Page 13: of Offshore Engineer Magazine (Sep/Oct 2020)

Read this page in Pdf, Flash or Html5 edition of Sep/Oct 2020 Offshore Engineer Magazine

LEADING OFF Digital Transformation oil and gas industry has increased the amount of information Today, it is possible to temporarily store high-frequency collected and worked to make better decisions and gain in- data locally and transmit data centrally at various sample sight into operations. In the current market, it is more critical rates, depending on the level of detail required for remote than ever for companies to identify ways to economize and analysis. High-frequency data can be used locally to perform improve project economics. This means reassessing processes initial validation steps while compute-intensive mining and and reimagining human resource management to get the best algorithm optimization can be performed centrally. This en- return on investment. ables companies to retain full granular data for on-demand

First, data must be collected at a reasonable cost. This means requests and asynchronous replication by physically moving identifying data points that can be collected reliably and con- hard drives (such as Import/Export offering from Amazon sistently. Once collected, data points need to be aggregated and Web Services) or using spare bandwidth when it is available. normalized so they can be validated. With the evolution of edge The resulting improved algorithm from central processing computing, data processing speed has improved exponentially. can then be pushed back to the edge units to keep improv-

Neural Decision Processors, by Syntiant, for example, feature ing the data quality of the edge processing. Algorithms at the a processor so small that 54 chips can ft on a penny. Today, place of collection also can help correlate near-real-time data this task can be performed on site where high frequency data is feeds to create synthetic data when a value is erroneous or available for algorithms to take advantage of the best granularity missing and eliminate outliers.

of data available. Edge processing power also makes it possible An example of this approach is how offshore platforms are to enhance cybersecurity, leveraging machine learning to dif- managing rotating equipment. Digital twins are run centrally ferentiate valid traffc from undesirable traffc. and compared to actual measurements of the units so discrep-


Offshore Engineer