Towards prediction of wind load on pylons for a neutral atmospheric boundary layer flow over two successive hills

Abstract

In strong wind situations for a neutral atmospheric boundary layers, wind loads on pylons depend on both the mean velocity and the turbulence fluctuations, predicted by analytical models only for simple topographies. For complex topographies, only experiments or numerical simulations give access to the local flow conditions around the pylon. We present experimental and numerical results using a RANS model for the parametric study of the flow near two successive hills of height H and width L for varying distance λ between the tops. We focus on the modification of flow conditions on the second hill by the presence of the first hill. For $\lambda /H\ge 19.2$, the first hill does not modify the flow structure around the second hill. For decreasing values of $\lambda /H$ below 19.2, a secondary turbulence intensity peak appears on the frontal part of the second hill. It is associated with the impact of the mixing layer generated at the top of the first hill, whose development is well described by analytical models of the planar mixing layer. These analytical models yields simple estimations of the turbulent kinetic energy k for the pylon’s wind load calculations. Different hill slopes and height ratios are also investigated.

Introduction

For pylon design, calculations take into account the load on the pylon. The wind load is a major contribution to the total load, in particular during strong wind atmospheric events, and yet, the hardest to predict and describe when flow occurs over complex topographies.

In strong wind situations, neutral atmospheric boundary layer conditions are generally observed. For such flows over shallow topographies, Jackson and Hunt (1975) proposed an analytical model to predict the mean flow perturbation of an upstream log-law velocity profile. Hunt et al. (1988) then improved this model and compared successfully their predictions with field measurements (Mason and Sykes (1979); Mason and King (1985); Taylor and Teunissen (1987)) and wind tunnel measurements (Gong and Ibbetson (1989); Finnigan et al. (1990)) for flows over hills. Shallow hills are generally defined as hills with slope S below 0.2, where S is the ratio between the hill’s height H and its width L (taken at mid-height). It is related to the observation made by Finnigan (1988) on a collection of field and wind tunnel data that flow separation downstream of the hill occurs for hills with $S>0.3$ (so called steep hill) whereas no separation is detected for hills with $S<0.2$. The success of the analytical model of Hunt et al. (1988) for shallow topographies motivated its integration into Eurocode formulae for the prediction of topographic effects on local wind velocity amplification.

For steep hills, the analytical model of Hunt et al. (1988) is generally applied with caution, and for the upstream face only. This approach is validated by several studies that have been conducted for flows around two dimensional isolated rough hills using Reynolds-averaged Navier-Stokes (RANS) equations with a k-epsilon turbulence model (Castro and Apsley (1997), Griffiths and Middleton (2010) and Safaei Pirooz and Flay (2018)) or a Reynolds stress model (Loureiro et al. (2008)), all in good agreement with experiments.

For two successive two-dimensional steep hills, the flow conditions on the second hill are still harder to predict, even for shallow hills. Previous studies like Bitsuamlak et al. (2006), by comparing their numerical simulations with the experimental data of Carpenter and Locke (1999), have shown that RANS equations predict well the mean velocity field. In the case of sinusoidal smooth hills,Lee et al. (2002) also use both RANS numerical simulations and experiments to analyze the impact of the first hill flow separation on the surface pressure of the second hill. They observe that even at a distance $\lambda =5H$ between the hills crest, the surface pressure at the second hill crest is about $25-30\text{\%}$ smaller than at the first hill crest. The difference of surface pressure then decreases as λ increases. Kim et al. (1997) also investigated experimentally the two hill configuration by varying the height of the first hill to observe the impact on the flow over the second hill. For a first hill of height $H_1=2H_2$ (where $H_2$ is the height of the second hill) and a slope of $S=0.5$, the size of the separation flow area behind the second hill is decreased by $20\text{\%}$ from that of the same isolated hill.

Several experimental studies have been conducted to study the separation flow behind the first hill and its disturbance of the flow over the second hill. Li et al. (2017) use a PIV system in a wind tunnel to investigate the velocity and the turbulence intensity around two-dimensional successive smooth hills for installation of wind turbines. For a slope of $S=0.5$, the measurements show an important production of the turbulent energy in the separation flow area of the first hill. Ferreira et al. (1991) present an experimental investigation of interaction between two 2D smooth hills by varying the distance λ. For a slope $S=0.5$, Ferreira et al. (1991) observe an increase of maximum velocity at the top of the second hill as the distance λ increases. The size of the separation flow area of the first hill tends to grow as both hills approach each other.

In this study, we present a parametric investigation of rough two-hills flow configurations using a RANS $k-\epsilon$ model and experiments. The experimental set-up and numerical method are presented in section 2. The flow structure and the comparison between the experiments and the numerical simulations are presented and discussed in section 3. Results of the parametric investigation performed with numerical simulations are presented and discussed in section 4. In section 5, the ability of a simple plane mixing layer model to explain and predict flow caracteristics in the vicinity of the second hill is investigated. Conclusion and perspectives are drawn in section 6.