larger absolute strength.

Depth A gives 0.22 times 2.00 to the power 0.80, or about 0.383, and therefore 30.6 kPa. Depth B gives 0.22 times 1.40 to the power 0.80, about 0.288, and therefore approximately 34.6 kPa. Depth A has the larger normalized strength because it has the larger OCR. Depth B nevertheless has the larger absolute strength because its current effective confining stress is greater. This is a useful example of why a SHANSEP plot and a depth profile can tell different-looking stories while both remain correct.

If someone translates this result as “overconsolidated soil is always stronger,” the wording is incomplete. Stronger than what, under which present stress and in which shearing mode? A safer sentence is: “In these hypothetical tests, normalised undrained strength increases with OCR, but calculated absolute strength also depends on the current effective vertical stress.” The revised version has a denominator, a condition and a scope. Those three additions prevent a common verbal error from becoming a quantitative misunderstanding.

How the laboratory evidence is assembled

A useful SHANSEP laboratory program commonly involves reconsolidating specimens to controlled effective stresses, creating normally consolidated and overconsolidated states, and shearing specimens under specified paths. Each specimen should have a traceable depth, identification, water content, quality record, loading history and selected failure criterion. Do not assume every specimen underwent an identical path simply because the final results appear on the same chart. Some may have different unloading stages or dwell times, and those distinctions can affect measured behavior.

For a normally consolidated reference, the current effective stress and maximum past effective stress are approximately equal in the simplified interpretation, so OCR is one. In the power model, OCR to any finite exponent is one, leaving the normalized strength equal to S. For overconsolidated points, the normalized result is plotted against OCR. A trend may be fitted to data, for example by taking logarithms of the dimensionless variables. The plotted relation should ideally show individual observations, not merely an elegant line and two coefficients.

Consider what a fitted line does not reveal by itself. It may hide a small number of tests, correlated errors, sample quality differences, uncertainty in the calculated OCR or differences between laboratories. It may also hide whether the engineer fitted peak strength, strength at a particular shear strain or another specified limit state. A title saying “SHANSEP strength” is not enough. A useful reader asks for the sample table and test definition so that the relationship can be independently interpreted rather than treated as authority by typography.

Why sample disturbance deserves its own warning

Marine clay is often soft, water-rich and sensitive to extraction, stress relief, transport and trimming. A specimen can look intact while having lost some of its field structure. High-quality block samples and conventional tube samples can produce materially different stress–strain responses after reconsolidation. Research on Norwegian marine clays by Lunne and colleagues found that disturbance affects measured responses and can change how peak and post-peak resistance appear. This means the quality of a SHANSEP curve is partly a quality-of-evidence problem, not merely a mathematical fitting problem.

Suppose a set of disturbed specimens systematically underrepresents peak resistance at small strain. Fitting a smooth power law through those data may yield a visually convincing curve, yet it has not repaired the specimen quality problem. The reverse mistake is to adopt the highest observed strength without considering whether it represents the design stress path and failure strain. A careful interpretation compares specimen quality indicators, field correlations, stress–strain curves, independent tests and geological consistency. Repeated calculations cannot restore information lost when the original soil structure was disturbed.

When translating a laboratory report, separate observation from attribution. “The sample reconsolidated with a larger void-ratio change” describes what was measured. “The sample was heavily disturbed” is an interpretation that requires a quality criterion appropriate to the clay. “The deposit is unsafe” is a design conclusion that cannot be drawn from that one indicator alone. Preserving those levels makes an explanation both more useful and more honest, especially when different specialists will review it later.

Three different failure modes, not three names for one value

In triaxial compression, the major principal stress increases relative to the minor stress under a defined path. In triaxial extension, the principal stress configuration changes differently. In direct simple shear, the specimen is sheared along a horizontal plane while principal stress directions rotate. Each test probes a different stress pathway. Natural clay fabric, depositional layering and prior K0 consolidation may make measured resistance depend on loading direction. A report should therefore label s_u,compression, s_u,extension and s_u,DSS rather than silently substitute one for another.

Imagine a hypothetical set at 100 kPa current effective stress and OCR 1.5: the direct simple shear relation predicts 30 kPa, the compression relation predicts 36 kPa, and extension predicts 25 kPa. These are invented teaching outputs, not an assertion about any particular marine clay. They are enough to demonstrate the interpretive trap: reporting “the strength is 36 kPa” loses the shearing-mode qualification and can materially change an assessment. If an application requires several stress paths, the relationship among them needs to be justified, not hidden behind a single average.

A table comparing tests should also specify the chosen strength criterion. Peak shear stress, a defined strain value and a residual or large-strain value can differ. The phrase “undrained strength” is not a guarantee that every row uses the same definition. Reading across datasets requires checking specimen geometry, drainage control, strain rate, consolidation stress, stress path and interpretation rule. Those checks are especially important when a graph was assembled from multiple research reports rather than from one coherent laboratory campaign.

A second worked example: recover the coefficient from observations

Suppose a normally consolidated direct simple shear test at effective vertical stress 100 kPa produces an interpreted undrained strength of 23 kPa. For this single idealised observation at OCR one, the normalized strength is 23 divided by 100, or 0.23, and it provides a candidate S. Suppose a second otherwise comparable observation at OCR four and effective stress 100 kPa gives 69.7 kPa. Its normalized strength is 0.697. Divide 0.697 by 0.23 to obtain approximately 3.03, which in the power model equals four to the power m.

Taking logarithms gives m approximately ln(3.03) divided by ln(4), or 0.80. This is a calculation exercise illustrating how a power exponent is recovered. Real fitting should use multiple suitable data points, uncertainty and appropriate weighting rather than letting just two observations dictate a model. In particular, the test conditions must be comparable, because mixing compression and extension data would make the apparent OCR exponent partly a stress-path effect masquerading as a stress-history relation.

Check the result by substitution: with S 0.23, m 0.80 and OCR four, the normalized prediction is 0.23 times about 3.031, or 0.697. Multiply by 100 kPa to recover 69.7 kPa. A reader may now translate the graph as a modeled relation between dimensionless normalized resistance and OCR, supported by a defined data set. That is more accurate than “strength rises exponentially.” A power law is not the same mathematical object as an exponential function in OCR, even if both can produce curves that rise.

A third worked example: uncertainty changes the conclusion

Suppose the present effective vertical stress is estimated as 80 kPa, while the best estimate of maximum past effective stress is 160 kPa. If the latter may plausibly range from 140 to 180 kPa, the OCR range, holding current stress fixed, is 1.75 to 2.25. With S equal to 0.22 and m equal to 0.80, the corresponding normalized strengths are approximately 0.344 and 0.421. The resulting absolute strengths are roughly 27.5 and 33.7 kPa, compared with the central estimate of 30.6 kPa.

This simple range illustrates sensitivity; it is not a statistical confidence interval. It considers uncertainty in just one input and temporarily treats the fitted S and m values and current stress as exact. A proper uncertainty appraisal would also examine the variability of the laboratory strengths, the selection of failure criterion, dependence among parameters, depth registration and perhaps model-form uncertainty. If samples with higher preconsolidation estimates also have systematically different S estimates, independent extreme-value combinations may be misleading.

An important distinction is the difference between measurement uncertainty and variability in nature. One can have accurately measured samples from genuinely variable clay layers; one can also have a homogeneous layer tested with imprecise equipment. More samples do not automatically solve poor measurement practice, and more decimal places do not turn an uncertain OCR into an exact number. A credible graph or explanation communicates which kind of uncertainty is being discussed and does not turn an illustrative band into a design safety factor.

Reading an OCR-depth plot without losing the denominator

An OCR-depth plot may show values that fall toward one at greater depth, rise near an erosion surface or vary sharply across stratigraphic units. Those features can invite geological stories, but first check how each ratio was formed. OCR equals preconsolidation pressure divided by current vertical effective stress. A rising effective overburden can reduce OCR even if preconsolidation pressure is also increasing. A ratio profile should therefore be read alongside the two underlying stress profiles, not isolated as a decorative line.

Suppose preconsolidation pressure rises from 150 kPa to 180 kPa between two depths, while current effective stress rises from 75 kPa to 150 kPa. The OCR falls from 2.0 to 1.2 even though the estimated prior maximum stress has increased. Nothing contradictory has happened: the denominator has grown faster. If an explainer says “the deeper material has lost past loading,” that conclusion is not supported by these numbers. It is better to say the deeper material has a smaller ratio of interpreted maximum past to current effective stress.

Geological mechanisms require separate investigation. Loading, unloading, erosion, aging, desiccation, cementation, changes in pore-fluid chemistry and sample disturbance can influence interpretation. Do not automatically rename apparent overconsolidation as historical erosion. An explanatory article should show which observations distinguish among possible causes. That approach teaches the reader how to ask the next useful question rather than encouraging a confident but untested origin story.

Translating the curve to a strength profile

To turn a fitted normalized relation into a strength-versus-depth profile, apply it at each depth using that depth’s effective stress and OCR. This requires a consistent stress model and appropriate layer boundaries. Do not calculate one profile with submerged unit weight but another with total unit weight, or mix kPa and MPa without explicit conversion. The normalized law is dimensionless; strength must regain the correct unit by multiplication with the local current effective stress.

Consider three illustrative depths with current vertical effective stresses of 40, 80 and 120 kPa and OCR values of 2.5, 2.0 and 1.4. With S equal to 0.22 and m equal to 0.80, the normalized strengths are approximately 0.458, 0.383 and 0.288 respectively. Their corresponding absolute strengths are about 18.3, 30.6 and 34.6 kPa. The profile strengthens with depth even while the dimensionless normalized strength decreases. That is precisely why plot axes and denominators should not be omitted when results are simplified for a general audience.

If the clay at the third depth actually belongs to a different geological unit, reusing the same S and m may be inappropriate. A line drawn across a stratigraphic boundary can conceal a model change. A better chart marks unit boundaries, tests and fitted parameters, making it clear where one relationship ends and another begins. If different layers have insufficient data, display that uncertainty rather than inventing a seamless profile.

What the curve cannot tell you on its own

SHANSEP does not automatically predict drainage behavior during construction. Undrained strength relates to loading sufficiently rapid that pore-water drainage is limited under the relevant conditions, but different construction sequences can permit consolidation and alter effective stress over time. Likewise, a stress-history correlation does not by itself establish settlement, permeability, cyclic degradation, creep, strain softening or interface friction. Each of those requires additional evidence and perhaps a different laboratory or field test.

It also does not remove the need to evaluate site-specific hazards. Marine soils can be highly variable, especially where tidal channels, mass-transport deposits, reworked sediment, buried lenses or shallow gas occur. A calculated 30 kPa strength at a sampling point does not ensure that the same value applies horizontally across a foundation footprint. Geometry, anisotropy, strain localization, time and spatial continuity all matter to interpretation. These are not footnotes to a universal formula: they define whether the formula has been applied in its appropriate domain.

The calculation is educational and should not be used as a standalone offshore foundation design. Design decisions need qualified engineering review, appropriate standards, verified site investigations and explicit limit-state assumptions. In teaching, this boundary is productive rather than obstructive: it asks the reader to distinguish a correct arithmetic operation from a justified engineering conclusion. A strong translator of technical results knows when to stop at the evidence.

Common errors and how to repair them

Error one is dividing undrained strength by total stress rather than effective stress. If total stress is 150 kPa and pore pressure is 70 kPa, current effective stress is 80 kPa. A measured strength of 24 kPa corresponds to a normalized ratio of 0.30 using effective stress, not 0.16 using total stress. The correction is to identify the stress convention and write the subtraction with units before fitting or applying a relationship.

Error two is taking OCR as present stress divided by prior maximum stress. That reverses the physical interpretation: an overconsolidated specimen with prior maximum 160 kPa and current 80 kPa should give OCR two, not one-half. The repair is to write numerator and denominator explicitly and check that OCR for a normally consolidated reference is approximately one. Values below one may indicate a misdefinition, unusual interpretation or other conditions warranting explanation rather than being silently accepted.

Error three is mixing strength criteria or shearing modes. If one row reports compression peak at 8% axial strain and another reports a direct simple shear value at a different criterion, a combined regression can acquire apparent curvature that is not solely related to OCR. The repair is to group comparable observations, record the failure criteria and treat each fitted relation according to its proven scope. A convenient spreadsheet merge should not determine material behavior.

Error four is translating “normalised” as “standardised” in the sense of universal validation. Normalisation here means scaling a measured stress by a specified reference stress. It does not imply that the result is automatically standardized across laboratories, clays or field conditions. The repair is to name the actual divisor each time: normalized undrained strength equals undrained strength divided by current effective vertical stress.

Error five is reporting a precise predicted strength from imprecise input. If current stress, OCR and fitted parameters have material uncertainty, printing 30.64183 kPa miscommunicates evidence. Report an appropriately rounded value and an honest uncertainty account. Error six is using a laboratory parameter without identifying sample quality. The repair is to preserve specimen provenance and place quality concerns alongside the fit rather than in a distant appendix.

Transfer exercise one: the hidden stress conversion

A report shows total vertical stress of 210 kPa, pore-water pressure of 90 kPa and preconsolidation pressure of 240 kPa. It lists S equal to 0.20 and m equal to 0.75 for the relevant illustrative laboratory mode. Calculate effective vertical stress, OCR, normalized strength and absolute undrained strength. Then explain why dividing by 210 kPa in the final step would be inconsistent with the model’s definition. Work the numbers independently before checking the answer.

Answer: effective stress is 210 minus 90, or 120 kPa. OCR is 240 divided by 120, or 2. The normalized strength is 0.20 times 2 to the power 0.75, approximately 0.336. The absolute predicted strength is about 40.4 kPa. The 210 kPa total stress includes pore pressure and is not the effective vertical stress used in the stated normalized relation. This example also reminds the reader that changing the pore-pressure assumption changes the denominator and OCR together.

Transfer exercise two: compare two interpretations

Two laboratory interpretations are supplied for the same present effective stress of 90 kPa and OCR of 1.8. For direct simple shear, the illustrative coefficients are S 0.21 and m 0.75. For compression, they are S 0.27 and m 0.70. Compute both strengths and propose a sentence that describes the difference without declaring that the soil has one universally correct undrained strength value. Do not round until the last step.

Answer: 1.8 raised to 0.75 is approximately 1.554; the DSS normalized strength is therefore 0.326 and the predicted absolute strength is about 29.4 kPa. For compression, 1.8 raised to 0.70 is approximately 1.509; the normalized strength is approximately 0.407 and the absolute strength is about 36.7 kPa. A defensible sentence is that different laboratory shearing modes yield different interpreted strengths at the same effective stress and OCR in this illustrative model. Neither value can replace the other without justification.

Transfer exercise three: decide what is not knowable

A report gives a smooth normalized strength line but no individual tests, no specimen quality description, no geological units and no defined failure criterion. It announces S 0.25 and m 0.82. Can you verify the quality of the fitted data from those two coefficients alone? Can you infer a representative project-wide strength profile without the effective stress and OCR profiles? What extra evidence would you request before moving from explanation to engineering use?

Answer: no. The coefficients define a curve but not the quality, representativeness or applicability of the observations underlying it. The strength-versus-depth calculation requires effective vertical stress and OCR at each depth, with geological and testing context. Request original test data, depth logs, stress and pressure interpretation, quality indicators, loading and unloading paths, shearing modes and failure criteria, plus field correlations and project-specific review. This answer tests restraint as much as algebra.

Transfer exercise four: catch an attractive but wrong summary

A simplified article says: “The OCR is higher in the upper layer, so the clay there is certainly stronger than clay beneath it.” At the upper level present effective stress is 50 kPa and OCR is 2.4. At the lower level current stress is 110 kPa and OCR is 1.2. With S 0.22 and m 0.8 for both, determine which has the larger normalized strength and which has the larger absolute strength. Then rewrite the summary in words suited to a careful non-specialist.

Answer: the upper normalized value is 0.22 times 2.4 to the power 0.8, about 0.443, and upper absolute strength is about 22.1 kPa. The lower normalized value is 0.22 times 1.2 to the power 0.8, about 0.255, and lower absolute strength is about 28.0 kPa. The upper layer has the larger normalized ratio, yet the lower has the larger calculated absolute resistance. A better summary is: “Higher OCR can increase strength relative to present effective stress, but the actual predicted strength also depends on that stress.”

Frequently asked questions

Does SHANSEP mean every soft clay follows the same two coefficients? No. The concept is a framework for investigating stress history and normalized behavior, not a universal pair of constants. Different clay units, specimen preparation and shearing modes may demand different relationships. If a published article gives a convenient formula without the data and conditions supporting it, treat it as an example until its provenance has been established.

Is OCR the same as sensitivity? No. OCR compares interpreted past maximum vertical effective stress with present effective stress. Sensitivity commonly compares intact with remoulded undrained strength under specified tests. A sensitive clay can also be overconsolidated, normally consolidated or influenced by aging and structure. The two quantities answer different questions and must not be substituted merely because both are dimensionless.

Why does a normalised strength graph sometimes show more scatter than the fitted line suggests? Natural heterogeneity, uncertain preconsolidation estimates, sampling damage, testing variation and model simplification can contribute. A neat trend line is a drawing of a chosen model, not a guarantee that observations are precise. Always ask how many specimens support it and whether residuals have any meaningful structure.

Can an OCR of one occur at depth? Yes. In the simplified framework, normally consolidated material has present effective vertical stress approximately equal to its maximum past effective vertical stress. Depth alone does not determine OCR. Loading history, depositional environment, stress changes and interpretation method all affect that ratio. A deeper sample can have a smaller OCR than a shallower one while still carrying a larger absolute effective stress.

Is the highest measured undrained strength the safest input? Not necessarily. An engineer must consider relevant deformation mode, data quality, uncertainty and the specific design context. Choosing the largest number because it appears reassuring defeats the purpose of measuring stress history carefully. In education, emphasize what each test represents, and distinguish interpretation from final parameter selection.

What is the simplest one-sentence translation? “SHANSEP relates a clay’s undrained shear strength, normalized by its present effective vertical stress, to its overconsolidation ratio using laboratory-supported parameters for a specified testing mode.” That sentence is longer than a slogan, but it preserves the ratio, material property and qualification. A few precise words prevent a large misunderstanding.

Where to go next

For a useful companion, read marine sediment oedometer consolidation and OCR to see where the stress-history estimate begins. Then read direct simple shear and anisotropy for the distinct strength mode. K0 consolidation and anisotropic stress paths adds another essential piece of the interpretation.

This series is about making difficult material readable without changing its meaning. The How to Translate Easily index collects related guides, while How English Works: Vocabulary and Meaning develops the habit of checking exactly which relationship a technical word names. A useful explanation should leave the reader able to repeat the calculation, question its assumptions and explain its limits.