Alamdari, Pedram, Khatamifar, Mehdi, and Lin, Wenxian (2026) Multi-objective optimization of bladed cylindrical solar tower receivers using random forest surrogates and NSGA-II. International Journal of Heat and Fluid Flow, 122. 110724.

Preview

PDF (Published Version) - Published Version Available under License Creative Commons Attribution. Download (3MB) | Preview

Abstract

Bladed cylindrical receivers have the potential to reduce external heat-losses in solar tower concentrated solar power systems, but their added blades also increase surface area and aerodynamic loading. These competing effects create a coupled thermal–aerodynamic–geometric design problem that has not been fully addressed in previous optimization studies. This work develops a surrogate-assisted multi-objective optimization framework to identify receiver geometries that balance heat-loss reduction, aerodynamic performance, and geometric compactness. Machine-learning surrogate models were trained using a CFD database spanning multiple receiver temperatures, wind conditions, blade lengths, and blade numbers, achieving high predictive accuracy with cross-validated coefficients of determination of about 0.98 for both thermal and aerodynamic responses. The optimization results reveal a consistent trade-off structure across all operating conditions. Very compact designs exhibit both higher drag coefficients and higher total heat-loss coefficients, while highly extended designs reduce the total heat-loss coefficient but introduce excessive surface area and geometric complexity. Between these extremes, a stable intermediate design region provides the best balanced compromise. The surrogate predictions indicate that the knee-point candidates have approximately 43%–52% lower drag coefficients and 33%–40% lower total heat-loss coefficients than the compact design extreme. The proposed framework offers practical design envelopes for early-stage screening of bladed receiver concepts, while indicating that intermediate geometries should be further assessed using high-fidelity simulations before final design selection.

Item ID:	94529
Item Type:	Article (Research - C1)
ISSN:	0142-727X
Keywords:	Solar towers, Machine learning, Multi-objective optimization, NSGA-II, Random forest, Total heat transfer coefficient, Drag coefficient, Heat-loss
Copyright Information:	© 2026 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Date Deposited:	30 Oct 2026 04:41
FoR Codes:	40 ENGINEERING > 4012 Fluid mechanics and thermal engineering > 401204 Computational methods in fluid flow, heat and mass transfer (incl. computational fluid dynamics) @ 50% 40 ENGINEERING > 4012 Fluid mechanics and thermal engineering > 401299 Fluid mechanics and thermal engineering not elsewhere classified @ 20% 46 INFORMATION AND COMPUTING SCIENCES > 4611 Machine learning > 461199 Machine learning not elsewhere classified @ 30%
SEO Codes:	17 ENERGY > 1708 Renewable energy > 170805 Solar-thermal energy @ 30% 17 ENERGY > 1701 Energy efficiency > 170199 Energy efficiency not elsewhere classified @ 70%
	More Statistics