Why a pipe can be stable yet vulnerable
Did you know that a submarine pipeline can be heavy enough to remain on the seabed in a current and still suffer fatigue when a short section hangs free above a seabed depression? These are different engineering questions. On-bottom stability concerns sliding or movement on the bed. Free-span assessment asks whether an unsupported length bends under its weight and vibrates under passing water. A useful translation of a technical report must identify that distinction before explaining any number. Neither a successful sliding calculation nor a photograph of a line touching the seabed elsewhere proves that every suspended span is acceptable.
What engineers mean by a free span
A free span is a length of pipeline with insufficient support from the seabed across part of its route. The gap may follow an uneven seabed, a scour feature, a crossing, an installation condition or later changes in sediment. A survey may report the unsupported length, gap height, position and neighboring contacts. Those values need definitions: where exactly does the span begin and end, and were the endpoints inferred from bathymetry, sonar or physical inspection? Treat an observed geometry and a later structural idealisation as related but nonidentical records. The model must tell the reader which supports it assumed.
Why the applicable method matters
DNV-RP-F105, Free Spanning Pipelines, is a first-party recommended practice for assessing in-line and cross-flow vortex-induced vibration, direct wave loading, ultimate limit state and fatigue limit state. The public DNV page identifies an edition dated December 2025. A project may contractually use another revision, so a report should name its governing edition instead of borrowing a criterion from whichever summary is convenient. Our examples teach how to decode engineering language and arithmetic, not how to certify a pipeline. Design approval requires the governing standard, validated inputs, engineering models and independent checking appropriate to the project.
Begin with the pipeline itself
Write down external diameter, wall thickness, steel grade, coating and any contents that affect mass. Keep the external diameter used in hydrodynamic expressions separate from the steel diameter used in stiffness or stress calculations. A concrete coating adds mass and changes the outside hydrodynamic diameter; a fluid-filled pipe differs from an empty pipe in inertia. Corrosion allowance changes net wall thickness assumptions, while stiff coatings may or may not be credited in flexural rigidity. A clear report states which contributions are included. Never substitute a nominal size label for an actual modeled outer diameter without checking the datasheet.
The span length is not every survey distance
A chart may show kilometers along a pipeline, coordinates for the edge of a depression, and the length that actually lacks contact. Those are different measurements. The free-span length used in a structural model is governed by contact and boundary representation, not simply by the distance between two convenient survey waypoints. Seabed stiffness can shift effective support positions. In a report, translate chainage into positions, identify span shoulders and specify the assumed length for each load case. If a table gives only a starting chainage and a maximum gap, ask for the missing endpoint rather than inventing the span length.
Understand gap height
Gap height tells us how far the pipeline underside lies above the local bed at a specified location. It is not the same as water depth, free-span length or clearance to another asset. A high gap can expose more of the circumference to flow, while very small gaps change local hydrodynamics and may experience changing contact in waves. State whether the figure shows maximum, mean, midspan or survey-registered clearance. These quantities can influence modelling assumptions and intervention decisions, but none can stand alone as an all-purpose pass or fail rule. A measured zero clearance also deserves a statement of survey resolution.
Seabed contact is a spring, not a verbal promise
The seabed is not usually an infinitely rigid clamp. Contact conditions depend on soil response, pipe embedment, local geometry and loading history. A model may use discrete springs, continuum soil support or a more elaborate nonlinear representation. If a report calls a support ‘fixed’, determine whether that is a computational idealisation or a demonstrated physical boundary. A short free span with soft shoulders may have a different natural frequency than the same geometric length between stiff supports. Translating ‘supported’ as ‘fully restrained’ can overstate actual stiffness and understate deflection or fatigue demand.
Separate three loads
A suspended pipeline experiences self-weight and buoyancy, hydrodynamic forces from waves and currents, and operational or installation loads such as thermal expansion, axial tension or pressure-related effects. Those actions may combine, but they should not be described as a single generic ‘sea load’. Gravity determines sag and contact tendencies. Steady current can generate vortex shedding. Oscillatory wave flow can drive time-dependent forcing, sometimes at frequencies unlike current-driven shedding. Thermal and axial effects may modify effective tension and natural frequencies. The analysis depends on which operating and environmental state is actually being evaluated.
Define in-line and cross-flow motion
In-line means motion generally along the local direction of water flow; cross-flow means transverse motion. These directions are relative to the current or wave velocity at the pipe, not necessarily north–south or horizontal–vertical on a map. Vortex-induced vibration can have responses in either direction. A cross-flow amplitude written in pipe diameters cannot be converted into millimetres until the chosen diameter is known. A field report should identify the axes, the sign convention and where displacement is measured. Without those definitions, a colorful motion plot can be visually impressive but mechanically ambiguous.
Where vortex shedding enters
Flow past a bluff cylinder can shed vortices alternately or in more complicated patterns, depending on the Reynolds number, proximity to the seabed, flow history and motion of the cylinder. Fluctuating forces can excite structural vibration. A common elementary scale is shedding frequency f_s approximately equal to St multiplied by U divided by D. Here St is a Strouhal number, U a characteristic velocity and D a reference outside diameter. This is a scale check, not a design response model. For oscillatory waves, combined waves and current, nonuniform velocity or moving structures, direct substitution into this relation is not sufficient.
Define frequency before comparing it
Natural frequency refers to a property of the modeled structural system in a defined configuration, including effective mass and support assumptions. Forcing frequency describes a component of the hydrodynamic environment. A vibration response can include modal coupling and nonlinear fluid–structure interaction, so a single equality is not a universal resonance verdict. Always record whether the value is in hertz, radians per second or cycles per minute. One hertz means one cycle per second, while angular frequency omega equals two pi times ordinary frequency. This simple unit distinction prevents serious mistakes when a model uses both notations.
First worked example: shedding scale
Imagine a hypothetical pipe with external diameter 0.40 m in a steady characteristic current of 0.60 m/s. For an illustrative Strouhal number of 0.20, the simple shedding estimate is f_s = 0.20 × 0.60 / 0.40 = 0.30 Hz. That is one illustrative vortex-shedding timescale, not a prediction of actual response amplitude, fatigue life or compliance with DNV-RP-F105. The unit cancellation is (m/s) divided by m, leaving inverse seconds. If a colleague reports 0.30 m/s as the resulting frequency, the calculation has been translated into the wrong physical quantity and must be corrected.
Second worked example: diameter sensitivity
Hold the illustrative current at 0.60 m/s and Strouhal number at 0.20. If the relevant hydrodynamic diameter increases from 0.40 m to 0.50 m, the simple frequency scale becomes 0.20 × 0.60 / 0.50 = 0.24 Hz. The calculation illustrates an inverse relationship, not a rule that thicker coatings always reduce fatigue. Added coating also changes mass, damping, gap geometry and response. A defensible comparison would recompute the full hydrodynamic and structural model for each diameter. Never transfer the reduced frequency alone into a claim about improved design safety.
Third worked example: current speed
For the same hypothetical 0.40 m cylinder and illustrative St = 0.20, changing U from 0.60 m/s to 0.90 m/s changes the simple shedding scale from 0.30 Hz to 0.45 Hz. Velocity matters, but sea currents vary in magnitude and direction with depth and time. A report should explain whether the input is a depth-averaged value, seabed-adjacent profile value, extreme current or statistical exposure bin. Fatigue analysis needs the distribution of conditions, not just a maximum. An extreme-condition point cannot be silently treated as every hour of a service lifetime.
Reduced velocity: a diagnostic ratio
Reduced velocity often appears as U_r = U/(f_n D) under a stated convention. It is dimensionless because the denominator has velocity units. The ratio compares flow speed with a structural timescale. For U = 0.60 m/s, f_n = 0.45 Hz and D = 0.40 m, U_r = 0.60/(0.45 × 0.40) = 3.33. This number is not a universal pass or fail boundary. The relevant response regime depends on the model, in-line versus cross-flow behavior, damping, Reynolds effects and flow conditions. A report must say which criterion it applies.
What a mode shape contributes
A suspended pipeline can vibrate in multiple modes, each with its own shape and frequency. The first vertical mode may resemble a single broad arc, but actual configurations can involve lateral response, axial coupling and interacting adjacent spans. A table that lists the first natural frequency alone may omit a higher mode with meaningful exposure. Ask which modes were retained, how they were extracted and how boundary conditions were verified. Modal frequency is not a directly observed universal property of ‘the span’; it is a result of an identified model with particular contents, soil support and tension assumptions.
The simple beam check is only a check
For a highly idealized uniform simply supported beam in bending, a familiar first-mode scale is proportional to sqrt(EI/m)/L squared, where E is elastic modulus, I is section second moment, m is mass per unit length and L is unsupported length. Real subsea spans require additional considerations including axial force, added mass, soil contact and hydrodynamic forces. The proportionality is useful for interpreting sensitivity: a longer span can lower a bending-dominated natural frequency sharply. It cannot independently calculate an approved fatigue life. If a design report uses a more complete model, identify its assumptions instead of replacing it with a classroom formula.
Fourth worked example: length sensitivity
Imagine two otherwise identical ideal bending-dominated beam models with unsupported lengths 20 m and 25 m. Under the simplified inverse-square length scaling, their frequency ratio f_25/f_20 is (20/25)^2 = 0.64. If the 20 m idealization had a hypothetical 0.50 Hz first-mode frequency, the 25 m idealization would suggest roughly 0.32 Hz. That is a directional sensitivity check, not a site-specific modal calculation. A real length change may also move contact locations and soil stiffness, so the original assumptions must be revisited before using a frequency ratio in an engineering decision.
Fifth worked example: stiffness
For a simplified model in which other terms stay fixed, frequency scales with the square root of bending stiffness. If effective EI were multiplied by 1.44, the estimated frequency multiplier would be sqrt(1.44) = 1.20. This does not justify assuming a coating or grout increases EI by a stated amount; that contribution must be verified structurally. Nor does a higher frequency automatically mean less fatigue under all current distributions. What matters is the relationship between excitation, response, stress range and exposure. A report that jumps directly from ‘stiffer’ to ‘safe’ is missing those connecting steps.
Sixth worked example: mass
Under that same simplified bending-dominated scale, multiplying effective vibrating mass per length by 1.21 while holding stiffness and span length fixed would reduce the estimated frequency by 1/sqrt(1.21), approximately 0.909. A hypothetical 0.44 Hz becomes about 0.40 Hz. The important distinction is between structural mass, internal fluid mass and hydrodynamic added mass. Some components vary across operating and flow conditions. Describe which mass definition the model actually uses, and avoid mixing weight per length measured in newtons per metre with mass per length measured in kilograms per metre.
What added mass really signals
A vibrating object in water accelerates surrounding fluid as well as its own materials. Engineers describe that inertial contribution through added mass, among other formulations. It is not an extra permanent layer physically attached to the pipe. A coefficient used for added mass has a chosen reference geometry and flow regime, and it can differ between directions and models. If a spreadsheet adds water volume twice, once as displaced fluid mass and again as the full added-mass term, the interpretation may be wrong. A translator should name each contribution and determine how the governing formulation defines it.
Damping is not one universal percentage
Energy is dissipated through structural mechanisms, contact with soil and interaction with the surrounding fluid. A report may group these contributions in an effective damping parameter, but that value may depend on response amplitude and direction. It should not be quoted without identifying the convention and the analysis stage. More damping can reduce some vibration responses, yet assigning an optimistic value without evidence can obscure a fatigue risk. When comparing two revisions, see whether the changed damping came from measurements, calibrated soil behavior or a modeling assumption. Do not describe the change as an observed improvement unless the evidence supports that claim.
Different consequences: ULS and FLS
Ultimate limit state, or ULS, concerns severe loading and strength-related consequences. Fatigue limit state, or FLS, concerns damage accumulated through repeated stress cycles. A pipeline can pass a maximum-stress check and still fail a fatigue acceptance criterion because many smaller cycles accumulate damage. A free-span report should therefore make separate statements about both, with their own load cases, factors and criteria. The most concise faithful explanation is not ‘the span passes’ but ‘the assessed configurations satisfy the specified ULS and FLS checks under the listed assumptions,’ if both have actually been demonstrated.
How stress range enters fatigue
Fatigue assessment begins with cycles of stress variation at relevant locations, including welds and features governed by appropriate fatigue classifications. Stress range is the difference between maximum and minimum stress within a defined cycle, not necessarily the maximum absolute stress. A response may have multiple frequencies and environmental bins, creating a spectrum of stress ranges. A report must say how structural motion becomes local stress and how counts are assigned. Using a single peak bending stress as though it were a lifetime fatigue spectrum can badly misrepresent damage. Detailed methods and appropriate S–N data must follow the governing design basis.
Cycle count, worked example
Suppose a hypothetical measured oscillation can be approximated as 0.45 cycles per second over one uninterrupted hour. The corresponding cycle count is 0.45 × 3,600 = 1,620 cycles. If the same state occurred for 10 hours, it would represent approximately 16,200 cycles. Neither calculation supplies stress amplitude, material fatigue resistance, weld classification or the frequency of this sea state over a design life. It is a unit-checking illustration only. A published fatigue conclusion requires more information. Do not multiply a transient laboratory frequency by the entire operational lifetime without a justified exposure model.
Miner’s rule, worked example
A common linear fatigue damage accounting form is D = sum(n_i/N_i), where n_i counts applied cycles in bin i and N_i is the resistance expressed as cycles to failure at that bin’s stress range under the selected fatigue curve. For a hypothetical pair of bins, 200,000/1,000,000 contributes 0.20 and 300,000/2,000,000 contributes 0.15. The sum is D = 0.35. This arithmetic shows the accounting principle, not an acceptance verdict; allowable damage, safety factors, design fatigue life and conservatism depend on the applicable criterion and project basis.
Do not invert the damage ratio
It is common for readers to confuse remaining fatigue life, usage fraction, safety factor and the damage sum. With the hypothetical D = 0.35, it would be wrong to announce a ’65-year remaining life’ unless the exposure and design-life basis happen to justify that conversion, which they usually do not. A damage value of 0.35 means that the specified bins sum to 0.35 under the chosen linear model. A statement about years requires an explicit exposure duration, damage rate and assumptions about future operating conditions. The symbols carry meaning only with a defined denominator.
Environmental bins are not optional decoration
The ocean does not apply one constant velocity forever. Wave height, period, current speed, current direction and their joint distributions determine exposure. A fatigue calculation may divide environmental conditions into bins with associated durations or probabilities. A reader should ask whether the dataset spans relevant seasons, whether currents align with the pipe axis and whether directional combinations are preserved. An averaged flow speed cannot usually replace the distribution because response and fatigue are nonlinear. Even if an average state seems moderate, a narrower set of conditions may align more strongly with a vibratory mode and contribute disproportionate damage.
Steady current and wave motion are different inputs
Current-driven vortex shedding can be approximately characterized through a current speed, while wave-induced flow is oscillatory and varies over a wave cycle. Combined wave and current loading can change both the velocity history and response regime. A report using a steady-flow screening relation should not imply it evaluated the full time-dependent wave case unless that analysis was also performed. Similarly, a significant wave height is not a near-bed orbital velocity. Translate wave statistics into the hydrodynamic quantity actually modeled, documenting the transformation, depth, period and applicable wave theory where required.
What the Keulegan–Carpenter number tells you
In oscillatory-flow problems, the Keulegan–Carpenter number is often formed as KC = U_m T/D, using a characteristic oscillatory velocity amplitude, period and diameter under an explicit convention. It compares the fluid excursion scale during a cycle with object diameter. For hypothetical U_m = 0.25 m/s, T = 8 s and D = 0.50 m, KC = 4.0. That result by itself is not a stability or fatigue acceptance rule. It helps identify the regime relevant to a hydrodynamic model. Check whether the report uses a different velocity convention before comparing numbers from separate sources.
Seventh worked example: angular frequency
An oscillation at f = 0.50 Hz has angular frequency omega = 2πf, about 3.142 radians per second. Conversely, an angular frequency of 1.57 radians per second corresponds to f = 1.57/(2π), about 0.25 Hz. These conversions matter when inputs appear inside equations for acceleration, dynamic stiffness or sinusoidal waves. One cannot simply paste 3.142 into a table labeled hertz. A report should indicate whether a velocity time series is described using ordinary frequency or angular frequency and whether phase is in radians or degrees. Small notation errors can propagate into major apparent disagreements.
Local current direction versus pipeline bearing
Flow orientation relative to the pipe controls the cross-flow component. If a horizontal current has magnitude U and makes angle theta with the pipeline axis, a simple horizontal decomposition gives perpendicular component U_perp = U sin(theta). With U = 0.80 m/s and theta = 30 degrees, the perpendicular component is 0.40 m/s. The along-pipe component is about 0.693 m/s. This vector calculation is not a replacement for three-dimensional hydrodynamics or a specific DNV response model. It prevents a basic mistake: treating the full current speed as normal flow without checking the route bearing.
Eighth worked example: orthogonal directions
Consider a hypothetical pipeline running east–west with a 0.70 m/s current toward the north. In the ideal horizontal map view, the current is perpendicular to the pipeline. If the same current instead travels parallel to the route, its horizontal cross-pipe component is zero. Real local flow around a curved or inclined seabed profile is more complicated, and waves may have a different direction from the current. The exercise shows why a report needs clear azimuth conventions. Saying ‘the flow is 0.70’ without units, bearing or vertical profile is not enough to characterize hydrodynamic forcing.
How a span survey becomes an analysis input
A survey commonly supplies route chainage, seafloor profile and free-span dimensions. Analysts then translate those observations into structural geometry, supports and environmental exposure. Each transformation has uncertainty. Sonar returns may have registration error, sediment cover may hide contact and a model may smooth the pipe profile. An audit trail should show the survey date, software processing or interpretation assumptions, the span identifier and the final model geometry. It is not enough to claim that ‘the model comes from survey data’ when the decision depends on a gap that is near the survey’s measurement resolution.
Ninth worked example: chainage
A surveyed unsupported region begins at chainage 12.480 km and ends at 12.506 km. A straightforward along-route separation is (12.506 − 12.480) km = 0.026 km = 26 m. That arithmetic does not prove the structural free-span length equals 26 m: the route coordinates, bend geometry, seabed contact and definition of endpoints may require adjustment. A reader can still use the calculation as a reconciliation check between survey tables. If a model reports 31 m with no explanation, the difference is a question to resolve rather than a license to alter the recorded data silently.
Tenth worked example: gap conversion
A pipeline inspection table lists maximum underside clearance as 180 mm. Dividing by 1,000 converts that to 0.180 m. If the external hydrodynamic diameter is 0.450 m, an illustrative gap-to-diameter ratio is 0.180/0.450 = 0.40. The ratio is dimensionless. It says nothing by itself about whether the span passes a fatigue check; near-bed hydrodynamics, actual flow, pipeline motion and support stiffness are still required. The reader’s task is to translate three columns into one defined quantity without mistaking clearance for total water depth.
When neighboring spans interact
A pipe may touch down briefly between two unsupported regions. That contact can couple the vibration or load redistribution of adjacent spans, especially if the intervening support is soft or intermittent. Do not automatically analyze every gap as an isolated simple beam. DNV-RP-F105 covers approaches to free-span analysis and structural/modal response; a project model must identify when multi-span interaction matters. The word ‘adjacent’ is a geometric statement; the phrase ‘independent’ is a modeling conclusion that needs support. Ask whether common modes, contact nonlinearity and relevant soil springs were considered before comparing independently calculated fatigue lives.
What operating condition does to the span
An empty line during installation, a seawater-flooded line during precommissioning and an operational line containing a different fluid may have different effective masses and axial force states. Temperature and pressure can change longitudinal loading and contact. A report should state the design phase for each assessment: as-laid, flooded, hydrotest, operating, shutdown or other specified condition. The same physical span might pass under one condition and require further analysis under another. Never borrow an installation acceptance statement to summarize a decades-long operational assessment without reviewing the actual modeled cases.
Eleventh worked example: internal contents
Take a hypothetical internal circular diameter of 0.30 m. The internal volume per metre is π(0.30)^2/4 × 1 m, approximately 0.0707 m³ per metre of length. If filled with a fluid of density 1,000 kg/m³, the contained mass is approximately 70.7 kg per metre. The arithmetic is a geometry-and-units check, not a complete effective mass calculation. Pipeline steel, coating, buoyancy-related definitions and fluid added mass still require separate treatment. A spreadsheet that uses 0.0707 as kilograms per metre has lost a density factor and will produce misleading dynamics.
Why direct wave loading must be separated
Direct wave forces can create bending stress without the same resonance mechanism as vortex-induced vibration. A report may assess direct wave response with an appropriate force model and may assess VIV with a response model. The relevant method depends on flow conditions and behavior; DNV-RP-F105 distinguishes these issues. A statement that ‘VIV does not occur’ cannot be stretched into ‘the span receives no dynamic wave stress’. Likewise, a direct-load stress calculation does not by itself establish a fatigue result for possible VIV excitation. Demand a clear map from each load mechanism to each evaluated limit state.
Fatigue location: the weld is not incidental
A pipeline span may experience large bending moment at certain positions, but fatigue resistance at a weld or structural detail also depends on a suitable stress concentration or classification and relevant fabrication quality. The highest geometric deflection does not automatically identify the controlling fatigue location. A reader should distinguish modal displacement plots from derived stress ranges and from damage tables at actual checking points. If a result is described as ‘maximum fatigue near the shoulder’, identify the coordinate, weld or detail considered. Otherwise, the sentence may hide a mismatch between a global response result and a local fatigue criterion.
Twelfth worked example: percentage exposure
Suppose a hypothetical fatigue exposure table assigns a flow bin 15 percent of a 20-year operational period. Under a deliberately simplified uniform time convention, its nominal duration is 3 years. That alone is not three years of continuous VIV because excitation may be conditional on wave direction, operating state and response threshold within the bin. Report both the probability or duration definition and the modeled stress response for that bin. If the report combines mutually dependent environmental conditions, a naïve multiplication of their individual probabilities can be wrong. The joint distribution needs to be stated.
How to read a screening table
A screening table may list spans that pass simplified avoidance checks and spans needing detailed analysis. ‘Screened out’ can mean ‘no further study under the chosen conservative criterion’, but the phrase can also mean an item was excluded for another reason. It is essential to read the column heading and method note. Failing a screening criterion is not equivalent to physical failure; it may simply trigger a more detailed response or fatigue assessment. Passing a simplified test is also conditional on its scope and input validity. Translate the administrative disposition and the physical interpretation as separate statements.
How to read an assessment envelope
Designers may compare several flow directions, velocity bins, operating states and support assumptions and then report an envelope or governing case. The envelope is a maximum or controlling outcome over the assessed cases, not a new measurement. When a table says ‘governing’, ask what quantity was maximized: stress, amplitude, fatigue usage, moment or another metric. Different cases can govern different metrics. Reporting that one sea state governs all failure modes without support collapses a multi-dimensional assessment into an unwarranted single story. A careful explanation names the controlling case next to the corresponding limit state.
Thirteenth worked example: converting a life basis
Suppose a hypothetical analysis estimates annual fatigue damage of 0.012 under a stationary exposure model. Over 25 years, simple linear accumulation gives 25 × 0.012 = 0.30. If the rate had been 0.020 instead, the same assumption gives 0.50. This is a numerical demonstration of a constant-damage-rate approximation, not an approved predicted life. Real service may involve changing seabed geometry, corrosion, interventions or operating profiles. To compare with an acceptance limit, use the project-design factors and chosen fatigue criteria, and disclose the assumptions that make linear extrapolation meaningful.
Fourteenth worked example: separating demand from capacity
A hypothetical fatigue table lists calculated cumulative usage 0.40 and a project-defined allowable usage 0.50. Under that explicitly stated acceptance rule, 0.40 is below 0.50. But calling the spare amount ’20 percent reserve’ without defining the denominator is ambiguous: the absolute margin is 0.10, and the ratio (0.50−0.40)/0.50 is 20 percent of the allowance. That margin is not automatically an engineering safety factor or reliable remaining life. A translation should preserve the exact comparison and not invent a more reassuring concept than the calculation supports.
Fifteenth worked example: ratio traps
A report says two spans have fatigue damage estimates of 0.08 and 0.12 for the same defined exposure basis. The second estimate is 50 percent greater than the first because (0.12−0.08)/0.08 = 0.50. The absolute difference is 0.04. Neither figure proves the second span is unsafe, because no acceptance threshold or relevant uncertainty has been provided. This distinction matters in engineering communication. Readers often confuse a large percentage relative change with an adverse absolute limit-state result. Good writing states the baseline, the unitless damage definition and any acceptance limit in the same paragraph.
When a pipe moves and the geometry changes
Free spans are not necessarily static over the asset life. Local scour can lengthen a gap, sediment deposition can improve contact, and pipe motion can modify soil support. A survey date and a design assumption date therefore matter. If an analysis uses the route as surveyed in a previous campaign, the report should say so and explain whether extrapolation is justified. Monitoring or inspection may be needed where the seabed is evolving. Avoid the claim that a single acceptable model is permanently valid. The relevant question is how the actual geometry could change before the next inspection or intervention.
The cost of mistaking measurement for prediction
An ROV or acoustic survey can measure or estimate clearance and geometry at inspection time. Modal frequency and fatigue usage are often outputs of models, sometimes supported by sensors, not directly read from a photograph. A sentence such as ‘the camera showed a fatigue life of 30 years’ is a category error. A camera might show a gap; a computational model used that gap with environmental, material and support assumptions to estimate a life measure. Separate observation, inference, model prediction and design acceptance. Doing so gives the reader a transparent chain of evidence rather than an artificially certain result.
Reading a sensor result
Strain, acceleration, displacement or vibration sensors may inform free-span assessment, subject to calibration and representativeness. Sensor location, sampling frequency, noise, drift and operating conditions influence interpretation. A reported vibration peak does not immediately establish its source or fatigue significance. Compare observed frequencies with modeled modal properties and environmental conditions; determine whether the sensors can capture relevant modes. Ask whether the monitoring interval covers the conditions that control lifetime damage. The DNV recommended practice includes guidance on sensor technology, but no single sensor trace can establish every future condition without a defensible interpretation and uncertainty framework.
Sixteenth worked example: sampling rate
Suppose a sensor records at 10 samples per second. Its sampling interval is 0.10 s and the nominal Nyquist frequency is 5 Hz. Signals above that limit can alias into misleading lower frequencies if not appropriately filtered. A target vibration near 0.6 Hz lies below 5 Hz, but the sampling rate alone does not guarantee calibrated amplitude, correct axis, synchronization or sufficient duration. A report must distinguish data-acquisition capacity from validated signal quality. If the sensor was mounted at a vibration node, a low measured amplitude there might not represent the maximum amplitude elsewhere on the span.
Measurement uncertainty as an input
Every reported span geometry has limits: sonar resolution, coordinate uncertainty, bed roughness and changing contact make exact boundaries difficult. Environmental input also has statistical and measurement uncertainty. Models add uncertainty in soil springs, added mass, hydrodynamic response and fatigue data. Rather than combining every issue into a vague ‘conservative’ label, a clear analysis distinguishes what was measured, what was assumed and what was varied in sensitivity studies. It is possible for one parameter to be conservative for one failure mode but not for another. Explain the direction and basis of each claimed conservative assumption.
Seventeenth worked example: survey uncertainty
A hypothetical survey estimates unsupported length as 24 m with an uncertainty interval of plus or minus 1 m. The endpoints 23 m and 25 m are not confidence bounds unless the source explicitly defines them statistically. Under ideal inverse-square bending frequency scaling, the frequency ratio between a 25 m and 23 m length is (23/25)^2, approximately 0.846. This shows why seemingly small geometry uncertainty can affect modal predictions. It does not mean the real span has a precisely 15.4 percent frequency uncertainty; that would require the complete model and a justified mapping from measurement error to response.
An intervention is a change to the evidence
Interventions may include supports, rock placement, local span correction or other measures chosen for the site. They can alter contact stiffness, gap geometry, current exposure and maintenance access. A report proposing an intervention must say which failed or uncertain check it is addressing and how the changed configuration will be verified. Rock placed near a pipe is not automatically proof that load transfer is correct or that all fatigue modes disappear. A post-intervention survey or appropriate verification may be needed. Translate an intervention proposal as a proposed mitigation until the installation and as-built condition have actually been assessed.
Avoid an unqualified ‘allowable span length’
The expression ‘maximum allowable span length’ sounds like a universal pipe property. In practice it depends on wall properties, masses, supports, environmental exposure, load cases and acceptance criteria. A length determined for one configuration cannot simply be copied across another location, depth or service state. When an engineering table states an allowable value, attach its assumptions and criterion. A measured span shorter than that value may satisfy the particular check used to derive it, but separate checks may still apply. Keep the phrase conditional: allowable under the specified methodology and operating envelope.
Eighteenth worked example: intervention prioritization
Imagine three hypothetical spans with calculated fatigue usages of 0.18, 0.44 and 0.29 under one stated model, and identical survey confidence. If the project intervention trigger were explicitly set at 0.40, the second span would cross that trigger while the first and third would not. In reality, the action plan might also consider model uncertainty, future scour, inspectability, installation feasibility and consequences of failure. The exercise explains how to read a ranked table without treating rank as a physical law. A risk-based maintenance decision must make the criteria and uncertainties explicit.
How model versions alter comparisons
Two analysts may report different fatigue estimates for the same physical span because they used different effective masses, environmental bins, support stiffness, response methods or standards editions. That difference alone does not prove one person made an error. Start with a controlled comparison: input data, method, governing edition, model version, units and output definition. If the difference remains after harmonization, investigate calculation details. Avoid choosing the smaller value simply because it is more convenient for a construction programme. A faithful translation explains which assumptions changed and why a revised estimate can be technically credible or still unresolved.
A worked example on decimal places
A calculated natural frequency of 0.374 Hz may be reported as 0.37 Hz to two decimals. That is reasonable for a summary if the full precision remains in the record and the rounding does not obscure a decision boundary. A frequency of 0.374 Hz is not equivalent to 374 Hz. An engineering plot whose x-axis shows ‘mHz’ would display 374 mHz, which equals 0.374 Hz. This is a straightforward yet common source of misreadings in multidisciplinary documents. Always preserve unit prefixes, and do not add more significant digits than the input quality supports.
A practice problem on load direction
A local flow velocity is 0.90 m/s and its horizontal angle to a straight pipeline axis is 60 degrees. In a simplified planar projection, the perpendicular component is 0.90 sin(60 degrees), approximately 0.779 m/s, while the along-pipe component is 0.90 cos(60 degrees) = 0.450 m/s. State that these are vector components, not a complete hydrodynamic-force solution. A practice answer that simply reports 0.90 m/s cross-flow has not interpreted route orientation. A separate current vertical profile and wave direction would be needed for a more realistic site assessment.
A practice problem on natural frequency
Two otherwise identical ideal bending beams have lengths 18 m and 24 m. Under inverse-square length scaling, f_24/f_18 = (18/24)^2 = 0.5625. If the shorter ideal beam has frequency 0.80 Hz, the longer one scales to 0.45 Hz. The numerical answer checks an algebraic relationship. It does not certify the two pipeline spans because real boundary conditions, effective mass, axial force and hydrodynamic effects may differ. A good explanation includes the conditional words ‘in the idealized model’ so the reader does not confuse a teaching exercise with a site calculation.
A practice problem on fatigue accounting
A hypothetical report lists three damage contributions for distinct exposure bins: 0.08, 0.11 and 0.06. Their linear sum is 0.25. If the project criterion were specified as D less than or equal to 0.50, that sum would be below the hypothetical limit; if a different criterion applied, the conclusion might change. Ask whether the bins overlap and whether the model already combines several mechanisms inside each bin. Double-counting would exaggerate damage; omitting relevant bins would understate it. The arithmetic is simple, but its meaning depends on well-defined exposure and mutually consistent contributions.
A practice problem on a change in clearance
A survey records a pipe gap of 0.12 m at one location and 0.20 m at the same location in a later survey. The reported increase is 0.08 m, or 80 mm. Relative to the initial gap, the increase is 0.08/0.12 = 66.7 percent. Do not claim that fatigue damage therefore rose by 66.7 percent: the relationship is not linear and may also depend on span length, support evolution and environmental conditions. A correct operational response first validates survey comparability and then reassesses the changed configuration under the relevant model.
A practice problem on ambiguous terminology
A project note says ‘the span was cleared’ without any accompanying evidence. That phrase could mean a hazard was removed, a survey accepted the span, the span passed screening or physical clearance increased. These are not interchangeable interpretations. The reader should request the action record, calculation disposition or survey measurement that resolves the phrase. In a translation, prefer the explicit statement ‘the 26 m span satisfied the stated screening check’ when that is what the record says, or ‘the gap was filled with an installed support’ only when construction evidence proves it. Precision prevents false assurance.
Independent transfer task: separate the checks
A fictional executive summary says a pipeline is safe because its submerged weight exceeds the predicted lateral hydrodynamic force, even though a 30 m length hangs over a depression. Identify the flaw. The force comparison may bear on on-bottom stability, subject to the proper methodology, but it does not assess the suspended span’s dynamic response and fatigue. The corrected summary should separately report the seabed-resting sections’ on-bottom check and the free span’s ULS and FLS results. The answer teaches that two legitimate calculations can address different physical questions without either replacing the other.
Independent transfer task: a current-frequency conflict
A fictional report gives U = 0.75 m/s, D = 0.50 m and illustrative St = 0.20, then claims f_s = 3.0 Hz. The simple scale gives 0.20 × 0.75 / 0.50 = 0.30 Hz, so the stated value is ten times too high. That calculation does not establish whether VIV occurs; it only catches an arithmetic or decimal-place discrepancy. The repaired explanation should label the estimate as a frequency scale for the chosen simplified conditions, then refer to the applicable response model for actual amplitude and fatigue. An error check is not a design conclusion.
Independent transfer task: the missing denominator
A field note says ‘the gap is 40 percent’ without explaining 40 percent of what. It might be clearance relative to pipe diameter, increase from an earlier survey, or a share of a corridor length. A faithful technical summary cannot invent the denominator. Request the definition and raw values. If the intended measure is 0.16 m clearance divided by 0.40 m diameter, report gap-to-diameter ratio 0.40. If the intended statement is a 40 percent change, identify the baseline time and quantity. These two interpretations have different physical meanings even though both use the same number.
Independent transfer task: fatigue-life language
A fictional engineer reports damage usage 0.40 in a specified exposure calculation. Another writer restates it as ’40 percent chance of rupture’. Diagnose the mistake: a calculated fatigue usage is not a failure probability. It arises from a particular stress-cycle model and resistance representation. The repaired text should say ‘calculated cumulative fatigue damage usage 0.40 under the stated methodology’ and then quote the relevant acceptance criterion if known. Probability of rupture needs a separate reliability model and uncertainty basis. Readers should never convert one dimensionless engineering ratio into a probability merely because both can be written as a percentage.
Independent transfer task: choose the next investigation
A survey identifies a longer unsupported gap after seasonal seabed erosion, while an old fatigue calculation still uses the former span profile. The correct next step is to reconcile the changed geometry and assess whether the previous model remains valid. It may require updated supports, environmental and structural input data, monitoring or a new fatigue check. The answer is not automatically to install rocks or declare failure from a photograph. Explain that action depends on the new condition and the governing criteria. This is an evidence-sequencing task, not a field engineering instruction.
Common error: false precision
A model may output fatigue usage to six decimals, but its environmental exposure and soil support may have much larger uncertainties than those digits suggest. Retaining full numerical precision internally for reproducibility is sensible; presenting a conclusion with unwarranted certainty is not. Show rounded figures consistent with decision sensitivity, and report uncertainty or sensitivity where meaningful. State the principal assumptions in a plain-language summary. A reader should be able to see whether the answer is robust to plausible changes, not merely admire the number of digits in the spreadsheet.
Common error: conflating guidance and certification
The existence of a DNV recommended practice does not mean that a calculation summary written by an unrelated author is DNV-approved, project-approved or fit for direct construction. Standards and recommended practices set out procedures and criteria within stated scopes and editions. Actual engineering decisions involve professional responsibility, site data, project specifications and independent verification. A world-facing educational article can teach how to question a report, follow dimensions and identify missing assumptions. It must avoid implying that worked examples replace an engineer’s project-specific free-span assessment or that a web page confers formal technical endorsement.
Frequently asked: can short spans vibrate?
Yes, depending on their geometry, dynamic properties and environmental conditions, even spans that look modest in length may require attention. ‘Short’ is a relative descriptor, not a universal VIV exemption. A given length may be sufficiently stiff and stable for one pipe but not another with different mass, boundary support or flow exposure. The useful question is not ‘How many metres are safe?’ but ‘Which analysis and acceptance criteria were applied to this particular configuration?’ If someone provides a universal safe length without specifying diameter, stiffness, current and support assumptions, ask for the missing basis.
Frequently asked: is vibration always a failure?
No. Detecting motion does not automatically mean a pipeline has failed or will fail. Engineers evaluate response levels, resulting stresses, cycles, fatigue resistance and applicable criteria. However, a small observed displacement at one sensor does not by itself establish safety either, because location, calibration and exposure duration matter. A sound summary can say that vibration was observed and quantify its frequency or amplitude while leaving the fatigue conclusion to a documented assessment. It is possible to be accurate about an observation without turning it into an unsupported reassurance or alarm.
Frequently asked: what does a pass mean?
A pass means the particular check met its stated criterion for the analyzed conditions and inputs. A free-span check can cover different phases, environmental loads and limit states, so a single word cannot substitute for the assessment scope. State the span identifier, survey or modeled geometry, operating state, ULS/FLS method and relevant criterion. If the report evaluated only one phase, do not silently broaden the conclusion to the full lifetime. If it evaluated many cases, report which govern. The most useful translation is one that keeps the boundaries of the evidence visible.
Conclusion: translate the chain of causation
A reliable free-span explanation runs from measured seabed geometry to the structural model, from waves and currents to dynamic response, and from response to stress cycles, fatigue usage and acceptance criteria. Keep on-bottom stability separate from suspended-span vibration. Use units for every velocity, frequency, length and ratio; distinguish measured observations from predicted outputs; and retain the assumptions behind any ‘pass’. The goal is not to make a complex offshore assessment sound easy by deleting nuance. It is to make the evidence readable enough that a careful person can recognize the important questions and the limits of the answer.
Original methods and verified neighboring reading
DNV-RP-F105 Free spanning pipelines, edition 2025-12. Use the full applicable source and project design basis for engineering work, rather than a brief web summary.
DNV overview of updated free-span recommended practice (2017 background). Use the full applicable source and project design basis for engineering work, rather than a brief web summary.
DNV FatFree engineering analysis module description. Use the full applicable source and project design basis for engineering work, rather than a brief web summary.
eduKateSG on-bottom stability guide (related but different check). Use the full applicable source and project design basis for engineering work, rather than a brief web summary.