Session: 05-05-02 Floating System for Renewable Energy II

Paper Number: 79353

79353 - Stochastic Weather Window Analysis in Operations and Maintenance Planning Policies for Offshore Floating Multi-Purpose Platforms 

Offshore floating multi-purpose platforms (MPPs) are becoming common in the "blue economy." Significant benefits can be derived from shared use of infrastructure for multiple services and resource extraction.  Renewable energy generation such as from wind, waves, currents, etc.  and other functions such as aquaculture, leisure, and transport functions are all possible at or  in proximity of MPPs.  The satisfactory performance of these inter-dependent multiple functions depends upon a diversified set of operations and maintenance (O&M) activities that take different amounts of time and might require different levels of calmness in weather conditions.  

Markov Decision Process (MDP) approaches can be used in MPP O&M planning.  This inherently involves stochastic weather window analysis that operators can employ to schedule work activities.  In the marine environment, wind, wave, current and other weather-related variables can be summarized in the so-called Beaufort scale and this then affects navigability, selection of appropriate service vessels, crew personnel planning, etc.  The proposed method seeks to balance lost revenue during downtime versus unsuccessful or incomplete O&M activities that can lead to risks and safety issues due to volatile weather conditions. The method is illustrated using observed metocean data at a planned site. 

The MDP approach fundamentally considers Markov chain transitions from one weather state to another.  Based on actual weather patterns, we define transition probability matrices.  Distinct weather patterns leading to different transition matrices are synthetically defined and employed to represent: (1) highly variable seas; (2) stormy seasons; and (3) generally calm seas.  The proposed MDP approach seeks optimal policies in each of these cases to assess robustness of the methodology.  This is done by comparing optimized policies for an observed metocean data set with policies derived for the synthetic data sets.

 

Presenting Author: Taemin Heo The University of Texas at Austin

Authors:

Taemin Heo The University of Texas at Austin
Ding Peng Liu The University of Texas at Austin
Lance Manuel University of Texas at Austin