Interpretación de resultados

Downbursts are localized events that have a wind field significantly different than the fields associated with synoptic wind systems. According to Wilson et al. (1984), a typical downburst event may have a diameter in the range of 1000 to 6000 m. Hjelmfelt (1988) reported a range for downburst sizes between 1500 and 3000 m based on 11 downburst events. Downbursts can lead to very high velocities near the ground. Previous field studies, such as the Federal Aviation Administration Lincoln Laboratory Operational Weather Studies (FLOWS; Fujita, 1985), showed that the maximum downburst wind velocities occur at the first 50 m above the ground, as indicated by Fujita and Wakimoto (1981), Wilson et al. (1984), and Hjelmfelt (1988). Savory et al. (2001) indicated that the maximum recorded wind speed during downburst events is equal to 67 m/s which is within the range of velocities corresponding to F2 tornado defined by Fujita and Pearson (1973). Fujita (1990) and Boss (2010) indicated that the probability of structural damage due to downbursts is higher than tornadoes because of their high frequency of occurrence. Various studies were conducted to characterize the downburst wind field. In general, these studies can be classified into field studies, experimental studies, and numerical simulations using the Computational Fluid Dynamics (CFD) tools. Although field measurements can provide accurate information regarding the velocity field, it is a challenging task to conduct site measurements for downbursts due to the unpredictability of the event occurrence in terms of time and space. This motivated many researchers to study downbursts either experimentally (Osegura and Bowles 1988, Lundgren et al. 1992, Alahyari and Longmire 1994, Yao and Lundgren 1996, Wood et al. 2001 and Chay and Letchford 2002) or numerically (Kim and Hangan 2007, Sengupta and Sarkar 2008, Mason et al. 2009, Mason et al. 2010a,b). Most of experimental and numerical studies on downbursts are based on the impinging jet (IJ) model suggested by Fujita (1985) and the cooling source (CS) model developed by Anderson et al. (1992). The IJ model, as inferred from its name, is based on the analogy between a strong downdraft that touches the ground during downburst event and a jet that impinges to a wall. The CS model is based on simulating the density perturbation happening in the cloud base by the cooling process. This continuously increases the density of the air inside the cloud base until it becomes heavier than the adjacent air and falls down forming the downburst. This can be

modeled numerically using a negative energy source located in the computational domain at the level of the cloud base, and it can be modeled experimentally by releasing a fluid parcel in a slightly denser ambient fluid.

There are some advantages of numercial simulations over physical experiments. Numerical simulations allow for simulating the actual size of downbursts, and thus, avoiding potential scaling effects, which is not the case for physical experiments. Also, numerical simulations allow for generating detailed information of the flow field in both time and space compared with physical experiments. The following discussion focuses on the numerical studies of downbursts.

Kim and Hangan (2007) employed the IJ model to simulate a laboratory-scale downburst with a diameter of 0.038 m and a jet velocity of 7.5 m/s. They solved the Unsteady Reynolds Averaged Navier Stokes (URANS) equations to obtain the time-dependent mean downburst velocities. These time-dependent mean velocities are typically referred to as the "running mean" velocities. Kim and Hangan (2007) extracted the maximum "running mean" velocity profile and compared it with the profiles obtained from previous experiments conducted by Donaldson (1971) and Didden Ho (1985) as illustrated in Figure 1.3.

Figure 1.3 Comparison between vertical velocity profile for downbursts. Reproduced from Kim and Hangan (2007)

0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 V/V_max Z /D j

Kim and Hangan (2007) Experimental (Ho 1985) Experimental (Donaldson 1971)

Sengupta and Sarkar (2008) used the IJ model to simulate downbursts. They employed K-epsilon, K-omega, Shear Stress Transport (SST) and Large Eddy Simulation (LES) turbulence models, and compared the resulting profiles with those obtained from experiments. The comparison showed a very good agreement between the profile obtained from LES and the profile obtained from the tests. Studies conducted by

Hadzˇiabdic´ (2005), Chay et al. (2006) and Gant (2009) also recommend using LES to 

simulate downburst wind fields.

Mason et al. (2009) employed the CS model to simulate downbursts in two-dimensional domain generated over various terrain roughness. Their study indicated that the increase in the ground roughness tends to decrease the maximum horizontal velocity and to increase the elevation where it happens. Later, Mason et al. (2010) studied the flow field of a translating downburst using a three-dimensional domain. For the two simulations mentioned above, the Shear Adaptive Simulation (SAS) method introduced by Menter and Egerov (2005) was utilized to account for turbulence while the neutral wall function was used to model terrain roughness. According to Teske and Lewellen (1977), the usage of neutral wall function in modeling terrain roughness is acceptable when the grid employed in discretizing the domain has a small height for the first layer above the ground. Mason et al. (2009, 2010) used a 1.0 m high first grid layer, which justifies the employment of the neutral wall function. However, the usage of that 1.0 m high layer shades doubt on their results for terrain aerodynamic roughness, z0, greater than 0.016 m, which include typical terrain exposures encountered by TLs (ESDU 2001). That is because of the following reasons; according to Richards and Hoxey (1993), Franke et al. (2004), Fluent Inc. (2005), Ansys Ltd., (2005), and Blocken et al. (2007), the height of the first grid layer, Δz, limits the ground roughness, ks, and the aerodynamic roughness, z0, as ks~30 z0, that can be modeled. Maximum roughness that can be modeled cannot exceed the mid height of the first grid layer, (ks or 30 z0) ≤ 0.5 Δz. This leads to a 0.016 m maximum allowable aerodynamic roughness z0 in the simulations conducted by Masons et al. (2009 and 2010).

Masons et al. (2009) compared the downburst profiles obtained using CS model with those using IJ model. The comparison revealed that the profiles obtained using CS model

are narrower and have a lower elevation for the maximum horizontal velocities than the profiles obtained from IJ method. This could be a result of employing a ramp function to enforce the flow in the simulations conducted using the CS model, compared with an instantaneous enforcing in the simulations conducted using the IJ model.

Vermeire et al. (2011a) simulated downbursts occurring over various terrains, with z0 equals to 0.001-0.1 m, using the CS model and employing LES to resolve for turbulence. Similar to Mason et al. (2009), they utilized the neutral wall function using a 1.0 m high first grid layer which shades doubt on their findings for terrain roughness z0 greater than 0.016 m. Later, Vermeire et al. (2011b) used the CS model to study the interaction between multiple downburst events and reported a 55% increase in the velocity magnitude compared to that of a single event.

All of the above simulations provide good insights on downburst wind field. However, none of these studies discussed the turbulent characteristics (such as turbulent intensities, length scales, spectra, and peak factors) of the flow near the ground. These characteristics are essential to quantify peak loads on structures including TLs and their responses as indicated by Chen and Letchford (2004a, b), Chay and Albermani (2005), Chay et al. (2006), Holmes et al. (2008) and Kwon and Kareem (2009). Holmes et al. (2008) analyzed the velocities of a downburst event recorded at the Wind Science and Engineering Research Center at Texas Tech University (Gast-Orwig and Schroeder 2005) and obtained the turbulent characteristics of the event. Unfortunately, these characteristics are for open terrain exposure only and limited to the locations where the velocities were measured. Detailed turbulent characteristics of downburst events happening over various terrain exposures typically encountered by TLs are still missing and this is one of the aspects covered in this thesis.

Obtaining the turbulent characteristics can be achieved using a high resolution LES with a careful modeling of terrain roughness. As mentioned earlier, terrain roughness was commonly modeled using wall functions which provide a constraint on the maximum roughness that can be modeled. This constraint has a significant effect especially when detailed flow characteristics close to the ground is needed. Methods such as terrain

following coordinates, immersed boundary methods (IBM) and canopy models do not have limitation on the roughness that can be modeled. However, they do not allow for modeling a terrain exposure with a prescribed aerodynamic roughness z0, which is needed for obtaining the turbulent characteristics of downbursts acting on different exposures. There is a need for a new method capable of modeling a prescribed aerodynamic roughness in LES without imposing a constraint on the roughness that can be modeled, and this is one of the topics covered in the thesis.