A reduced order modelling methodology for external cylindrical concentrated solar power central receivers

The use of supercritical carbon dioxide (sCO2) power cycles for concentrated solar power (CSP) applications is becoming increasingly attractive since these cycles may offer lower capital costs and increased thermal efficiency. However, there are currently no utility-scale sCO2-CSP tower plants in operation. Therefore, to aid in the design and analysis process, there is a need to develop sufficiently accurate and computationally inexpensive models for such plants. This dissertation presents a reduced order modelling methodology for external cylindrical concentrated solar power central receivers. The methodology is built on a one-dimensional thermofluid network to model the heat transfer through the tube walls, coupled to a fluid flow network of the solar salt flowing inside the tubes. This is combined with a neural network surrogate model to determine the radiative heat flux impinging upon the tube surfaces. The receiver geometry is discretized along the height and around the circumference and each increment is represented by an equivalent thermal resistance network that represents the heat transfer within the tube walls. The heat transfer network parameters are calibrated using a detailed computational fluid dynamics model, which enables the calculation of the maximum tube wall temperatures. The heat transfer network is connected to the fluid flow network that solves the mass, energy, and momentum balance equations to determine the mass flow rates, pressure drops and temperature distributions. The radiative heat flux profile impinging on the receiver is typically calculated for a specific location and specific time of the day using a tool such as SolarPILOT. However, this can be computationally expensive since the central tower is surrounded by thousands of individual heliostats that are all sources of radiative flux, which depends on the position relative to the sun and relative to the receiver, as well as the direct normal irradiation (DNI) at that location and time. To reduce the associated computational expense, a multilayer perceptron (MLP) surrogate model is developed that allows the prediction of the flux profile for a range of plant configurations and atmospheric conditions at a specific location. The application of the methodology is demonstrated via a case study. The methodology may be used in future studies where sCO2-CSP tower plants are investigated, especially those with an interest in the detail design and analysis of the central receiver.

Grantor

Department of Mechanical Engineering

Year dc:date.issued

2023

Author dc:creator

• Heydenrych, James

Advisors dc:contributor.advisor

• Rousseau, Pieter
• Du Sart Colin

• Engineering

Handle dc:identifier.uri

http://hdl.handle.net/11427/39520

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/39520