Page 29 - Layout 1

THOUGHT LEADERSHIP

Big Data is popularly deined as involving data

sets so large or complex that traditional data

processing applications are inadequate, but

it often refers simply to the use of predictive

analytics or certain other advanced methods to

extract value from data.

we have a lot of when Big Data specialists and experts in management systems, equivalent to
it. Without contextdrill-ship operation combined forces. The inancial controlling systems, giving insight

and without smart Big Data algorithms had picked up many to asset managers by allowing global

algorithms, a lot of data isanomalies, notably one concerning lube overviews of e.g. a leet of vessels with
useless - nobody looks at it, nobody oil temperature peaks on certain shifts. drill-down capabilities to sub-systems.

gets any wiser. The irst task is converting The vessel experts quickly identiied a We often think only of ship operators as

Big Data into Small Data. Humans can slight handling issue by one operator, the beneiciaries of Big Data technology.
process only Small Data: We look at which could lead to a premature and very But, again, Big Data can do more.

business trends.costly equipment failure unless rectiied. Maritime authorities should monitor the

Explanation of the error and its potential developments closely. For example, DNV
Beyond Big Data lie the treasures
consequences to the crew member solved GL recently performed a study for the
Big Data software can perform some tasks the problem.Norwegian authorities on near-collisions in |27

automatically in the process from turning Norwegian territorial waters. By combining
Big Data can do more than
a lot of data into smart decisions. Typically human intelligence on vessel behavior
tasks are averaging, arbitrary curve itting Condition-Based Maintenance before near-collisions, AIS data could be

(for trends, etc.), and detecting anomalies Many Big Data applications in the shipping iltered to yield a reasonably small set of

(outliers in raw data, in trends, in trends industry focus on condition-based cases to be screened more closely. There
of trends, etc.). While this goes far beyond maintenance:were some false positives, such as a ferry

your average Excel analysis, it could still • Main and auxiliary engines, e.g. ABBreturning to pick up a stranded passenger

be described as “gloriied statistics”. In (Nakken and Prytz (2016))who had missed the last ferry of the

many applications, we need to add more • Hull condition in terms of corrosion,day. But mostly, the algorithm correctly
knowledge to unleash the power of Big e.g. Saipem (De Masi et al. (2016)), USidentiied near-collisions – and twice as

Data. In analogy: If you want to ind a Navy (Donaldson (2014))many as had been reported.

needle in a haystack it helps to know that • Hull and propeller condition in terms
needles are magnetic.of fouling, e.g. Jotun and DNV GL Conclusion – We need Big Insight

As an example, consider an anonymized (Krapp and Bertram (2015)), Akzonobel Big Data – whatever that may be – is

case study. A brand-new drill-ship had (Ramsden (2016))here and bound to thrive, but the key to
collected 1.3 PetaBytes (= 1 million The applications generally build upon success is turning Big Data to Big Insight.

Gigabytes) data from 10,000 sensors in data that is already available in assorted And this requires combining huge data

under six months. The irst insight fromsub-systems and reporting schemes,sets intelligently with smart algorithmic

a Big Data analysis showed that aboutbut generate additional insight through processes (see DNV GL’s ECO Insight
1/3 of all of the data violated simple the smart fusion of information and by as a prime example). The future lies in

plausibility checks. While it was not the generally compiling individual “snapshot” information fusion; where we combine and

insight the customer aimed for, the serious data to time-lines with trend prediction cross-check data sets from many diferent
problem with the data quality was a and automatic screening for anomalies.stakeholders, adding expert knowledge to

valuable insight as warranties could still be There is a trend towards merging them. Then we will see Big Data turn into

claimed. But the real break-through cameinformation in integrated technical assetBig Insight.

ISSUE 50 | WAVES