DNV-RP-F107 is the reference method for assessing dropped object risk to subsea assets, and most offshore risk consultancies have a study or two built on it. Yet the recommended practice is exactly that — a practice, not a recipe — and it leaves the analyst a series of judgement calls. Reviewers, whether client-side or regulator-side, tend to find the same weaknesses again and again. This article walks through six of the most common ones, and what a defensible study does differently.

1. Letting the concentric rings do the thinking

The classic implementation divides the sea floor into 10 m annular rings centred on the drop point and distributes hit probability across them. That was a sensible simplification when studies were built in spreadsheets, but it has a real cost: everything inside a ring is treated as equally exposed. Close to the drop source — precisely where the hit frequency is highest and most of the risk lives — a 10 m ring can span the difference between a direct hit on a manifold and a harmless landing on open seabed.

A ring-based answer is not wrong so much as blunt. If the targets that matter are small relative to the ring area (and pipelines, umbilicals, and valve skids almost always are), the ring average can materially under- or over-state the frequency on the specific asset. Higher-resolution approaches — a fine cellular grid rather than coarse annuli — keep the same F107 probability model but stop averaging away the geometry. That is the single biggest practical improvement available over the traditional method.

2. Applying the wrong lateral deviation category

F107's excursion model spreads dropped objects across the seabed according to object category — flat/long objects deviate differently from box-shaped ones, and weight class matters. A surprisingly common error is categorising the lift inventory once, early, and never revisiting it: the drill collar treated like a container, the container treated like a spool piece. Because the deviation parameter drives how far objects wander from the drop point, a mis-categorised object population quietly reshapes the entire hit-frequency map. A defensible study documents the lift list, the category assigned to each item, and why.

3. Taking generic drop frequencies on faith

The base drop frequencies in the recommended practice are derived from historical North Sea lifting data. They are per-lift frequencies — so the study is only as good as its lift count. Reviewers regularly find studies where the number of crane lifts per year is a round-number guess, or where drilling-programme lifts and platform-crane lifts have been mixed into one undifferentiated figure. The frequency table is not the weak point; the exposure estimate is. Get the lift schedule from the operations team, split it by crane and by load type, and state the basis in the report.

4. Stacking conservatisms until the answer is meaningless

Each individual assumption — heaviest object in the class, worst-case impact energy, no protection credit, everything hits end-on — can be defended as "conservative". Multiplied together they produce a risk number that no longer describes the operation, and that can push a client toward expensive protection structures the risk does not justify. The remedy is not optimism; it is transparency. Carry best-estimate and conservative cases side by side, or better, run the assessment probabilistically so the conservatism is explicit in the distributions rather than hidden in point values.

5. Ignoring the aggregate picture across multiple sources

Real facilities rarely have one crane and one drop zone. Platform cranes, vessel cranes alongside, and drilling activities each contribute, and the asset at risk sees the sum. Single-source studies answer the question "what does this crane contribute?" while the operator's actual question is "what is the hit frequency on this pipeline?" If the study cannot aggregate multiple drop sources into one exposure map, someone ends up adding independently-produced numbers by hand — or, more often, nobody adds them at all.

6. Results a reviewer cannot retrace

The most common regulator comment on dropped object studies is not about the maths — it is about traceability. Which targets were assessed, with what dimensions? Which energy method was used, DNV energy distributions or object-specific kinetics, and why? What damage capacity was assumed for the pipeline, and does it credit coating or burial? A study that presents a single risk number without the audit trail invites a round of questions that costs more time than the original analysis. Modern tooling helps here: visual layers a client can open themselves (including formats like KML for Google Earth), exportable calculation tables, and rerunnable sensitivity cases turn review from correspondence into a working session.

What good looks like

None of these failure modes is exotic; each comes from treating a recommended practice as a fixed procedure rather than an engineering method. A dropped object study that stands up to review typically has five properties: resolution fine enough that geometry matters (metre-scale, not 10 m rings), a documented and categorised lift inventory, an exposure basis agreed with operations, explicit rather than stacked conservatism — ideally probabilistic — and outputs the reviewer can interrogate without writing to the author.

This is the gap DORAS was built to close: DNV-RP-F107 methodology on a 1 m² cellular grid, multi-source aggregation, Monte Carlo simulation alongside the traditional deterministic method, and both DNV energy distributions and physics-based kinetics so reviewers can reconcile the numbers against what they know. Studies run in seconds, which means sensitivity cases get run rather than argued about. A free 30-day evaluation is available at dropulyzer.com — no card required.