Optimal Microgrid Sizing of Offshore Renewable Energy Sources for Offshore Platforms and Coastal Communities

Technical Summary: Optimal Microgrid Sizing of Offshore Renewable Energy Sources for Offshore Platforms and Coastal Communities

Problem Statement
Isolated offshore communities and platforms, such as islands and oil and gas rigs, traditionally rely on mainland grids or diesel generators, resulting in high energy costs and significant carbon emissions. While offshore renewable resources (wind, wave, tidal, and floating photovoltaics) offer a sustainable alternative, their intermittent nature complicates system design. A critical challenge in microgrid planning is the accurate modeling of Battery Energy Storage System (BESS) degradation. Traditional optimization models often oversimplify degradation costs as a fixed percentage of capital investment or rely on limited input parameters, leading to inaccurate lifetime cost estimations and suboptimal system sizing. Furthermore, existing optimization tools often struggle with the computational complexity introduced by nonlinear degradation factors and the stochastic nature of marine energy resources.

Methodology
The authors propose a Renewable Energy Microgrid Optimizer (REMO) designed to determine the optimal mix and sizing of offshore renewable generation and storage resources. The methodology integrates five key components:

1. Resource Modeling: The model utilizes five years of historical data from the National Data Buoy Center (NDBC) and NOAA to project power outputs for Offshore Wind Turbines (OWTs), Wave Energy Converters (WECs), Tidal Energy Converters (TECs), and Floating Photovoltaic (FPV) systems across six US regions (Texas, Alaska, California, Florida, Hawaii, New Jersey).
2. Optimization Framework: The REMO model minimizes the total lifetime cost, comprising pre-commissioning, capital, operations and maintenance (O&M), decommissioning, and battery degradation costs. It employs a mixed-integer linear programming approach to select optimal capacities from a set of commercially available and prototype generation units.
3. Deep Neural Network-Based Battery Degradation (DNN-BD): To address the limitation of simplified degradation models, the authors integrate a DNN-BD module. This module is trained on 945 aging simulations using five critical input variables: ambient temperature, charge/discharge rate, state of charge (SOC), depth of discharge (DOD), and state of health (SOH). The network predicts cycle-by-cycle degradation to calculate precise use-based costs.
4. Decoupled Heuristic Algorithm: Recognizing that the DNN-BD model introduces nonlinearity and non-convexity, making direct optimization computationally demanding, the authors employ an iterative decoupled heuristic. This algorithm solves the REMO model, analyzes the BESS operation schedule to group cycles, calculates degradation costs via the DNN, and iteratively tightens operational constraints (limiting max power and total output) to reduce degradation until convergence.
5. Model Simplification: A "REMO-simplified" model is developed by analyzing the Levelized Cost of Energy (LCOE) of all resources. Based on findings that WECs and TECs exhibit significantly higher LCOEs in most regions compared to OWTs and FPVs, these resources are excluded in the simplified model to reduce computational time without sacrificing economic optimality for the tested scenarios.

Key Contributions

• Integration of Advanced Degradation Modeling: The paper introduces a novel integration of a Deep Neural Network into a microgrid sizing framework. Unlike traditional models that treat degradation as a static cost, this approach dynamically accounts for the nonlinear interactions between environmental factors (temperature) and operational stressors (SOC, DOD, C-rate).
• Iterative Optimization Strategy: The development of a decoupled heuristic algorithm allows for the inclusion of complex, non-convex degradation models within a standard optimization workflow, balancing computational feasibility with modeling accuracy.
• Comprehensive Regional Analysis: The study provides a comparative analysis of four distinct marine renewable technologies across six geographically diverse US regions, offering specific insights into resource viability (e.g., the dominance of OWTs in Alaska vs. FPVs in Hawaii).
• Cost Sensitivity Analysis: The paper quantifies the cost reduction thresholds required for emerging technologies (WECs and TECs) to become economically competitive, identifying specific price points (13% and 10% of current costs, respectively) where they could displace other resources in specific regions.

Results
Simulations conducted on offshore oil and gas platforms and a large coastal town yielded the following findings:

• Cost Reduction: The integration of the REMO-DNN-BD approach reduced the total lifetime cost by 7.7% and reduced BESS degradation costs by approximately 42% compared to unconstrained or simplified degradation models.
• Regional Variability: Optimal system configurations varied significantly by region. For instance, Alaska's strong wind resources allowed for lower lifetime costs with fewer storage requirements, whereas New Jersey required heavier reliance on FPVs and BESS due to seasonal wind variability.
• Resource Viability: OWTs and FPVs consistently demonstrated the lowest Levelized Cost of Energy (LCOE) across most regions. WECs and TECs were found to be economically unviable at current cost levels, requiring drastic cost reductions (to 13% and 10% of current levels, respectively) to become competitive.
• Computational Efficiency: The REMO-simplified model, which excludes WECs and TECs based on LCOE analysis, reduced computation time by 16% to 25% compared to the full REMO-classic model.
• Sensitivity to Wind: A 10% variation in wind speed resulted in a 16% decrease in lifetime costs for increased wind and a 33% increase for decreased wind, highlighting the system's sensitivity to resource availability.

Significance
The paper claims that the proposed REMO framework, coupled with the DNN-BD model, offers a more realistic and resilient design solution for offshore microgrids. By accurately accounting for the complex, use-based degradation of batteries, the model prevents the economic inefficiencies associated with premature battery replacement or oversizing. The study underscores that while offshore renewables hold immense potential, their successful integration requires optimization tools that move beyond static assumptions to capture the dynamic interplay between environmental conditions, operational patterns, and component aging. The findings provide a data-driven basis for policymakers and engineers to prioritize mature technologies (OWTs, FPVs) in the near term while identifying the specific cost-reduction targets necessary for emerging technologies (WECs, TECs) to contribute to future offshore energy systems.