Offshore CPTu Tip Resistance, Friction Ratio and Dissipation Explained

A CPTu report is a measurement chain, not a soil-name machine

How to translate easily an offshore CPTu report starts by identifying which values are measured and which are calculated. A piezocone is pushed into seabed material while tip force, sleeve friction and pore-water pressure are recorded with depth. A chart can be made to look like a geological section, but the instrument does not directly measure mineral composition, stratigraphic age or an engineering safety factor. Preserve the sensor observations before interpreting what they might mean about soil behaviour.

Did you know that a pore-pressure reading can alter the corrected cone resistance even though it is not an extra push by the operator? The standard area correction accounts for water pressure acting on unequal projected areas within the tip assembly. In a common convention q_t = q_c + (1 − a)u_2. Here q_c is uncorrected tip resistance, a is the calibrated net area ratio, and u_2 is pressure measured behind the tip. Each term is a pressure; units must agree.

Offshore deployment creates a second question: where is the zero of the depth scale? A sounding might be indexed from the seabed, sea surface, a vessel reference, or an external chart datum. The location of the probe below the mudline cannot be inferred safely from an unlabeled ‘depth’ column. Preserve seabed pick, tide correction and any rod or launch offset. Comparing two curves without their vertical references can create a fictitious displaced layer or erase a real one.

The response of a cone is influenced by material stiffness, strength, compressibility, stress history and drainage during penetration. A clay-rich deposit can generate pronounced excess water pressure; some sands drain more rapidly. Transitional silt, cemented carbonate, gas-rich mud or biogenic layers can depart from simple categories. A report should introduce the material context, penetration speed and instrumentation before using the labels ‘clay-like’ or ‘sand-like behaviour’. Those labels are inferences, not direct compositional assays.

Read the instruments and their pressure units

The raw CPTu channels normally include penetration depth, q_c, sleeve friction f_s and u_2. Some datasets also contain tip inclination, temperature, rod force, corrected q_t, derived ratios and soil-behaviour classes. Resist the temptation to begin from a polished classification plot. Make a data dictionary listing each original symbol, unit, sensor location, processing status and reference pressure. Where a vendor uses alternative names, confirm the conversion rather than assuming that similar letters identify the same physical quantity.

Tip resistance q_c is force over a defined cone area. Sleeve friction f_s is force over the sleeve’s contact area, and pore pressure u_2 is measured at the shoulder filter in the standard arrangement. One export may state q_c in MPa and f_s in kPa. For numerical ratios, convert one of them before dividing. If q_t = 2.4 MPa and f_s = 36 kPa, q_t is 2,400 kPa and the simple percentage ratio is 100 × 36/2,400 = 1.50%.

The ordinary friction ratio is commonly written R_f = 100 f_s/q_t, but a normalized friction ratio F_r may use q_t − sigma_v0 as its denominator. That difference is not a typographical detail. Suppose q_t = 2,400 kPa, f_s = 36 kPa and total vertical stress sigma_v0 = 300 kPa. Then ordinary R_f is 1.50%, whereas normalized F_r is about 1.71%. If two reports use the same phrase ‘friction ratio’, inspect the formula before comparing their numbers.

Pore pressure during penetration is not necessarily equilibrium pore pressure. The latter, u_0, is estimated from groundwater or seawater conditions at the relevant reference. The penetration sensor reading u_2 may contain excess pressure associated with rapid deformation and imperfect drainage. A difference Delta u = u_2 − u_0 indicates a pressure deviation relative to the selected baseline. Describe whether a number is absolute, gauge, excess or a derived pressure difference. Those words prevent otherwise plausible but wrong stress calculations.

A successful pore-pressure measurement depends on filter saturation and sensor response. A filter that has lost saturation may register a sluggish, unrealistically flat or distorted pressure history. Negative excess pressure in a particular soil is not automatically an error, and a zero pressure change does not automatically prove sand. Check the pre-test preparation record, zero checks, temperature, penetration rate, interruption history and independent geological evidence. The purpose of quality control is to distinguish material behaviour from instrument behaviour.

The corrected tip resistance is sensitive to the calibration parameter a. Consider q_c = 1.80 MPa, u_2 = 900 kPa and a = 0.80. Convert 900 kPa to 0.900 MPa; the adjustment is 0.20 × 0.900 = 0.180 MPa, giving q_t = 1.980 MPa. If a different net area ratio is later discovered, every dependent metric should be revisited from the retained raw channels. Never edit q_c itself to hide the correction.

Normalize stresses without mixing denominators

Geotechnical interpretation often compares cone resistance with the stresses already acting within the ground. Total vertical stress sigma_v0 depends on overlying material and water reference; equilibrium pore pressure u_0 supports part of it. Effective vertical stress sigma’_v0 is the difference sigma_v0 − u_0 under the stated simplification. If total stress is 350 kPa and equilibrium pressure is 250 kPa, effective stress is 100 kPa. These values must refer to the same elevation and sign convention.

A net cone resistance is q_net = q_t − sigma_v0. A simple form of normalized tip resistance is Q_t = (q_t − sigma_v0)/sigma’_v0. With q_t = 2,477 kPa, total vertical stress 350 kPa and effective stress 100 kPa, the normalized value is 21.27. It is dimensionless. This number is not directly measured by the cone; it inherits uncertainty from both the corrected resistance and the assumed overburden stress model.

An example of normalized pore-pressure ratio is B_q = (u_2 − u_0)/(q_t − sigma_v0). If u_2 = 350 kPa, u_0 = 250 kPa and net resistance is 2,127 kPa, B_q is approximately 0.047. A low positive B_q is not a universal soil classification. If the denominator is small, measurement noise or stress-model error can make the ratio unstable. State the definition and interpret it with the friction ratio, corrected resistance and geological evidence.

Soil-behaviour type charts place empirical combinations of normalized variables into categories. Their value is a structured comparison between soundings, not perfect identification of grains. A core might show shell fragments within a unit whose CPTu classification is ‘silt-like behaviour’. The defensible report states both the laboratory lithology and the interpreted penetration behaviour. Disagreement prompts a material investigation, rather than a decision to erase whichever description is less convenient.

Water depth and soil unit weight are crucial to stress normalization. A shallow offshore deposit under a high water column should not have hydrostatic pressure computed from depth below seabed alone unless the chosen pressure datum allows it. Build a simple stress table by depth, including overlying water when needed and clearly differentiating total, pore and effective pressures. If a sounder’s elevation or seabed reference is ambiguous, provide a provisional range instead of presenting a single normalized index to three decimal places.

Penetration, thin layers and the limits of apparent resolution

A piezocone does not respond at one infinitesimal point in the ground. Its tip and sleeve have finite dimensions, and penetrating a thin hard layer between softer units can create a response influenced by both materials. A sharp peak in q_t does not guarantee a layer of equal thickness to the plotted peak. Similarly, a very thin sand lens in clay might fail to display its full intrinsic tip resistance. Match the sounding to core descriptions and acknowledge the cone’s zone of influence.

An offshore log can contain operational marks at rod additions, pauses, changes in rate or tilt. A pressure step immediately following restart is not automatically a stratigraphic surface. Examine the time series as well as the depth plot, note deployment events and retain raw observations. A polished line that omits interruption records can be less scientifically useful than an imperfect but traceable original graph. Quality flags should identify the reason a data interval was removed or downweighted.

Marine mud may retain delicate natural structure that is disturbed by sampling. An undisturbed-sample laboratory strength can still be affected by unloading, vibration, trimming or storage. The CPTu test takes place in situ but has its own assumptions and correlation uncertainty. Comparing the two is valuable precisely because their limitations differ. A disagreement is not proof that either method is defective; it may signal stress-path differences, fabric, sensitivity, anisotropy or sample damage.

Biogenic silica, shell fragments, carbonate cement, gas, organic matter and laminations all complicate correlation. For example, a shell-rich layer may be locally strong but crush during loading; gas may affect compressibility and pressure response. A site-specific calibration and independently recovered samples can improve interpretation. The reader should see which conclusions rest on general empirical charts and which are supported by material tests from the actual offshore area.

A sounding gives a vertical response profile at one position. Connecting units between widely spaced soundings is a geological hypothesis that must acknowledge distance and seabed morphology. A channel, scour surface or old slide deposit can separate apparently similar profiles. Cross sections should show probe locations and distinguish observed intervals from inferred boundaries. Turning every colour band on adjacent charts into one uninterrupted geotechnical layer creates a deceptive impression of certainty.

Dissipation is a transient pressure test

A dissipation test stops penetration at a selected depth and tracks pore pressure with time. The decay or evolution is a response to relaxation and drainage after the penetration disturbance, not a new direct measurement of permeability or shear strength. Log the stop depth, pressure sensor, sampling frequency, initial reading and how equilibrium pressure was estimated. Some curves are nonmonotonic, so one must not force every result into a simple exponential narrative.

One commonly used metric t50 identifies the time for half of an explicitly defined excess pore pressure to dissipate. If initial pore pressure is 500 kPa and estimated equilibrium pressure is 200 kPa, initial excess is 300 kPa. After half of that excess dissipates, 150 kPa excess remains, giving an absolute sensor pressure of 350 kPa. The target is not half of the absolute initial 500 kPa. Draw the equilibrium baseline on the curve before extracting t50.

A model may connect characteristic dissipation time to a horizontal consolidation coefficient c_h. A generic dimensional structure is c_h = T r²/t, where dimensionless T depends on a particular model, r is an appropriate characteristic length in metres and t is time in seconds. This pattern helps with units but cannot substitute for the actual piezocone geometry and analytical solution. Report the selected time-factor chart and corresponding assumptions, not just the final c_h value.

Dissipation inferred horizontal consolidation behaviour may differ from a vertical laboratory consolidation test because layered soils can be anisotropic. A consolidation coefficient is not identical to hydraulic conductivity; compressibility and fluid properties are involved in any conversion. If a document calls a t50-derived estimate ‘measured permeability’, revise the language to identify the indirect inference and its assumptions. Where no reliable equilibrium state was reached, describe the curve as incomplete instead of inventing a terminal value.

An unexpected dissipation plateau can reflect a real hydraulic boundary, a filter problem, a temporary stoppage or an inadequately long record. Compare with neighbouring depths and laboratory data. A curve may be useful even when it does not meet the ideal model: it might indicate relative drainage rates or motivate additional tests. The correct translation preserves what was actually observed, and separately states the confidence placed in a model-based parameter.

Worked calculations and independent transfer questions

Practice one: q_c = 1.90 MPa, u_2 = 600 kPa and a = 0.75. The corrected tip resistance is 1.90 + 0.25 × 0.600 = 2.050 MPa. Do not add the entire pore pressure or treat 600 kPa as 600 MPa. If a reader cannot reconstruct this elementary calculation, the later classification is not auditable.

Practice two: q_t = 2.40 MPa and f_s = 36 kPa. Using R_f = 100 f_s/q_t gives 1.50%. If total vertical stress is 300 kPa, the normalized F_r with denominator q_t − sigma_v0 becomes 100 × 36/2,100 ≈ 1.71%. The important lesson is to label the denominator, not to memorize one expected percentage.

Practice three: q_t = 2,500 kPa, total vertical stress 400 kPa and equilibrium pore pressure 260 kPa. The effective vertical stress is 140 kPa; net resistance is 2,100 kPa; Q_t using the stated definition is 15.0. Dividing q_t by total stress would give a different index and should not be represented as the defined normalized resistance.

Practice four: a measured u_2 of 680 kPa decays toward an estimated equilibrium 280 kPa. Half of 400 kPa excess pressure is 200 kPa, so the 50%-dissipation absolute reading is 480 kPa. If the test ended at 520 kPa, the half-excess target was not observed. An extrapolated t50 requires a defensible model and must be identified as estimated.

Practice five: in c_h = T r²/t, take T = 0.20, r = 0.018 m and t = 200 seconds. The result is 0.20 × 0.000324/200 = 3.24 × 10⁻⁷ m²/s. This arithmetic checks dimensional consistency only. The real test still needs the correct characteristic radius and time factor from an applicable consolidation interpretation.

Practice six: an exported depth curve uses elevation relative to chart datum while the recovered core uses depth below seabed. A 0.4-m apparent mismatch is not evidence of a geological fault until the seabed elevation, vertical offsets and recovery disturbance are reconciled. Reconstruct both depth axes to one reference and show the residual uncertainty.

Practice seven: a core contains shell-rich silt while the CPTu chart labels that depth ‘clay-like behaviour’. The correct explanation reports an empirical behavioural classification and independent lithological observation. It does not erase the shells or call the cone wrong without evidence. Material identification and mechanical response classification are different questions.

Practice eight: a pressure trace becomes flat after a long instrument pause. Check filter saturation, sensor health, changes in penetration speed and the operation log. A flat curve is not conclusive proof of a freely draining sand horizon. Flag the suspected interval and reconsider any corrected q_t or B_q values that depend on the questionable pressure.

Practice nine: a very low resistance layer is proposed as a potential weak horizon in a seabed slope. The sounding may help guide investigation, but slope failure requires a geometry and loading model, credible shear strength under relevant drainage conditions, groundwater assumptions and an uncertainty assessment. Report the layer as indicated, not as a verified future failure surface.

Practice ten: a consultant says ‘the clay strength was measured by CPTu’. A more accurate statement distinguishes the directly recorded tip and sleeve responses from any inferred undrained strength obtained through an empirical or semi-empirical correlation. Add the correlation and calibration basis. The correction matters most when decisions depend on the numeric strength rather than the illustrative shape of the log.

Questions a careful reader should ask

Does q_t always exceed q_c? A positive u_2 and a net area ratio less than one normally increase the corrected value under the common convention. Different pressure responses and definitions can change the relationship; check the formula and signs rather than assuming.

Can a classification chart identify the exact soil composition? No. It organizes empirical penetration behaviour. Grain-size, mineral, biological and geochemical observations from independent samples are needed for a more complete material description.

Why might two friction ratios disagree? Their denominators, processing versions, tip correction or units may differ. Recalculate from raw q_t, f_s and total stress before concluding that the soil changed.

Does a dissipation t50 equal permeability? No. It is a characteristic time from an observed or modelled pressure response. Converting it into a consolidation or conductivity parameter requires geometry, compressibility and model assumptions.

What makes a CPTu translation trustworthy? The author keeps measured channels separate from corrected values and inferred soil parameters, declares units and datums, checks mathematical examples, reports sensor quality and preserves uncertainty and source records.

What the friction ratio cannot tell you by itself

Suppose a report labels a high friction-ratio interval as organic clay. That label may be plausible in its classification system, yet the mechanical ratio alone does not measure organic carbon or confirm that the soil is clay. It is a pattern of relative sleeve and cone resistance. A core, sample description and laboratory indicators are the appropriate means of checking composition. In a scientific translation, preserve both the empirical ‘clay-like behaviour’ label and independent material observations, and do not silently promote one into the other.

Similarly, a low friction ratio and high tip resistance can suggest granular behaviour, but cementation and crushing complicate a simple density interpretation. A limestone-derived shell sand and a quartz sand may generate different cone responses at the same effective stress. If the client needs relative density, explain which correlation was used and the material population for which it was developed. A ratio of resistance channels is evidence for classification; it is not a universal conversion into a laboratory-determined void ratio or grain composition.

The sleeve and cone respond over different geometries and slightly different depth intervals. When a thin seam appears, dividing an instantaneous sleeve value by a tip response from the same nominal data row can combine signals influenced by different portions of the soil column. A sharp ratio peak may therefore involve genuine geology, sensor geometry or both. Interpret short high-frequency spikes cautiously, especially where sampling does not resolve the seam. Avoid drawing categorical boundaries finer than the practical resolution of the method.

One chart uses normalized resistance and friction ratio to classify behaviour, while another uses pore-pressure information as well. If classifications differ, the report should identify each chart, its input values and the reason for disagreement. It is not appropriate to choose the chart that gives the tidiest cross section without a material argument. A defensible summary may label the interval as transitional and specify the conditions that limit confidence. Uncertainty is not an editorial problem; it is part of the interpretation.

Recovering field meaning from calibration and equipment records

Calibration is more than a document stamped with an expiry date. The cone area ratio, sensor linearity, pressure zero, sleeve calibration and appropriate range determine how raw measurements become engineering quantities. If the instrument’s net area ratio is recorded as 0.78 but processing assumes 0.80, the corrected tip resistance shifts by 0.02 times u_2. At a pore pressure of 1,000 kPa that difference is 20 kPa, which may matter for very low-resistance material. Document the parameter actually used.

The water pressure filter must be appropriately saturated for meaningful dynamic response. In fine sediment, pore pressure can change rapidly during penetration and more slowly during dissipation, so a delayed or incompletely saturated filter can distort both stages. A plausible-looking u_2 line is not enough to demonstrate good data. Inspect the full preparation and zero-return record, and treat improbable straight runs or steps as investigations. If a result is qualified, identify which derived ratios inherit the problem rather than dismissing the entire sounding.

A pressure measurement may also be altered by temperature changes or sensor drift. The vessel deck, water column and seabed can impose different thermal conditions. Checking a baseline after the sounding can reveal drift that was invisible on the final formatted profile. A report should separate original acquisition values from documented corrections. If interpolation or trimming was performed to remove a known instrumentation artefact, retain the interval limits, procedure and reason. A defensible correction improves traceability rather than concealing what happened.

Field records should show the penetration speed and pauses. Conventional CPT procedures typically use a controlled push rate, but offshore operations may encounter variations from platform motion or capacity limits. Fast and slow penetration can alter drainage response in transitional soils, and a long pause can create a separate pore-pressure relaxation event. Do not treat the depth axis alone as sufficient metadata. Every observation occurred at a time and under an operational state, and those conditions may explain features that appear geological.

A laboratory comparison that respects stress paths

Engineers often estimate undrained shear strength from net cone resistance using a cone factor, schematically s_u = q_net/N_kt. This is an interpretation requiring an appropriate cone factor, not a direct pressure sensor reading of shear strength. If q_net is 600 kPa and the selected project-calibrated factor is 15, the illustrative estimate is 40 kPa. A different credible factor would change that estimate, so the report must state its source and uncertainty. Do not imply that the numerical result is unique.

Compare a CPT-derived undrained strength with a field vane test only after examining strain rate, anisotropy, remoulding and correction conventions. The vane mobilises resistance through rotational shear surfaces, while a cone creates a three-dimensional penetration mechanism. Both are useful but not interchangeable as raw physical measurements. If vane strength and q_net/N_kt disagree, the question is which test and correction best represent the engineering mechanism at issue. Averaging them without understanding their differences is not a repair.

A laboratory oedometer may indicate compressibility or an apparent preconsolidation stress, but the CPTu stress history interpretation still needs calibration. Overconsolidation can affect pore-pressure response and tip resistance, yet an empirical cone-based estimate of overconsolidation ratio is not identical to the history directly observed in a controlled sample. Organic, diatom-rich or cemented sediments can complicate both methods. Preserve the phrases ‘estimated from CPTu’ and ‘inferred from laboratory consolidation’ as distinct evidence streams.

A useful interpretation chart contains field and laboratory points at matching depth references. State whether a sample depth is the recovery location, its centre or the nominal drilling interval; state whether the CPT depth is below mudline. Align both before correlating strength. Sample disturbance, anisotropy and natural variability may explain differences after registration. A graph that draws a smooth best-fit line through dissimilar material types can conceal the very exceptions the engineering decision must understand.

Dissipation interpretation without a magic formula

Imagine a curve that begins at u_2 = 760 kPa and approaches a measured equilibrium value near 360 kPa. The excess pressure at start is 400 kPa, so a 50% excess-pressure target is 560 kPa. If the reported t50 is taken where the curve crosses 380 kPa, the analyst probably halved the absolute reading rather than the pressure difference. The correction is conceptual, not just arithmetic: define the baseline and excess before measuring any fraction of the response.

When a dissipation curve starts below the estimated equilibrium pressure, a conventional ‘decay from a positive excess’ description may not fit. Pressure may rise toward equilibrium instead. Such behaviour can follow stress history and dilation, though measurement problems must also be considered. Do not force a negative initial excess into a positive-pressure decay formula without checking the selected interpretation. Describe the sign and shape before deriving a time factor. The model must describe the observed process, not reshape the process to suit a memorized graph.

Dissipation times are not intrinsic units of permeability. A test may have a short t50 because of a more permeable soil, different compressibility, geometric influence or an instrumental effect. A coefficient of consolidation incorporates mechanical and hydraulic behaviour, while hydraulic conductivity has different units and requires a compressibility relationship for inference. If a client requests a conductivity estimate, explain what extra measurements and model assumptions are needed. A time measurement alone cannot determine a unique conductivity across every material.

Time-series sampling matters. Early pressure changes can be rapid; late measurements may approach the limit of sensor drift. A logarithmic display makes a wide range of times convenient but can hide how many actual readings informed the apparent early trend. When estimating t50, use the recorded samples and document interpolation between them. If the test ends before the target crossing, the t50 was not observed. A report may offer a modelled range, but must not present it as a directly read time.

More independent numerical and reasoning exercises

Exercise eleven: the recorded q_c is 0.80 MPa, the sensor pressure is 1.20 MPa and the calibrated net area ratio is 0.76. What is q_t? The correction is (1 − 0.76) × 1.20 = 0.288 MPa, giving 1.088 MPa. This illustrates why a large pore-pressure correction can be consequential in soft marine deposits. If the instrument is questionable, the corrected value inherits that uncertainty rather than becoming more authoritative simply because it is computed.

Exercise twelve: q_t is 1,800 kPa, sleeve friction 27 kPa and total stress 300 kPa. The ordinary R_f equals 1.50%, while normalized F_r with denominator q_t − sigma_v0 equals 1.80%. A table that labels both merely ‘friction ratio’ creates a 20% relative difference in the percentage value. The solution is not to choose an average; it is to retain the distinct definitions and use each with its own classification convention.

Exercise thirteen: a soil layer has estimated total vertical stress 510 kPa and pore pressure 390 kPa. Effective stress is 120 kPa. If an export uses 390 MPa, the calculation produces a physically implausible result for this context. The repair is a source-unit audit, not selective editing of the output to force positive values. Every pressure series should include a unit label and a range check before derived normalized resistance is trusted.

Exercise fourteen: a sounding’s u_2 is 520 kPa, its equilibrium u_0 is 400 kPa, corrected q_t is 2,200 kPa and total stress is 500 kPa. B_q under the stated convention is 120/1,700 ≈ 0.071. If a colleague uses 120/2,200 instead, they have not used net tip resistance. This is an ideal audit question because the calculation is simple while the denominator choice is conceptually important.

Exercise fifteen: a 10-mm-thick sand seam appears as a 100-mm-wide q_t anomaly on a plotting system. Should the seam be redrawn as 100 mm thick? No. Penetration influence zones, sensor spacing, recording interval and the plotting algorithm may spread the response. The independent core should guide the seam thickness. State the sounding’s mechanical response interval and geological layer thickness separately rather than claiming they are identical.

Exercise sixteen: a field operator reports no stable equilibrium before a dissipation test ended. Can the analyst publish a precise t50 from the same run? Not as a directly observed half-excess time unless the required baseline and crossing are established. A model-dependent extrapolation can be considered, but it needs its assumed asymptote, fit, uncertainty and sensitivity clearly marked. The distinction protects the integrity of both the observation and the engineering inference.

Writing the final interpreted report for real readers

A readable results section can follow the reasoning in the same order as the measurement: identify location and depth datum, describe deployment and instrument checks, show raw q_c, f_s and u_2, apply area corrections, estimate total and effective stresses, compute clearly named ratios, and only then discuss material behaviour. The sequence prevents a classification label from becoming a substitute for a measured value. Even a reader unfamiliar with CPTu can follow the chain by asking which assumption each new quantity introduces.

Where one parameter is used in a foundation or slope model, communicate the transfer explicitly. For example, a CPTu estimate of undrained strength may be selected after calibration, with a range reflecting material variability and cone-factor uncertainty. A slope analysis then requires load cases, boundary conditions, geometry and drainage. The original sounding does not by itself pronounce the final design safe. Distinguish evidence, engineering estimate and project decision in separate sentences rather than mixing them in a single confident headline.

An internal review should independently recompute at least one row of each derived column, test units at both ends of the profile, inspect plausible continuity of stress with depth, and confirm that each colour legend names its classification method. The report should document excluded data and operational pauses. When a source is translated across languages, both the original labels and the target-language explanations should be consistent. Transparent arithmetic and terminology outperform decorative precision.

The most valuable translation is often the one that tells the reader precisely what is not known. Without saturation records, an abnormal pore-pressure trace may remain suspect. Without cores, soil behaviour cannot be equated with exact mineral composition. Without a pressure baseline, t50 may be uncertain. Without field geometry, a weak layer does not prove failure. These limitations do not make the test useless; they define how responsibly its results can be used.