Session: 02-01: AI for Energy Sustainability I

Paper Number: 155317

155317 - Optimizing the Performance of Liquid-Based Medium-Temperature Volumetric Solar Thermal Receivers Using Genetic Algorithms: A Step Towards Digital Twinning 

Abstract: 

Concentrated solar power (CSP) systems have traditionally utilized tubular receivers that are limited by low capture efficiencies and low temperature uniformity, especially at temperatures above 400◦C, where radiative heat losses start to dominate. These drawbacks of tubular receivers have led to high levelized cost of electricity due to substantial thermal losses at high temperatures. Volumetric receivers have been recently investigated as a promising alternative to tubular surfaces. Unlike traditional tubular receivers absorbing solar energy on a surface, volumetric receivers absorb incoming solar energy through a semi-transparent medium that is directly irradiated upon in an open-tank configuration, allowing a volumetric absorption of solar energy. Their design leads to a high capture efficiency, more uniform temperature distribution, and an increase in tolerability of higher solar fluxes, all of which are important for improving the overall performance of a CSP plant. The behavior of volumetric receivers is still being investigated and attempts to accurately predict their thermofluid response, which dictate the performance of these receivers, has proven to be challenging. The interactions between radiation, natural convection, and volumetric heating over extended periods of time leads to a highly complex and dynamic environment. A mechanistic model based on the fundamental physics of the system was developed to capture these heat transfer interactions in liquid-based volumetric receivers under various input conditions. Later, experimental validation under a 6.5kW solar simulator demonstrated that the model accurately captured receiver thermofluid response within reasonable accuracy for short durations of time (<7.5 minutes). However, the optimization of system performance in terms of capture efficiency and avoidance of hot spot development inside the receiver was not investigated in previous studies. The system performance and design optimization is essential to advance these next-generation clean energy systems to a high technology readiness level. This work aims to bridge this critical gap by employing a systemic approach based on the theory of evolution. Previous studies found five critical parameters related to the performance of volumetric receivers, indicating that optimizing these variables could maximize efficiency and maintain temperature uniformity inside the receiver under various operating conditions. These five key parameters included: the attenuation coefficient of semi transparent media k, the incoming solar flux q0, the depth of the volumetric receiver y, top surface emissivity ϵ, and heating time t. To optimize the performance, we employed the Evolutionary Computing approach, which is a branch of computational intelligence focused on global optimization through algorithms influenced by natural biological evolution. Specifically, a metaheuristic Genetic Algorithm (GA) was implemented in this study to optimize the five identified parameters effectively. GAs mimic natural biological evolution by continuously selecting, recombining, and mutating solutions to find the most optimal set of parameters. By simulating a survival of the fittest process, GAs can efficiently explore large and complicated solution spaces, making them ideal for optimizing the performance of volumetric receivers. Here a Non-Dominated Sorting Genetic Algorithm (NSGA-II) was used that utilizes additional concepts of non-dominated sorting, crowding distance calculations, and elitism to further improve the optimization process. This specific algorithm helps maintain diverse, high-quality solutions as the algorithm goes through different generations of key parameters. To implement this algorithm for optimizing volumetric receivers and their multivariable systems, MATLAB’s Global Optimization Toolbox was used which includes the gamultiobj solver, a variant of the NSGA-II. Using this algorithm, multiple solutions were found with varying trade-offs between efficiency and avoidance of hot spots inside the receiver. The ability to find multiple optimal solutions is particularly useful, as it provides a range of possible outcomes that can be adjusted to fit different operational conditions, ultimately improving the performance of volumetric receiver systems. More details about the MATLAB model and the working of genetic algorithms in the optimization process will be discussed in the upcoming ES 2025 paper.

Presenting Author: Thasanka Kandage University of Alberta

Presenting Author Biography: I am an undergraduate student at the University of Alberta, pursuing a major in Software Engineering. I am currently working as a Research Assistant under the supervision of Dr. Muhammad Taha Manzoor, where I am involved in research related to genetic algorithms and the creation of AI models for optimizing complex systems and predicting experimental outcomes.