Can AI cut the cost of subsea engineering?

After my previous blog, Phil Cooper of Xodus wrote, ‘No need to worry about experience leaving the industry. Automated design, IoT and Machine Learning will take care of everything!’ Now I am not sure if he was kidding or not, but that comment made me ponder ‘could the offshore industry in general, and the subsea sector more specifically, operate with fewer engineers and depend on computers and robots to answer all the questions and solve all the problems?’

So I set my brain to whirring – just like some robot – on the question of whether automata could solve the many and varied challenges that confront engineers during the course of a project or the life of a field. I thought back about things that I had witnessed over my 30 plus years of reporting on the offshore and subsea sectors.

At the back at the end of the 1980’s, I spent some time with Seanor, a small-ish Norwegian engineering company. It was involved in a design project for a subsea manifold and on the Friday I was in its office, the team discovered a pipe clash in the drawings. The team now had to reproduce all of the structural documentation minus the clash and as it had one of the first – and maybe the first? – computer-aided design (CAD) programs on a computer linked to a printer, it set the machine to run over the weekend in order to have the new drawings available on Monday morning.

There are several points to ponder here. Firstly, CAD – everyone has CAD systems now, but back then it was special. Secondly, it was early on in the advent of computer power, so it was still going to take some time to re-print the drawings, although faster than a draughtsman with a slide rule and a pencil. Thirdly, this was well before the internet was commonplace and being ‘online’ had almost no meaning at all, so there was no worry about someone hacking the system while it was running for three days. While this was quite a long time ago and at an early juncture in the use of computers, it proved their value for such a task – reproducing a document based on a repetitive task.

Scroll forward four or few years to the installation phase of two of Norway’s landmark subsea projects – Norsk Hydro’s TOGI (Troll Oseberg Gas Injection) and Saga’s Snorre subsea production station. Both projects – at the time Norway’s deepest multi-well systems based on template-manifolds – experienced significant problems prior to first production, although it might be said to be generous to describe the issues so kindly.

At TOGI, during the diverless pipeline pull-in which at the time was one of the biggest (20in) ever carried out at such water depths (303m), seabed debris fouled the surface of one of the end facings, preventing proper sealing and completion of the clamping. Kvaerner had to come up with a scheme to clean the pipe end face. I can not remember exactly how long it took, but it came up with a solution, completed the pull-in operation and the field came onstream. At Snorre SPS, there was an inherent fault with the control system cabling which all had to be replaced before the field could come into production.

I don’t believe that either of these problems could have been sorted out by a robot. The first was a problem solving issue matched with an operational scenario involving remotely operated tooling. The second was simpler – a retrieval and re-installation operation – but required painstaking work and many dives by an rov. How could AI have expedited this?

There are many examples of such scenarios which require the lateral thinking commonly associated with human beings and currently beyond the grasp of a robot. So yes, robots and AI could assist humans and fill the gaps for some tasks that humans can do, but could be bored by. Maybe robots could do some platform maintenance tasks that do not require ingenuity. The more complex, problem solving will remain, for the time being, the purview of humans, along with the types of physical jobs that might be beyond the capability and flexibility of robots. Just think about those early Dr Who episodes where the Daleks were stumped by a flight of stairs.


My Google search, like those of many others, is set to pick up any internet references to offshore, subsea, pipelines and floating production systems. It seems, though, that the majority of the items that appear nowadays are about a variety of marketing reports on subsea production, subsea controls, subsea grids, subsea anything, along with similar reports on future pipelay installations and concrete weight coatings and future floater projects. People have been making money from market reports for quite a long time, but these days, there seem to be more market reports than there are projects. And who are these people producing these reports anyway?


I spoke at two Subsea UK events at the end of last year about the history and future of technology. What struck me was that a number of young engineers came up to me to say that they did not know the the origin of the technology that they were specifying. Maybe someone with a long perspective on the subsea business needs to run a subsea history course. Hmm.


As I am new to his blogging business, I failed to have a visit counter installed for the first one, but managed to get it up and running not long after posting No 2. Even I am impressed – 310 visits after just three days – so time for potential supporters, ie advertisers, to pile in. Hint, hint.

Article views:1085 Article views: 1

Leave a comment

8 thoughts on “Can AI cut the cost of subsea engineering?”