The existence of ocean current greatly affects the performance of any deep-water operation and should be carefully defined during analysis and design of any deep-water system. This study focuses on introducing and testing a new proposed simplified method; namely the weighted current profile (WCP); in filtering vertical current profiles from a large database so the response analysis of a deep-water system can be carried out more efficiently. The method is tested for underwater operation of a deep-water work class Remotely Operated Vehicle (ROV); connected to surface vessel through an umbilical line; under more than ten thousand current profiles. The study reveals that the proposed WCP method can be used successfully in eliminating thousands of unnecessary current profiles from a big database.


Ocean current is one of the important aspects that is essential to consider when designing any deep-water system. Due to the scarcity of measured current speeds and directions along the water column in the past; a vertical current profile was commonly assumed to follow one of the simplified profiles which can be conservative for deep-water systems. In the last decades; several measurements have been conducted on different deep-water locations producing different current databases that can be used to define a more precise multidirectional current profile for designing and executing a deep-water operation. However; each database usually consists of thousands of fine current profiles which are cumbersome to analyze through conventional methods due to the large computational time. Performing statistical analysis on the deep-water current database is also still prohibitive without sacrificing the coherence of current speed and direction between depths along the water column.

Different approaches have been tested to capture the vertical current profile that produces the largest structural response of a deep-water system; known as extreme current profile; from a big current database. One of the approaches is by forming a family of extreme current profiles through conditional current analysis (Winterstein et al.; 2011; Winterstein et al.; 2009). The other approach is by performing the average conditional exceedance rate to define a statistical current profile with certain return period (Liu et al.; 2018). One popular method is the Empirical Orthogonal Function (EOF) which utilizes the mode-based analysis to extract the characteristic extreme current profile from database (Forristall & Cooper; 1997; Jeans et al.; 2003; Jeans et al.; 2012; Mattioli & Pizzigalli; 2016). This method was later compared with Current Profile Characterization (CPC) and an advanced clustering procedure called Self-Organizing Map (SOM) algorithm in reducing the big current database when analyzing the vortex-induced vibration of a steel catenary riser (Prevosto et al.; 2012). The clustering procedure showed a great advantage compared to the other two options since it can extract the extreme current profile from a big database in a more efficient way even when the database consists of fine current profiles. Due to the promising result shown by the clustering procedure; Jeans et al. (2015) compared different clustering algorithms in an attempt to extract a fine extreme vertical current profile from a deep-water current database in Gulf of Mexico for riser design. However; most applications of clustering in extracting the extreme current profile still excluded the information of current direction along the water column. Hunt et al. (2020) then tried to measure the significance of current directions during the clustering process when finding the extreme current profile for a vertically fixed flexible structure in deep-water with different fine current profile databases. They showed that the inclusion of current directions on the clustering procedure may improve the resulted extreme current profile. They also tested the combination of the unsupervised clustering approach and different dimensionality reduction methods to define the current profile that produces the largest structural response during deep-water operation of a Remotely Operated Vehicle (ROV). In this case; they tried three different dimensionality reduction methods: one is similar to the EOF procedure while the other two applies the artificial neural network with different approaches.

This content is only available via PDF.
You can access this article if you purchase or spend a download.