Benchmarking Derivative-Free Optimization Solvers for CSP Plant Design Using the SOLAR Simulator
- 1Group for Research in Decision Analysis
- 2Polytechnique Montréal
- 3Hydro-Québec
Abstract
This work presents a case study where four well-known derivative-free solvers are benchmarked on several instances based on the solar suite of problems: an easy to use, freely available and open-source Concentrated Solar Power (CSP) plant simulator. The solar simulator provides a collection of ten test problems designed to reflect real-world optimization challenges. The design of CSP plants involves solving complex and often computationally expensive optimization problems. These problems rely on highly diverse models which may be difficult to interface with optimization solvers. As a result, standardized benchmark problems for optimization solvers in the CSP design field are scarce. Such benchmarks are essential for selecting the most appropriate solver for a given optimization problem. The solar problems model numerous components of the power plant, ranging from the design of the heliostat field to the choice of turbine.
Related
- Neural Network Solvers for Nonlinear ODEs: Benchmarking MLPs and B-Spline/Chebyshev KANs
- MUTACO: Simulator-Efficient Response Matching in Stochastic Nonlinear Systems via Compatibility-Aware Sequential Design
- Differentiable Schwarz-Coupled Solvers for Thin Boundary Layer Transport: Inverse Modeling via Agentic AI Optimization
- CFL-aware edge design for graph neural solvers of hyperbolic transport
- Reducing the arithmetic cost of quantum derivative pricing with language-model circuit design
BibTeX
@article{lebeuf2026benchmarking,
title = {{Benchmarking Derivative-Free Optimization Solvers for CSP Plant Design Using the SOLAR Simulator}},
author = {Xavier Lebeuf and Sébastien Le Digabel and Philippe Brillon and Miguel Diago and Charles Audet and Christophe Tribes},
journal = {SolarPACES Conference Proceedings},
year = {2026},
month = oct,
url = {https://www.tib-op.org/ojs/index.php/solarpaces/article/view/2973},
note = {CC BY},
}
RIS
TY - JOUR
TI - Benchmarking Derivative-Free Optimization Solvers for CSP Plant Design Using the SOLAR Simulator
AU - Xavier Lebeuf
AU - Sébastien Le Digabel
AU - Philippe Brillon
AU - Miguel Diago
AU - Charles Audet
AU - Christophe Tribes
PY - 2026
DA - 2026/10/23
JO - SolarPACES Conference Proceedings
UR - https://www.tib-op.org/ojs/index.php/solarpaces/article/view/2973
AB - This work presents a case study where four well-known derivative-free solvers are benchmarked on several instances based on the solar suite of problems: an easy to use, freely available and open-source Concentrated Solar Power (CSP) plant simulator. The solar simulator provides a collection of ten test problems designed to reflect real-world optimization challenges. The design of CSP plants involves solving complex and often computationally expensive optimization problems. These problems rely on highly diverse models which may be difficult to interface with optimization solvers. As a result, standardized benchmark problems for optimization solvers in the CSP design field are scarce. Such benchmarks are essential for selecting the most appropriate solver for a given optimization problem. The solar problems model numerous components of the power plant, ranging from the design of the heliostat field to the choice of turbine.
SN - 2751-9899
ER -