IMCA S 026, Unpacked: What an Out-of-Straightness Survey Actually Requires

What an OOS Survey Is Actually For

An out-of-straightness (OOS) survey defines the longitudinal and plan profile of a rigid subsea pipeline so a buckling analyst can find the imperfections that matter. The physics is the reason the work exists: hydrocarbons move at temperature and pressure above ambient, the line wants to expand, and that axial driving force looks for the path of least resistance. Where the pipeline already deviates – over a boulder, a slumped trench wall, a crossing – far less force is needed to buckle it. Those deviations are the OOS events. S 026 splits the job in two: a vertical OOS survey (on trenched, backfilled or rock-dumped line) to catch upheaval buckling, and a lateral OOS survey (on exposed line) to catch lateral “snaking”.

The point S 026 makes early, and returns to, is that survey accuracy is not academic: it feeds directly into how much restraint – trench backfill or rockdump – the analyst calls for. Noisier data means a larger assumed error, which means more rock. So the document frames the whole exercise as a cost trade-off, and it says so explicitly: acceptance criteria should be set “based on client specifications and a review of survey costs versus buckling mitigation costs,” because “a more costly, more accurate OOS survey may save more than it costs by permitting a reduction in rockdump.” That speed-accuracy-cost framing is the spine of the standard, not an afterthought.

The Pre-Survey Checks It Does Specify

This is where S 026 is more concrete than its reputation suggests. Two checks are mandated before and during a survey, and both carry real acceptance figures.

The Static Bathymetry Check lands the ROV on the seabed (or a structure) and logs the depth sensors while stationary. The survey procedure has to define its frequency and an acceptance criterion – “typically SD <25 mm” – that permits the survey to proceed, tied to the noise the project will tolerate. Sensor stability is checked by differencing the depths, with a typical acceptance of a change of less than 0.1 m. It is repeated at the end of a run, or mid-survey, if the sea state shifts or a sensor is suspected of drifting.

The Dynamic Test (also called the Repeatability Test) pilots the ROV several times over a section that includes a real OOS event, such as a trench transition, as a full dress rehearsal of the survey. Its acceptance criterion for vertical OOS is the standard deviation of the depth differences between runs being below a value – “such as 25 mm” – agreed with the buckling analysts and tied to the OOS data acceptance. It does double duty: it proves the positioning and profilers, familiarises the pilots and online surveyors, and its final run can become the start of the survey proper. For horizontal OOS it measures lateral repeatability the same way.

Note the pattern: S 026 gives you the check, the method and a representative number, then explicitly hands the final value to project agreement. That is the document’s whole posture, and it is worth recognising rather than mistaking for silence.

Choosing the Sensor, and the Accuracy You Are Buying

Far from ducking sensor selection, S 026 lays out the families and their honest trade-offs, then quantifies them.

• Contact wheel – depth sensors on the axle of a wheel rolled along the pipe, with the undercarriage using the pipeline as a rail. Cheap, no profiler-offset processing error, good for high-accuracy lateral work, but slow (around 500 m/hr), demanding to pilot, and it needs an accessible top of pipe. It warns about the real failure mode: rolling over a bracelet anode or piggyback spacer pitches the ROV up and breaks wheel contact, degrading the profile.
• Profilers (MBES and laser) – the ROV flies above the pipe and the transverse profiler measures the offset. Faster (up to 1000 m/hr), easier to pilot, suited to piggybacked lines. Laser gives an order of magnitude more resolution than MBES; dual-head systems see round the pipe and measure embedment. The document is candid about MBES limits: sparse returns from the pipe crown, fish blocking returns, shadow zones from camera booms, and low-density insulation coatings that can absorb the acoustic energy.
• Pipetracker – an induced-magnetic-field sensor that works on buried, rock-dumped or concrete-covered line, with its own onshore-then-offshore calibration routine.
• Instrumented pig and photogrammetry/SLAM are both covered too – the pig for fast INS-based runs (corrected by odometer or electromagnetic weld detection), photogrammetry where no precise positioning is available, each with its stated catch.

The accuracy is tabulated, not hand-waved. Single-sensor accuracies (2σ) run from 0.01 m for scanning laser and 0.03 m for MBES to 0.10 m for a dual-head scanning sonar or a TSS440 pipetracker, combined through an error budget of √(Sensor² + ROV Offset² + ROV MRU²). And the speed-versus-accuracy trade is laid out directly: a pipetracker degrades from about 3.5 cm at 500 m/hr to 7 cm at 1000 m/hr; a scanning laser holds near 1 cm across the same range. This is exactly the framework a survey manager needs to price accuracy against vessel time – and it is already in the document.

Managing Noise and Swell – the Part People Skip

S 026 treats survey noise as a first-class problem, not a footnote. Swell causes pressure fluctuations at the depth sensor, worst in shallow water (<50 m) but visible deeper with a long period; it lists the noise sources (swell, motion-correction latency, inaccurate top-of-pipe picking, sensor imprecision).

Crucially, it gets the filtering subtlety right. It recommends a time-domain filter equal to the swell period – but pairs that with a risk assessment of removing a genuine OOS feature whose wavelength matches the noise produced by the swell period and ROV speed. That is the real trap of naive filtering, and the standard flags it. It also offers the cleaner route: use INS z-values aiding (or aided by) the pressure sensor to suppress swell without smoothing the signal.

Then it does something many readers miss – it ties QC to a specific external standard. De-spiking is constrained to removing only single points “clearly deviating in an unphysical manner,” with a spike commonly defined as an amplitude greater than twice the SD, and it explicitly notes that DNV-RP-F110 states the survey company should not smooth or process the data. The residual noise band after de-spiking is not discarded; it becomes the measure of data quality, feeding the safety factor and the size of the “non-detectable OOS” that could hide within the noise. A typical vertical OOS specification is an SD below about 30 mm.

The Deliverables That Carry the Risk

The deliverables section is precise about the things that actually break analyses. A vertical OOS listing must carry, as a minimum, KP, Easting, Northing and top-of-pipe depth. Two details matter more than they look:

• Top of pipe to three decimal places. Reporting to two decimal places introduces depth steps that flatten real OOS peaks and troughs – the document illustrates exactly this. Depths must also be tidally reduced to the project datum.
• Seabed water temperature, recorded for the buckling analysis (it sets the tension restrained by pipe-soil friction), plus internal pressure if the line is live.

Embedment is handled properly here, and this is the detail an earlier generation of commentary got wrong: the “five-point” and “seven-point” references in S 026 are not a QC method – they are recognised conventions for defining pipeline embedment and the profile of the adjacent seabed (port and starboard), which feed lateral-buckling analysis. Depth of cover is measured at an agreed interval (typically 5 m) and reported as maximum, average or minimum depending on whether the cover is rockdump or blocky backfill.

Where S 026 Stops, and Your Procedures Begin

So what is fair to say against it? Not that it is incomplete on method – it is not. The honest limitations are narrower:

• It is a framework, not a spec sheet. Almost every acceptance number is “agreed with the buckling analysts” or “defined in project procedures.” You cannot lift a ready-made acceptance specification out of S 026; you have to set the values yourself against the buckling case and the cost trade-off. That is deliberate for a guidance document, but it means the standard is a starting point for your procedures, not a substitute for them.
• No single worked end-to-end example. Appendix 1 gives a one-page OOS survey process flow, but there is no fully worked case carrying raw data through to a final deliverable with the awkward bits (position loss, sensor disagreement, data gaps) shown.
• No explicit rule for reconciling disagreeing sensors. When a laser and a pipetracker disagree on the same feature, the error-budget tables rank their expected accuracy, but the document does not give a procedure for which to trust – that judgement is left to the surveyor.
• It leans on DNV-RP-F110. The QC and buckling rationale assumes you also have the relevant DNV recommended practice to hand; read S 026 alone and some of the “why” sits off the page.

These are real, but they are the limits of a guidance document doing its job – not the gaps a careless reading might invent.

Bottom Line

IMCA S 026 is a stronger document than its brevity suggests. It tells you what to measure, the checks to run with representative acceptance figures, the error budget that prices accuracy against vessel time, and the QC philosophy that ties survey noise to the rock you will eventually pay for. Read it for the method and the discipline, then bring your own project procedures for the numbers and your buckling analyst for the acceptance criteria. For the surrounding detail, S 026 points to its IMCA companions – S 003 on multibeam, S 013 on deep-water acoustic positioning, S 017 on USBL, S 019 on subsea metrology, S 021 on peripheral survey sensors and S 025 on shared ROV/survey sensors – and the buckling QC ultimately answers to DNV-RP-F110. Used that way, S 026 does most of what a survey manager needs; the rest is the part it always intended you to supply.

---

This article provides independent analysis of IMCA S 026. It is not endorsed by or affiliated with IMCA. For the complete guidance, refer to the official IMCA publication at imca-int.com.