The Business Network for Offshore Wind threw me a curve yesterday. Big Data, AI & Blockchain was the name of the conference held in Boston. What does that have to do with offshore wind? As it turns out, plenty. (Well, not so much blockchain.)

We spent almost seven hours on Big Data and AI (no one ever even defined it as artificial intelligence). Offshore wind development relies on these to de-risk projects by making them more efficient and removing uncertainty. Two examples will have to suffice here: site assessments and O&M (another term that has simply passed into the lexicon).

Site Assessments

BNOW gathered a variety of experts with experience in offshore wind site assessments. One speaker, Julia Robinson Willmott of Normandeau Associates, explained wildlife surveys. Ms. Willmott led the Normandeau team in completing for NYSERDA a survey of the 16,000 square mile New York Offshore Planning Area, the largest aerial wildlife survey ever.

Traditionally, to conduct a survey of wildlife in an ocean surface environment one uses a boat. The engine noise can scare off what the researcher is seeking, or attract it (think dolphins leading a boat). Sometimes both can happen, as when one species is attracted (dolphins) and another is scared off (schooling fish). Further, human observers are not always reliable. They can fall asleep when there is little to see, or be overwhelmed by numbers, or misidentify an endangered species. BOEM has reported traditional observers missed 75% of observations of sea turtles.

A wildlife survey is improved by a low-flying plane with a camera. BOEM concludes this is a significant advantage because it does not rely on human recall and mitigates human preference/bias. More ground can be covered faster leading to reduced costs, but there is still the problem with noise, and after the wind farm is up, the survey cannot be replicated. Which leads to the Big Data solution: a high-flying plane with a high-resolution digital camera is faster, safer and provides more data. Swathes of the ocean are locked down with GPS coordinates as is the location of the observer. The permanent record that is created is auditable and can be reviewed by multiple observers. And it leads to Big Data. As reported in the press release concerning the NYSERDA survey: “Halfway through this survey, which is the world’s largest and most detailed digital aerial survey of offshore marine and bird life, more than two million ultra-high resolution images of birds, sharks, sea turtles, fish and marine mammals have been captured, including two blue whales and six North Atlantic right whales.” Two million is a lot. Ms. Willmott converted those numbers into digital terms: terabytes of data.

O&M

Big Data does not end after the shearwaters, Kemp’s Ridley sea turtles and basking sharks are counted. It will show up in the “digital twin” prepared for each turbine and monopile. Dr. Athanasios Kolios, a consultant to Ramboll (Ramboll is part of the EU’s ROMEO project, an initiative to reduce the O&M costs of offshore wind farms), explained the concept. A digital twin is “a digital model continuously monitoring how the structure is doing and updated with real time information about the loads affecting the structure.” With numerous sensors continuously monitoring the structure, operators can better estimate the remaining useful life of their equipment and its components. This allows optimization of the maintenance schedule, deferring some items and accelerating others. The goal is to drive unplanned maintenance to zero, reduce turbine downtime, and extend project lifetime. Stated another way, the goal is to reduce “tower climbs.”

Dovetailing with Dr. Kolios’s talk was that of Ali Askari, of ULC Robotics. ULC’s unmanned aerial vehicle (aka drone) flies up and down and around the monopile and nacelle taking pictures all the time. The magic of digital allows the camera to adjust the focal length so that it can take a picture of the front side of a pipe and then, when on the other side of the structure, take a picture of the back side. When the flight is done and the data downloaded, AI correlates all the photos so that where a concern arises over a particular weld, for example, a reviewer has at hand all the corresponding photographs. So, instead of hanging a technician over the side, reviewing manually over 60 hours, achieving at best 97% accuracy and being limited to only a side-by-side comparison, ULC can send its drone out from the deck of the workboat, conduct a digital review in one hour, achieve better than 99% accuracy and apply multiple analytics at the push of a button. And then the data can be applied to the digital twin.