^{*}

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (

The effect of tropical cyclones on the turbulent flow over 2D continuous rolling hills was numerically investigated based on a field test analysis of the coastal region of Southeast China. A computational fluid dynamics (CFD) method was first developed and verified using previously published experimental results. Then two typical beneficial and destructive cyclone cases were studied above different locations of the hills. Results showed that the continuous hilly flow was much more drastic and variable than previously reported normal wind; the mean and turbulent magnitudes became the strongest around the hill top, with the maximum speed-up ratio, turbulence intensity and gust-speed ratio of 1.1, 0.32 and 1.6; the flow over lower hill was greatly affected by the nearby higher hills; the mean and fluctuating quantities were mostly smaller than the corresponding single hill case. These phenomena were considered to be related with the rather strong detachment and attachment of the cyclone flow around the two hills. In addition, the mean and fluctuating wind velocities were found to be underestimated by at least 20% if the widely accepted IEC standard equations were utilized, suggesting the necessity to supplement the field test analysis in the standard for more reasonable wind resource evaluation within the Southeast China coastal area.

It is well known that the wind resources in the coastal area of Southeast China are the most abundant in the country, with an average wind power density of more than 300 W/m^{2} and a utilization hour of more than 6000 h. The economy is rather developed in the region, but the electric power shortage problem is becoming increasingly serious, which has led to the fast development of wind energy resources in recent years. However, this region is often subject to strong tropical cyclones, posing a serious threat to the normal construction of wind farms. Considering the Beaufort scale [

Generally speaking, beneficial cyclones (wind scale 6–7) can be very profitable for offshore wind farms. For example, the electricity generated by the beneficial cyclone “Morak” during 6 days accounted for 2/3 of total electricity of the wind farms in one month when it passed by Zhejiang and Jiangsu Provinces in 2009. In contrast, a destructive cyclone (wind scale greater or equal than 10) may frequently destroy wind farm components such as blades and towers, and even the whole turbine system, resulting in a great loss to the local wind developers. Typical examples include typhoon “Cuckoo” in 2003, typhoon “Pearl” in 2006, typhoon “Saomai” in 2006 and Typhoon “Megi” in 2010, and so on. Therefore, to fully utilize beneficial cyclones as well as guard against the defensive and the destructive ones, in depth systematic research on the characteristics of turbulent flow need to be conducted. Nevertheless, due to lack of well-documented field test data, little work has been done in the past.

On the other hand, large land areas in the Chinese coastal region are covered by continuous rolling hills, causing flow separation behind hills and recirculation in valleys, which may further enhance the influence of tropical cyclones on the construction and the operation of a wind farm. For instance, according to news reports, a total of 13 turbines in the Shanwei wind farm in Guangdong Province were essentially destroyed by the typhoon “Cuckoo” in 2003, among which nine turbines were located on rolling hills with relative high altitude. This makes it very necessary to analyze the role that this kind of complex terrain plays on the cyclone wind flow.

Previously, under normal wind conditions, many researchers have focused their interests on 2D and 3D continuous or single terrain cases to understand the fundamental flow phenomena, and this work can be roughly divided into experimental tests and numerical simulations. For the former, experiments were often carried out in wind tunnels to measure the distributions of mean and fluctuating magnitudes, e.g., mean velocity, fractional speed-up ratio, turbulence intensity, turbulent kinetic energy and Reynolds stress, using Particle Image Velocimetry (PIV), hot wires or Pitot tubes,

Furthermore, the design of the modern wind turbines mostly abides by some international standard,

To cover the aforementioned points, systematic numerical and experimental studies need to be conducted to examine the effect of the tropical cyclones on the turbulent flow over complex terrains within the southeast coastal area in China. In the first stage, the mean and fluctuating characteristics of turbulent cyclone flow over 2D double hills were numerically investigated in this paper. Considering its complicated situation, the experimental work will be carried out in the near future. To this end, we first developed a CFD method and then verified it by comparing the numerical results with the experimental ones by Kim

The CFD method was built on the basis of the Reynolds-averaged Navier-Stokes equations using the commercial FLUNT code. In order to simulate well the concerned atmospheric boundary layer, a classical two-equation high Reynolds number _{μ}_{ε}_{1}, _{ε}_{2}, _{k}_{ε}

Deploying a method similar to that of Zhang [_{*}was equal to the surface friction velocity _{τ}_{0}. For the viscous sub-layer:
^{+} can be expressed as
_{τ}_{p}

Note that the hybrid RANS/LES method, as reference [

For the fully turbulent region:

As mentioned before, the CFD scheme was developed based on the commercial FLUNT software. The inlet flow was driven by the external force, following the given equations of the kinetic energy ^{+} < 5, so that the grid density was greatly enhanced near the hill surfaces to deal with the complex flow separation phenomena. Based on the grid independence test, the average grid resolution of 1,030 × 590 was used. In addition, the finite-volume discretization in the boundary-fitted coordinates, the SIMPLEC algorithm to couple pressure and velocity on a non-staggered grid and the QUICK scheme for the convection terms, were utilized to further improve the accuracy of the calculation.

It was also worth to mention that, using the CFD scheme, the construction of a good equilibrium flow in the present paper was not an easy task and we made a lot of efforts by adjusting our CFD scheme. Even so, the horizontal equilibrium of the resulting flow was not perfectly achieved, especially for the long-distant turbulent evolution along the streamwise direction, but it was not the case for the regions around the presently investigated double hill models under all incoming conditions, which had been tested in terms of mean and turbulent quantities (not shown) before the double hill models were put into the computation domain.

To verify the developed CFD method, we used the same situation,

The longitudinal log-law velocity distribution,
_{*} = 0.33 m/s and the aerodynamic roughness height _{0} = 0.05 mm, was set at the inlet of the computation zone, mimicking a neutrally stratified boundary layer. The corresponding kinetic energy _{z}_{z}

Four double hill models, designated as S3H4-S3H7, S3H7-S3H4, S5H4-S5H7 and S5H7-S5H4, were utilized, which were composed of four single 2D symmetry hill models with various heights and slopes, individually labeled as S3H4, S5H4, S3H7 and S5H7. Each single hill has a profile given by:

The coordinates

For the purpose of verification, the comparisons of CFD computation with the experimental results by Kim

From these results, the developed CFD algorithm was verified and it would be deployed in the following sections, since the turbulent flow of the cyclone is of similar order of Re (based on hill height, around 10^{5}) as those in Kim

As mentioned before, it was very necessary to investigate the effect of the cyclones on the turbulent flow over rolling hills based on the field test data. For this reason, we analyzed the cyclone data within southeast coastal area from 1949 to 2010 [

In addition, the 10 min-field data for typical 29 cyclones and 88 anemometer towers since 2003, was also analyzed statistically, leading to a curve-fitted 5th interpolation function representing the variation of the mean turbulence intensity (_{hub}_{hub}_{hub}

From these analyses, the typical beneficial and destructive cyclones, corresponding to _{hub}_{z}_{z}_{z}_{0} (=0.4) is the roughness height for the incoming flow and _{z}

Using the similar method in Sørensen [_{z}_{z}

The profile of the turbulence intensity distribution described by _{70}at _{hub}

Since separated flow phenomena over continuous hills are very common in practice, two double hill cases with higher slope, _{0} was also assumed to be 0.4 to represent the typical feature of vegetations over model hill surface.

This is associated with the increased turbulence at the downwind S5H4 hill crest from the separation of the upwind S5H7 hill, leading to the effective impairment of the detachment from the downwind S5H4 hill. Correspondingly, the flow reattaches to the ground much quicker behind downwind S5H4 hill than that behind downwind S5H7 hill, attributed to the fact that the flow momentum for the latter is much smaller than the former and thus the reattachment bubble may extend further downstream until the flow returns to a fully turbulent state. Here the mean reattachment location was determined to be the nearest grid point to the wall where

_{z′}_{0} is the reference wind speed at the same height above the flat surface. In fact,

Because of the influence of the neighboring hill, the speed-up ratio over the double hill top is found to be smaller than that at the single hill top. The momentum loss due to the adverse pressure gradient and viscous dissipation in the separation regions between the hills and behind the second hill are thought be responsible for this, which may be also deduced from the results in _{z′}

Moreover, the typical profiles of the Reynolds stress
_{z′}

_{z′}_{z′}_{z′}_{z′}_{z′}_{z′}_{z′}_{z′}_{z′}

Considering the importance in the normal wind resource evaluation, the mean gust speed _{gz′}_{gz′}_{g}_{0} represents the gust speed at same height above hill surface and flat ground, respectively.

As we know, the wind characteristics of tropical cyclones are not explicitly regulated in the IEC standard, the most widely accepted one for wind turbine design. Instead, people often deploy the existing IEC standard equations to approximately evaluate the cyclone flow. This undoubtedly makes it very difficult to obtain a cost effective and safe design under these conditions. To illustrate the difference in the wind evaluation results based on IEC standard and our field test analysis in Section 3.2, the vertical profiles of flow over the top of the double hill were compared as a typical example, where the wind energy utilization was regularly paid most attention.

It is worth mention that, for IEC case, the velocity profiles for the incoming beneficial and destructive cyclone flow were still expressed using _{z}_{g}_{b}_{70} based on

Here _{ref}_{ave}

_{z′}_{z′}_{gz′}_{z′}_{z′}

_{z′}_{z′}_{70} is about 0.26–0.31, which are 15.4%–27.3% higher than those calculated using IEC NTM and EWM models.

_{gz′}_{z′}_{gz′}_{hub}_{z′}_{z′}_{gz′}

It was easy to note that, using the IEC standard, the magnitudes of the wind velocity, the turbulence intensity and the gust speed were evidently lower than those based on the field test analysis, especially for the destructive cyclone case. In fact, the turbulent flow separation around the double hills tended to be much stronger for the destructive cyclone case than the beneficial cyclone case, as illustrated in

This paper presented a numerical investigation of the effect of typical Chinese tropical cyclones on the turbulent flow over the 2D continuous hills using a developed CFD method. In addition, the representative results were also compared with those based on the IEC standard equations. Some conclusions could be drawn as follows:

The mean and fluctuating characteristics of double hilly flow due to tropical cyclones are generally much stronger and more fluctuant than the previously reported normal wind conditions;

Although rather intense flow separation occurs between and behind the two hills, the speed-up phenomena are very similar for beneficial and destructive cyclone case, resulting in: (a) the relative powerful speed-up appears at the crest of higher hill, with the maximum fractional speed-up ratio

The turbulent quantities of destructive cyclones are much larger than those of beneficial cyclones in magnitude, but they tend to have similar features, mainly attributed to increased turbulence and momentum transport induced by the detachment of the hilly flow: (a) the kinetic energy _{z′}_{z′}

In contrast to the field test analysis, both the averaged velocity and the turbulence strength over the representative hill tops are underestimated by up to more than 20% based on the IEC standard, which makes it very necessary to supplement the standard using the statistic field test formula for more reasonable wind evaluation within the southeast coastal area in China;

Using a similar method to Sørensen [

This work was supported by the National Natural Science Foundation of China (Grant No. 51076153).

The authors declare no conflict of interest.

Schematic computational domain of the wind flow over 2D double hill.

Vertical profiles of mean longitudinal velocity over the double hills: (

Comparison of predicated and measured turbulence kinetic energy profiles over single hill S3H7 (symbol, experimental data; line, CFD results).

Distributions of streamwise velocity (

Distributions of streamwise velocity (

Variation of fractional speed-up ratio with height

Variation of Reynolds stress with height

Variation of turbulence intensity with height

Variation of gust speed ratio with height

Variation of wind velocity with height

Variation of turbulence intensity with height

Variation of gust speed with height

Classification of tropical cyclones.

Beneficial | Tropical depression | 10.8–17.1 | 6∼7 | 15.3∼24.3 |

Defensive | Tropical storm | 17.2–24.4 | 8∼9 | 24.4∼32.0 |

Destructive | Severe tropical storm | 24.5–32.6 | 10∼11 | 32.1∼42.8 |

Typhoon | 32.7–41.4 | 12∼13 | 42.9∼54.4 | |

Severe typhoon | 41.5–50.9 | 14∼15 | 54.5∼66.8 | |

Super typhoon | 51.0–61.2 or above | 16∼17 or above | 66.9∼80.4 or above |