Investigation of the wind load on heliostats in stow position

Abstract

In recent years there has been an increasing effort to lower the capital costs of concentrating solar thermal (CST) power tower (PT) plants to make their levelised cost of electricity (LCOE) more competitive with base-load energy systems. The field of heliostats contributes the most to the capital cost of a PT plant, hence the costs of manufacturing and installation of heliostat components can be reduced by optimising their structural design to withstand the maximum wind loads during high-wind conditions in the stow position. Heliostats are moved from operational positions and aligned parallel to the ground in the stow position during periods of high wind speeds to minimise the effect of fluctuating pressures and unsteady forces that can lead to structural failure. While the effects of temporal variation of turbulence on heliostat design wind loads have been investigated for operational positions, further knowledge of the spatial distribution of turbulence in the frequency domain of the atmospheric boundary layer (ABL) and the dynamic wind loads in stow position needs to be developed. The purpose of this thesis is to develop an understanding of the turbulence effects in the ABL that can lead to maximum wind loads during gusty high-wind conditions. This is achieved by studying the dynamic effects of the wind arising from turbulence characteristics in the lowest 10 m of the ABL and their influence on the peak wind loads on stowed heliostats. With this knowledge, reduction of the design wind loads in the stow position can allow the optimisation of the size and cost of heliostats with respect to the turbulent flow approaching them. One of the principal factors in the design criteria for a heliostat field is the design wind speed at which the heliostats are moved from operating positions to the stow position. The sensitivity of both the capital cost of heliostats and the LCOE of a PT plant was investigated by developing a statistical model assuming quasi-steady wind loads and simplified cost-area proportionality exponents. A parametric study using wind speed and solar irradiation data showed that a significant reduction in the design wind speed at windy sites can be achieved with only a small reduction in the capacity factor, thus offering potential to reduce the cost of heliostats and the LCOE of a PT plant. Optimal heliostat sizes were found to decrease significantly with increasing stow design wind speed, such as from 50 m2 to 25 m2 when the design wind speed increases from 10 m/s to 20 m/s. Velocity measurements in the lowest 25 m of a low-roughness atmospheric surface layer (ASL) were analysed to further understand the relationships between turbulence characteristics and their effect on the velocity gust factor that is widely used in design wind codes to estimate the peak wind loads on physical structures. It was found that the peak wind speeds associated with low-frequency gusts were under-predicted by 5% at heights below 10 m in desert terrain. Hence, simplified gust factor methods and semi-empirical turbulence models can under-estimate the peak wind loads, which are proportional to the square of peak wind speed, by 10% at heights below 10 m in low-roughness terrains where heliostat fields are located. Subsequently, an experimental investigation was carried out to identify the effects of turbulence characteristics in the lower ABL on the peak wind loads on stowed heliostats. The temporal and spatial turbulence characteristics were characterised over a wide range of turbulence intensities and integral length scales (𝐿𝑢 𝑥 ) in a simulated part-depth ABL using two geometries of spires and roughness elements. A range of square heliostat mirror chord lengths (𝑐) was used to investigate the effect of the length-scale-to-chord-length ratio (𝐿𝑢 𝑥 /𝑐) on the peak wind loads on an isolated heliostat in stow position and on a second downstream heliostat in a tandem configuration. It was found that both the peak lift coefficient and the peak hinge moment coefficient on the isolated heliostat increased linearly as 𝐿𝑢 𝑥 /𝑐 increased from 2.5 to 10 and at longitudinal turbulence intensities greater than 10%. In contrast, the peak lift forces and hinge moments on a second downstream heliostat were up to 30% lower than those on an isolated heliostat at 𝐿𝑢 𝑥 /𝑐 of 10. Peak energies of measured pressure spectra were an order of magnitude smaller on the downstream heliostat than the isolated heliostat and showed a shift to higher frequencies corresponding to smaller vortices. Peak wind loads on the downstream heliostat were observed to increase to 10% above those on an isolated heliostat as the spacing between the heliostat mirrors in tandem, defined by the gap ratio 𝑑/𝑐, was lowered below 1. In contrast, peak wind loads on a downstream heliostat were up to 30% lower than an isolated heliostat at 𝑑/𝑐 of approximately 2 at intermediate field densities. When stowing the heliostat mirror at a range of elevation heights, it was found that both peak lift forces and hinge moments can be minimised at an elevation axis height equal to or less than half of the mirror chord length. Hence, optimisation of the mirror chord length and the elevation axis height of the stowed heliostat for the approaching turbulence can significantly reduce design wind loads for high-wind events in the atmospheric surface layer and the cost of manufacturing the heliostat components. The results of the research in the current thesis can be used to optimise the spacing between stowed heliostats at different field densities and the critical scaling parameters of the heliostat, based on known characteristics of the approaching turbulence in a given ABL. Peak wind loads on isolated heliostats in stow position are likely to be over-designed for in-field heliostats positioned in lowdensity regions of a field, thus offering the potential to manufacture in-field heliostats from lower strength, lighter and cheaper materials. Consequently, uniform designs for heliostats in the first two rows of a field need to consider the critical wind load that leads to failure modes to determine the pylon length and mirror chord length of the heliostat.