TY - JOUR
T1 - Artificial Neural Network–Particle Swarm Optimization (ANN-PSO) Approach for Behaviour Prediction and Structural Optimization of Lightweight Sandwich Composite Heliostats
AU - Fadlallah, Sulaiman O.
AU - Anderson, Timothy N.
AU - Nates, Roy J.
PY - 2021/12/1
Y1 - 2021/12/1
N2 - The necessity to diminish the heliostats’ cost so that central tower concentrating solar power (CSP) systems can stride to the forefront to become the technology of choice for generating renewable electricity is obliging the industry to consider innovative designs, leading to new materials being implemented into the development of heliostats. Honeycomb sandwich composites offer a lightweight but stiff structure that appear to be an ideal substitute for existing heliostat mirrors and their steel supporting trusses, avoiding large drive units and reducing energy consumption. However, realizing a honeycomb sandwich composite as a heliostat, among a multitude of possible combinations can be tailored from, that delivers the best trade-off between the panel’s weight reduction (broadly equates to cost) and structural integrity is cumbersome and challenging due to the complex nonlinear material behaviour, along with the large number of design variables and performance constraints. We herein offer a simulation–optimization model for behaviour prediction and structural optimization of lightweight honeycomb sandwich composite heliostats utilizing artificial neural network (ANN) technique and particle swarm optimization (PSO) algorithm. Considering various honeycomb core configurations and several loading conditions, a thorough investigation was carried out to optimally choose the training algorithm, number of neurons in the hidden layer, activation function in a network and the suitable swarm size that delivers the best performance for convergence and processing time. Carried out for three case scenarios, each with different design requirements, the results showed that the proposed integrated ANN-PSO approach provides a useful, flexible and time-efficient tool for heliostat designers to predict and optimize the structural performance of honeycomb sandwich composite-based heliostats as per desired requirements. Knowing that heliostats in the field are not all subjected to the same wind conditions, this method offers flexibility to tailor heliostats independently, allowing them to be made lighter depending on the local wind speed in the field. This could lead to reductions in the size of drive units used to track the heliostat, and the foundations required to support these structures. Such reductions would deliver real cost savings, which are currently an impediment to the wider spread use of CSP systems.
AB - The necessity to diminish the heliostats’ cost so that central tower concentrating solar power (CSP) systems can stride to the forefront to become the technology of choice for generating renewable electricity is obliging the industry to consider innovative designs, leading to new materials being implemented into the development of heliostats. Honeycomb sandwich composites offer a lightweight but stiff structure that appear to be an ideal substitute for existing heliostat mirrors and their steel supporting trusses, avoiding large drive units and reducing energy consumption. However, realizing a honeycomb sandwich composite as a heliostat, among a multitude of possible combinations can be tailored from, that delivers the best trade-off between the panel’s weight reduction (broadly equates to cost) and structural integrity is cumbersome and challenging due to the complex nonlinear material behaviour, along with the large number of design variables and performance constraints. We herein offer a simulation–optimization model for behaviour prediction and structural optimization of lightweight honeycomb sandwich composite heliostats utilizing artificial neural network (ANN) technique and particle swarm optimization (PSO) algorithm. Considering various honeycomb core configurations and several loading conditions, a thorough investigation was carried out to optimally choose the training algorithm, number of neurons in the hidden layer, activation function in a network and the suitable swarm size that delivers the best performance for convergence and processing time. Carried out for three case scenarios, each with different design requirements, the results showed that the proposed integrated ANN-PSO approach provides a useful, flexible and time-efficient tool for heliostat designers to predict and optimize the structural performance of honeycomb sandwich composite-based heliostats as per desired requirements. Knowing that heliostats in the field are not all subjected to the same wind conditions, this method offers flexibility to tailor heliostats independently, allowing them to be made lighter depending on the local wind speed in the field. This could lead to reductions in the size of drive units used to track the heliostat, and the foundations required to support these structures. Such reductions would deliver real cost savings, which are currently an impediment to the wider spread use of CSP systems.
KW - ANN
KW - Fluid–structure interaction (FSI)
KW - Heliostat
KW - Honeycomb
KW - PSO
KW - Sandwich composite
UR - http://www.scopus.com/inward/record.url?scp=85114180939&partnerID=8YFLogxK
U2 - 10.1007/s13369-021-06126-0
DO - 10.1007/s13369-021-06126-0
M3 - Article
AN - SCOPUS:85114180939
VL - 46
SP - 12721
EP - 12742
JO - Arabian Journal for Science and Engineering
JF - Arabian Journal for Science and Engineering
SN - 1319-8025
IS - 12
ER -