Text

theses

Despite the promising outlook, the large-scale adoption of offshore wind (OSW) energy is hampered by its high operations and maintenance (O&M) expenditures. On one hand, offshore-specific challenges such as site remoteness, harsh weather conditions, and high crew/equipment transportation requirements significantly inflate O&M costs. On the other hand, the uncertainties associated with key environmental and operational parameters largely obscure the wind farm operator's ability to identify optimal O&M actions. In response, the overarching aim of this dissertation is to formulate, develop, and extensively test a set of data-driven optimization methods that can adequately address the unique O&M challenges and uncertainties faced by OSW farm operators. Towards that goal, three research efforts, corresponding to Chapters 3, 4, and 5 of this dissertation, are proposed.In Chapter 3, an offshore-specific maintenance optimization method, called the holistic opportunity-based strategy (HOST) is proposed. The method is rooted in ``opportunistic'' maintenance optimization, wherein the objective is to optimally group maintenance tasks at opportune time windows. Formulated as a multi-staged mixed integer linear program (MILP), HOST leverages favorable weather conditions and onsite maintenance resources in order to produce economically optimal maintenance schedules, thereby realizing total cost improvements of up to 6.8% compared to prevalent maintenance strategies.
In Chapter 4, the mathematical model of HOST is extended, to accommodate the multi-source uncertainties in key environmental and operational parameters, including electricity prices, asset degradation, and weather conditions. The stochastic holistic opportunistic scheduler (STOCHOS), is a rolling-time-horizon-based MILP that employs a sample average approximation method to model uncertainty, coupled with a scenario generation method which constructs stochastic scenarios that effectively characterize the temporal dependencies of the input parameters. Extensive numerical tests show that STOCHOS achieves up to 6.3% improvement in total costs relative to its deterministic counterpart, HOST, demonstrating the value of considering uncertainty in OSW maintenance planning.
Finally, Chapter 5 extends and couples the methods developed in Chapters 3 and 4, with production control decisions in order to model the emerging dependencies between operations and maintenance. Turbine control, such as yaw optimization, has the potential to alleviate the fatigue load variations of critical components, potentially at the cost of reduced power production. The proposed model, called the production optimized STOCHOS by yaw decision control (POSYDON), addresses this interesting trade-off between the short-term revenue maximization (via turbine control) and the long-term cost minimization (via maintenance scheduling). Results show a superior performance of POSYDON, as prolonged maintenance cycles and production loss savings of more than 10% are achieved over methods that overlook those O&M dependencies, thereby confirming the value of jointly considering production and maintenance optimization in OSW farm operations.
Overall, this dissertation offers both methodological and applied contributions to the literature and practice of OSW maintenance planning. On the methodological side, a set of novel methods and models that holistically account for OSW specific challenges and uncertainties are proposed. In terms of applied contributions, the numerical experiments conducted herein use real data and state-of-the-art forecasts from the US North Atlantic – a region where more than 11 GW of OSW energy projects are planned for operation by 2040. We therefore hope that the models, analyses, and findings in this dissertation will provide valuable insights to the operators of those future OSW farms.

http://dissertations.umi.com/gsnb.rutgers:12335

Ph.D.

Includes bibliographical references

doi:10.7282/t3-pne3-3a95

The author owns the copyright to this work.