Session: 11-01: Process Heat for Desalination and Industrial Decarbonization

Paper Number: 121387

121387 - Application of Multi-Objective Bayesian Optimization to Elucidate the Trade-Off Between the Solar Fraction and Cost of Parabolic Trough Solar Industrial Process Heat With Thermal Energy Storage 

Abstract: 

Decarbonizing process heat in the industrial and building sectors is the next frontier in the energy transition.  With over 50% of industrial processes requiring process heat at temperatures below 300℃, parabolic trough concentrating solar collectors and other moderately concentrating solar systems hold out promise as technology to usher in a transition away from fossil-based heat sources and towards sustainable ones.  Such solar concentrating systems are commercially-available and operate efficiently at 300℃ by concentrating solar beam radiation around 100 times and minimizing heat loss with an evacuated tube solar absorber.  Furthermore, with integrated thermal energy storage parabolic trough solar collector plants can increase their capacity factor beyond the ~25% limit imposed by the diurnal variability of the solar resource.  

Prior techno-economic studies of solar industrial process heat have focused on estimating its levelized cost and/or solar fraction considering the impacts of geospatial variability in the solar resource, weather, and available land and the hourly process heat demand with the aim of comparing solar industrial process heat to alternatives.  With a focus on geospatial factors and load characteristics, prior studies consider a single or limited number of parabolic trough collector plant designs.  For example, it is common that prior studies consider just two plant designs, one with a fixed size of the solar field without thermal energy storage and one with a larger solar field size with thermal energy storage fixed in size and typically on the order of 6 hours.  (The plant stores enough energy that it could output at its design capacity for 6 hours when fully charged).  

In the present work, we demonstrate how to extend prior techno-economic modeling of solar industrial process heat to incorporate optimization of the plant design for the competing objectives of minimum levelized cost and maximum solar fraction considering several design parameters, including the size of the solar field and the size of the thermal energy storage system.  A multi-objective Bayesian optimization (MOBO) algorithm is employed in conjunction with the System Advisor Model (SAM) parabolic trough collector plant model to guide the selection of design parameters towards those that lead to Pareto solutions, solutions for which the solar fraction cannot be improved by changing the design parameters without also increasing the levelized cost of the process heat.  The MOBO algorithm is implemented to find Pareto solutions for the design of parabolic trough collector solar process heat plants located at sites in the continental United States of America with an exceptional, average, and poor concentrated solar resource.  The locus of Pareto solutions elucidate a highly non-linear trade-off between the levelized cost of solar industrial process heat and solar fraction at all three sites.  The Pareto solutions also show that higher solar fractions are possible at a levelized cost less than $0.2/kW-hr_th than has been previously reported for solar industrial process heat.

In addition to identifying the Pareto solutions for three representative locations, we show that the MOBO algorithm is more computationally efficient at identifying Pareto solutions than alternative algorithms such as the genetic algorithm.  As such, it is the preferred algorithm for incorporating into future techno-economic analyses.

Presenting Author: Mario Ramos Valparaiso University

Presenting Author Biography: Mario is a rising senior at Valparaiso University (Valpo) where he is studying mechanical engineering in pursuit of a Bachelor's of Science in Mechanical Engineering. He joined the Solar Energy Research Group at Valpo in the summer of 2023. He is a recipient of the Richardson Scholarship.

Authors:

Mario Ramos Valparaiso University
Jesse M Sestito Valparaiso University
Jonathan Ogland-Hand Carbon Solutions, LLC
Nathan Holwerda Carbon Solutions, LLC
Luke J Venstrom Valparaiso University