Optimal Configuration Method for the Installed Capacity of the Solar-Thermal Power Stations

Yan Wang, Zhicheng Ma, Jinping Zhang, Qiang Zhou, Ruiping Zhang, Haiying DongEnergy 2023 · DOI

TL;DR

This paper proposes a multi-objective optimization method to determine the ideal installed capacity of solar-thermal (CSP) power stations within large-scale renewable energy bases. By balancing grid stability and operational costs using chance-constrained programming, the researchers demonstrate that adding 1 GW of solar-thermal capacity to a specific base in Gansu, China, can significantly improve voltage recovery, reduce peak-shaving stress, and increase the export limit of long-distance DC transmission lines.

What Problem Does This Paper Address?

As the world shifts toward renewable energy, regions like Gansu Province in China are building massive "new energy bases" dominated by wind and photovoltaic (PV) power. However, these sources are notoriously volatile and random. This creates two major headaches for grid operators:

  1. Peak Regulation: Wind and solar don't always produce power when demand is highest, requiring other plants to "shave" the peaks and fill the valleys.
  2. Voltage Stability: Large-scale renewable integration can make the power grid "fragile," especially for Ultra-High Voltage Direct Current (UHVDC) lines that transport power over thousands of kilometers. If a fault occurs, the system may struggle to recover its voltage.

While conventional thermal (coal) plants usually handle these issues, the goal is to replace them. Solar-thermal power is a promising alternative because it uses a traditional steam turbine (providing mechanical inertia) and includes thermal storage to provide steady power even when the sun isn't shining.

Methods

The authors developed a three-stage framework to find the "Goldilocks" capacity for a solar-thermal plant:

  1. Uncertainty Modeling: Using Chance-Constrained Programming (CCP), they modeled the inherent randomness of wind and PV forecast errors as normal distributions. This allows the grid to plan for a "confidence level" (e.g., being 95% sure constraints are met) rather than assuming perfect forecasts.
  2. Scenario Reduction: To make the math manageable, they used K-medoids clustering to condense 365 days of meteorological data into four "typical scenarios": Large Wind/Large Solar (LWLS), Small Wind/Small Solar (SWSS), and the two mixed variations.
  3. Multi-Objective Optimization: They built a model to minimize two competing factors:
  4. Transient Voltage Recovery Index (TVRI): A measure of how quickly and effectively the grid recovers from a fault.
  5. Comprehensive Operating Cost: Including investment, operation, and peak-shaving costs.

The resulting non-linear problem was linearized and solved using the CPLEX solver within the MATLAB-YALMIP environment.

Datasets

The study utilized a real-world case study from a new energy base in Gansu Province, China (specifically the Jiuquan-Hunan ±800 kV UHVDC project).

  • Wind/PV Data: Actual forecast and output data from the Gansu base as of July 2021.
  • Solar-Thermal Data: Technical parameters and performance data generated using SAM (System Advisor Model) software.
  • Grid Parameters: Standard parameters for a 100 MW solar-thermal unit, including a 10-hour thermal storage capacity.

Key Insights and Ideas

  • The Stability-Cost Trade-off: As solar-thermal capacity increases, the grid's voltage stability (TVRI) improves significantly at first but eventually hits diminishing returns. Meanwhile, investment costs continue to climb. The "optimal" point is where the stability gains no longer justify the massive capital expenditure.
  • CSP as a "Flexibility Provider": Unlike PV, solar-thermal plants can store heat. The study highlights how these plants store energy during peak PV hours (10:00–18:00) and discharge it during evening peaks (19:00–24:00), effectively stabilizing the UHVDC export line.
  • Deterministic Conversion: One of the technical hurdles in chance-constrained programming is dealing with probabilities. The authors successfully converted these probabilistic constraints into deterministic ones using distribution functions, making the model solvable via standard Mixed-Integer Linear Programming (MILP).

Key Results

Using the Gansu base as a model, the researchers found that the optimal configuration for solar-thermal capacity was 1,000 MW ($1 \times 10^6$ kW). At this level:

  • Voltage Stability: The TVRI reached 0.359, indicating strong voltage support during system faults.
  • Transmission Efficiency: The UHVDC transmission limit increased by 23.75%.
  • Renewable Integration: New energy consumption (reducing "curtailment") increased by 7.12%.
  • Peak Shaving: The peak-to-valley difference in the system was reduced by 29.96%.
  • Cost: The total comprehensive operation cost was calculated at 22.76 billion yuan.

Limitations and Future Work

While the paper provides a robust framework, a few areas remain for future exploration:

  • Simplified Investment Costs: The authors acknowledge that solar-thermal investment costs are complex; they simplified the model by focusing on four main subsystems (light field, power gen, heat absorption, and storage).
  • Storage Duration: The study fixed the thermal storage at 10 hours. Future research could optimize the duration of storage alongside the capacity of the plant.
  • Geographic Specificity: The results are highly dependent on the solar/wind profiles of Gansu. Applying this method to regions with different climates (e.g., coastal wind vs. desert solar) might yield different optimal configurations.

Summary updated

Abstract

Because of the randomness of wind power and photovoltaic (PV) output of new energy bases, the problem of peak regulation capability and voltage stability of ultra-high voltage direct current (UHVDC) transmission lines, we proposed an optimum allocation method of installed capacity of the solar-thermal power station based on chance constrained programming in this work. Firstly, we established the uncertainty model of wind power and PV based on the chance constrained planning theory. Then we used the K-medoids clustering method to cluster the scenarios considering the actual operation scenarios throughout the year. Secondly, we established the optimal configuration model based on the objective function of the strongest transient voltage stability and the lowest overall cost of operation. Finally, by quantitative analysis of actual wind power and photovoltaic new energy base, this work verified the feasibility of the proposed method. As a result of the simulations, we found that using the optimal configuration method of solar-thermal power stations could ensure an accurate allocation of installed capacity. When the installed capacity of the solar-thermal power station is 1 × 106 kW, the transient voltage recovery index (TVRI) is 0.359, which has a strong voltage support capacity for the system. Based on the results of this work, the optimal configuration of the installed capacity of the solar-thermal power plant can improve peak shaving performance, transient voltage support capability, and new energy consumption while satisfying the Direct Current (DC) outgoing transmission premise.
BibTeX
@inproceedings{wang2023optimal,
  author = {Yan Wang and Zhicheng Ma and Jinping Zhang and Qiang Zhou and Ruiping Zhang and Haiying Dong},
  title = {{Optimal Configuration Method for the Installed Capacity of the Solar-Thermal Power Stations}},
  booktitle = {Energy},
  year = {2023},
  doi = {10.32604/ee.2023.025668},
}
Figure