- The paper demonstrates an optimization framework using a multi-objective modified firefly algorithm to size PV, battery, and hydrogen systems for enhanced economic viability and self-sufficiency.
- It reveals that while current battery systems yield higher net present value for moderate self-sufficiency, hydrogen storage excels for near 100% renewable penetration under future cost scenarios.
- The study shows that an optimized long-duration strategy effectively mitigates component degradation and manages seasonal variations to improve overall system performance.
This paper (2211.07833) addresses the critical challenge of intermittency in renewable energy systems by focusing on the optimal sizing and operational strategies for grid-connected photovoltaic (PV) systems integrated with energy storage. The study specifically compares Battery Energy Storage Systems (BESS) and Hydrogen Energy Storage Systems (HESS), taking into account crucial real-world factors like component degradation, variable electricity prices, and seasonal storage needs.
The core of the paper presents an optimization framework utilizing a Multi-Objective Modified Firefly Algorithm (MOMFA) to determine the optimal sizes of system components (PV capacity, battery capacity, or electrolyser, tank, and fuel cell sizes) and, for HESS under a specific strategy, operational parameters. The optimization aims to maximize two key objectives: Net Present Value (NPV), representing economic viability, and Self Sufficiency Ratio (SSR), indicating the technical reliability and environmental benefit of the system by maximizing renewable energy utilization.
Two main system configurations are investigated:
- PV/BESS System: Composed of rooftop PV, load, grid connection, and a BESS. Modeled with a conventional operational strategy (CS) where excess solar charges the battery and load deficits are met by the battery or grid. Battery degradation is modeled as a constant annual depreciation rate.
- PV/HESS System: Composed of rooftop PV, load, grid connection, and a HESS consisting of a PEM Electrolyser (PEMEL), Hydrogen Tank, and PEM Fuel Cell (PEMFC). This system is more modular as charging power (electrolyser), storage capacity (tank), and discharge power (fuel cell) are independent. Degradation of PEMEL and PEMFC is modeled based on polarization curves and operating time, assuming a proportional change in cell voltage.
Two operational strategies are explored:
- Conventional Strategy (CS): Applied to both BESS and HESS. Simple rule-based strategy: charge storage when PV output exceeds load, discharge when load exceeds PV output, as long as storage state is within limits.
- Optimised Long Duration Strategy (OLDS): Proposed specifically for HESS to leverage its long-term storage capability, particularly for seasonal storage. This strategy introduces two distinct operational modes for "sunny" and "cloudy" periods, defined by optimized start and end times (
tstart, tend). During these periods, the system's charging and discharging behavior is controlled by storage limits (LIMIT_SUNNY, LIMIT_CLOUDY). A key modification in the "cloudy" period discharge rule is avoiding discharge during low-cost "off-peak" hours to prioritize economic benefits. The optimization process determines not only the component sizes but also these operational parameters (tstart, tend, LIMIT_SUNNY, LIMIT_CLOUDY).
The optimization problem is formulated to maximize NPV and SSR over a 25-year project lifetime. The NPV calculation considers capital costs, annual O&M costs, replacement costs (modeled with replacement factors rrep,i or set to 1 for ultimate cost scenarios), and revenue from electricity bill savings based on time-of-use rates. The SSR is calculated as the percentage of load met by the PV-ESS system, reducing reliance on grid import.
The optimization is performed using MOMFA, an enhanced Firefly Algorithm incorporating chaotic maps for initial population generation and Lévy flight for improved search capability. The multi-objective nature is handled using Pareto optimality and a Non-Dominated Sorting Algorithm (NDSA), allowing the algorithm to find a set of optimal trade-off solutions (Pareto front) between NPV and SSR. The paper demonstrates MOMFA's robustness and accuracy compared to the widely used NSGA-II algorithm for this complex problem.
Practical implementation requires detailed modeling of system components, including their efficiencies and degradation over time, along with external factors like solar irradiance profiles, load demand patterns, and dynamic electricity pricing.
- Input Data: Hourly or sub-hourly time series data for PV generation (or solar irradiance), load consumption, and time-varying electricity prices are essential. Historical data or forecasts are needed for solar degradation and price changes over the project lifetime.
- Component Models: Mathematical models for charge/discharge behavior and state of charge/health updates for BESS and HESS components are required (Eq 1-4 for BESS, Eq 5-7 and degradation models for HESS).
- Cost Models: Equations for calculating capital, O&M, and replacement costs based on component sizes and unit costs (Eq 11-13).
- Objective Calculation: Implementation of NPV and SSR formulas (Eq 8-10, 14-16) that integrate the simulated system operation over the entire project duration.
- Optimization Algorithm: Implementation of the MOMFA (or another multi-objective evolutionary algorithm like NSGA-II for comparison). This involves:
- Defining decision variables (component sizes, operational parameters for OLDS).
- Setting bounds and constraints for these variables.
- Implementing the objective functions (NPV and SSR calculation).
- Coding the algorithm's search process (initialization, movement rules, attractiveness calculation, non-dominated sorting).
- Running the optimization for a sufficient number of iterations and population size to converge to a well-distributed Pareto front.
The paper applies this framework to a real-world warehouse case study in Ho Chi Minh City, Vietnam (tropical climate with consistent solar), and a synthetic case in Melbourne, Australia (subtropical climate with high seasonal variation), using real and simulated data respectively.
Key Findings and Practical Implications:
- MOMFA Performance: Demonstrated to be more accurate and robust in finding the Pareto front compared to NSGA-II, suggesting it's a suitable choice for complex energy system optimization.
- BESS vs. HESS:
- At current costs and for achieving low-to-medium SSR (around 50%), BESS generally offers better economic performance (higher NPV) than HESS due to lower capital costs.
- For achieving very high SSR (approaching 100%) under future "ultimate" cost scenarios, HESS becomes significantly more viable than BESS. BESS struggles to achieve very high SSRs economically, especially in locations with high seasonal variability, due to the need for impractically large PV and battery capacities. HESS, with its long-duration storage potential, is better suited for high renewable penetration scenarios requiring seasonal energy shifting.
- Impact of Operational Strategy (OLDS for HESS):
- The proposed OLDS significantly improves the economic performance (NPV) of HESS compared to the conventional strategy, particularly in climates with high seasonal solar variation (MEL).
- OLDS effectively manages storage to optimize hydrogen use during high-value periods (peak/shoulder hours) and seasons (cloudy months), leveraging HESS's long-term storage capacity.
- OLDS can also reduce component degradation by controlling operation, leading to increased overall system efficiency over the project lifetime.
- Impact of Location: Tropical climates with consistent solar resources generally allow for achieving higher SSRs and NPVs with smaller system sizes compared to subtropical climates with significant seasonal variations, highlighting the crucial impact of local solar profiles.
Implementing this research requires substantial computational resources, especially for the optimization process over a 25-year hourly simulation period. The complexity of the HESS models, including degradation, adds to the computational load. Deployment in real-world systems would involve translating the optimized sizes and operational strategies into control logic for the ESS components and PV inverter, potentially integrated into an energy management system (EMS). The parameters tstart, tend, LIMIT_SUNNY, LIMIT_CLOUDY determined by the OLDS optimization would become key inputs to the EMS's HESS control logic.
Overall, the paper provides a practical framework and valuable insights for designing grid-connected PV-ESS systems, confirming BESS as currently more economical for lower renewable penetration but highlighting HESS, especially with optimized long-duration strategies like OLDS, as a critical solution for achieving high levels of renewable energy self-sufficiency in the future, particularly in regions with significant seasonal variations.