Melvin Oduro believes one of the most consequential decisions in offshore engineering is still made too often on instinct. When a project picks the equipment that will be tested, installed and run on a platform or a subsea system, the choice tends to rest on the judgement of experienced hands and on whatever suppliers happen to have available. In a paper published in January, the commissioning engineer sets out an argument, and a method, for doing it differently: choosing by evidence that has been modelled and stress tested before anything is committed. It is the latest in a five year body of work that has made him a persistent advocate for putting offshore commissioning on a more quantitative footing, and it is the reason a British technology readership has cause to take note of him.
The paper, in the Computer Science and IT Research Journal, proposes a predictive decision model that links the way equipment is selected to how a commissioning project actually performs. Oduro and his co author, an independent researcher in the United Kingdom, review how selection parameters, reliability, the ability to withstand harsh marine conditions and lifecycle cost, map onto measurable outcomes such as downtime, schedule adherence and reliability in service, and then build a framework to connect the two. His contribution is not a single new algorithm but a synthesis: a way of drawing methods that already exist into one coherent decision tool for the engineers, project managers and procurement teams who, on a large project, rarely reason from the same data.
What marks the work out is the toolkit Oduro assembles. The model leans on multi criteria decision analysis, for weighing factors that resist a single score, on Bayesian networks, for reasoning under uncertainty, and on fuzzy logic, for the in between judgements engineering so often involves, all of it supported by simulation and by digital twins, software models of physical assets fed by live sensor data. Few working engineers write this kind of thing down. Fewer still publish it in the open rather than leaving it inside a contractor’s proprietary software, and that openness is a large part of what gives his work its reach beyond any one company.
Equipment choices on these projects are frequently made on judgement and availability. The argument is that the same choices can be tested against data before they are committed, so the risks become visible while there is still time and money to act on them.
His standing rests on consistency rather than a single breakthrough. Since 2021 Oduro has published a steady run of peer reviewed reviews on offshore commissioning, on risk, on flange management, on hydrate prevention, on the human factors behind safety, and on decommissioning, each adding to a coherent case that the field should move from reacting to failures toward anticipating them. The equipment selection paper is the most data intensive of the set, and the affiliation it carries, Halliburton Energy Services in Houma, Louisiana, marks how far the work has travelled from the earliest papers, written while he was based in Takoradi, Ghana. The throughline is his own: a sustained effort to make tacit commissioning expertise explicit enough to be taught, checked and improved.
The relevance to British industry helps explain why his work is being read here. The UK is in the middle of an enormous offshore decommissioning effort, with the eventual cost of cleaning up the North Sea now put at around forty four billion pounds, and operators have been urged to adopt new technology and share data to bring that figure down. The digital twin and probabilistic methods at the centre of Oduro’s framework are precisely the capabilities that agenda calls for, and his co author’s UK base places the work within the national research conversation. But the value for a British audience lies in the transferable method he has set out, not in any single basin; the same approach would apply to North Sea decommissioning as readily as to a new development half a world away.
Oduro is unusually candid about what his paper is and is not. It is, by his own description, conceptual: a framework drawn from existing literature and practice rather than a finished, validated tool. He presents it as a foundation for the quantitative models the field still lacks. That candour matters, because the usefulness of any such model depends entirely on the data behind it. A digital twin built on sparse or unreliable readings can mislead as confidently as a good one informs, and the discipline of gathering, cleaning and sharing offshore data remains uneven across the industry.
There is a fair scepticism about data driven decision tools generally, that they can lend a false air of precision to what remain judgement calls, and that a poorly populated model is worse than none because it conceals its own uncertainty. Oduro’s framing meets that objection rather than dodging it. By insisting the work is a direction of travel rather than a product, he avoids the overclaiming that has dogged the field, and he is explicit that a model is only as honest as the people feeding it, on projects that have not always rewarded candour about bad news.
I publish these in the open on purpose. If a method only ever lives inside one company, no one outside can test it, and a method nobody can test is not one I would want a project to rely on.
The scale of the problem he is addressing is well documented. A widely cited study by the consultancy EY, which examined three hundred and sixty five oil and gas megaprojects, found that sixty four per cent ran over budget and seventy three per cent missed their schedules, with completion costs averaging fifty nine per cent above the original estimate, and deepwater projects faring worst of all. Much of that slippage gathers in the final phases, when many separately built systems must be integrated and proven to work together, the very stage Oduro’s research concerns. His work is, in effect, an attempt to close part of that gap from the front end, at the point where the equipment is chosen.
Whether his approach earns a place on real programmes will depend, as he concedes, on something less glamorous than the methods themselves: whether operators invest in the sensors and data systems such models need to function. For now, the significance of Oduro’s work is that an engineer in active practice has taken the trouble to write down, openly and with appropriate caution, how a recurring and expensive decision might be made better. In a field where that kind of thinking usually stays locked inside companies, the act of publishing it is itself part of why his name travels.
Last updated: June 12, 2026
