Los medios y la modificación de los hábitos perceptivos y de los gustos

The general aerodynamics of an isolated wheel in contact with a moving ground have been discussed in section 1.2.3. This section will look at the flow features and physics for the currently used geometry as reference for the combined wing - wheel study. Results from the experimental and computational approach will be combined to give a global overview. The longitudinal tyre grooves only seem to have a local influence and will not be discussed in more detail.

Contact patch region The contact patch region, aroundθ= 90◦_{, is mainly character-}

ized by the positive and negative pressure peaks respectively upstream and downstream of the area of contact. The unusualCP-value above unity for the positive pressure peak can

be explained from the fact that the Bernoulli law no longer holds in this region, because the energy is not conserved along a streamline. The energy that is added to the flow by the wheel rotation and ground movement increase the total pressure and this leads to the local high CP-value. A computational simulation in which only ground movement and

wheel rotation were modeled, but no flow velocity, showed that the peaks do still appear under these conditions; see table 3.1. This is in agreement with reasoning by Fackrell [61]. In contrast, for the stationary wheel in wind-on conditions the positive peak does not ap- pear and theCP-value does not even reach stagnation conditions due to the total pressure

losses as a result of the ground boundary layer; see table 3.1 again.

The viscous action of the two converging surfaces form the flow physics behind this phenomenon that is unique for the rotating wheel. The flow that is drawn into the corner by the moving surfaces is deflected outwards along the sides of the wheel in the form of viscous jets [34]. The rise in pressure has also been compared to a ‘pumping’ action [17]. The negative pressure peak results from the same, but oppositely acting, physical principles and was predicted by Fackrell [34] although it did not occur in his experimental results. The reason for this was suggested to be lifting of the moving ground belt due to the induced low pressure. In the current results the negative peak occurs for both the experiments and

Reference results

the simulations. The low pressure regions just upstream and to the outside of the contact patch (depicted as Z in figure 3.7) as a result of the local acceleration are a secondary characteristic of the contact patch region. These can be recognized in the computations (see figure 3.7) and in the experimental pressure distributions, especially for sensor location P2 (see figure 3.2).

Stagnation point The centreline stagnation point appears just below the most forward point of the wheel, near θ= 5◦_{. Fackrell’s results [62] indicate that the stagnation point}

moves closer to the ground due to wheel rotation. No stationary experimental pressure data has been obtained within this research, however the computational results show no effect of rotation on the centreline stagnation position. This is most likely a direct result of the grid resolution in this region. The CFD results predict the global stagnation position to be on the centreline, despite the wheel camber and resulting flow asymmetries.

Flow separation Flow separation from the top of the wheel takes place at around

θ= 275◦ _{in both the experiments and the simulations. As the flow separates from the top}

of the wheel in the simulations, the following base pressure in the wake reaches a fairly constant value of aroundCP =−0.4. In the experiments however the pressure coefficient

drops again after the local maximum and can reach lower values for a considerable section of the wheel before it reaches the more constant wake value. This broad second minimum can be distinguished for all of the tyre tread sensor locations and is completely missed by the SRANS simulations, independently of the used turbulence model. A similar second minimum can be distinguished in Fackrell’s experimental results for a rotating wheel [34, 62], but seems to occur only for the rotating case and not for the stationary5_.

A physical explanation for this feature can be found by studying the DES results for the isolated wheel. This unsteady simulation captures a second local minimum, even though the magnitude is underpredicted in a similar way as the suction value that occurs upstream of separation. It is therefore expected that this feature is unsteady in nature. However McManus’ URANS simulation of Fackrell’s geometry [17] does not capture the extra suction that is shown in Fackrell’s experimental pressure distribution either. From this it can be deduced that the mechanism behind this flow feature must be related to large

5_{The small local minimum for some of the stationary results of Fackrell is most likely the result of a}

local separation islet on the top of the wheel, as reported by Zdravkovich [25] for low aspect ratio cylinders. This same phenomenon can also be distinguished as the non-dried patch on the top of the wheel in the stationary oil flow results presented in figure C.6.

Reference results

scale unsteady eddy structures downstream of the separation position. Further analysis of the instantaneous PIV results of figure 3.6 (a), presented in figure 3.9, revealed that the separation process consists of irregular vortical structures that are being shed from the separating shear layer. The additional time-averaged suction that these structures gener- ate compared to the non-vortical recirculation depicted in the CFD results of figure 3.6 (a) is the physical reason behind the occurrence of this flow phenomenon. The statistical modeling instead of resolving of large scale unsteady vortical structures in RANS simula- tions will thus always lead to an underprediction of the lift for an isolated wheel due to the inability to capture this flow feature.

The CFD simulations show no obvious signs of separation over the sides of the wheel. The relevant pressure distributions, P3 and P5, present a large suction at θ = 180◦_,

indicating attached flow, which curves around the wheel into the wake. The experiments are more ambiguous about side separation, but there is not enough data to come to a decisive conclusion. Figure 3.6 (c) seems to show that the flow has separated and this is partly confirmed by the pressure distribution for P5 in figure 3.2, but on the other hand the sensor location on the opposite side, P3, seems to show only limited separation from θ = 230◦ _{to 190}◦_{. No PIV data is available for this side to check whether separation}

occurs here. Nevertheless it is clear that the complicated interaction of separation and reattachment, which has been described for stationary finite cylinder ends in section 1.2.1, does not occur for the rotating wheel. Finally, McManus et al. [17] also refer to a lower separation region around the front of the wheel close to the ground; in the area where a horseshoe vortex would be located for the stationary case. The current simulations do not show any flow reversal in this area, but a similar shaped ‘bow wave’ region can be recognized in theQ-iso-surface of figure 3.8.

Wake Several models have been proposed in literature to describe the trailing vortices and wake for a rotating wheel in ground contact. The majority of these [48, 60, 70] have been of a theoretical nature and are incomplete with respect to the occurring flow features. Recently McManus et al. [17] formulated a new description based on URANS simulations. The time-averaged arch shaped vortex at the top of the wheel, which he discovered, can be recognized in figure 3.8 as well. The flow in the upper near wake rotates around the arch shaped vortex core, instead of in a pair of counterrotating longitudinal vortices as proposed in the previous mentioned theoretical models. The lower extremes of the arch shaped vortex do however turn towards the freestream flow direction, but the vorticity

Reference results

along these legs dies out quickly, within a streamwise distance of 2/3D.

The lower wake is dominated by two longitudinal counterrotating vortices close to the ground. In figure 3.8 these vortices are represented as the two structures that continue the furthest downstream. The formation of these vortices results from a vortex in the centrelinex - z plane, which is bent in an arch shape in a horizontal plane, with the legs formed by the longitudinal vortices. The origin of this vortex in thex - z plane does not occur near the contact patch in the wheel - ground wedge, but more downstream around θ = 160◦ at 0.2D from the wheel surface. At this location the downstream downwash favours and strengthens this vortex, in contrast to the arch shaped vortex in the upper half of the wake, which is weakened by the local upwash at the downstream side. In figure 3.10 the longitudinal vortices are just being formed and are therefore not yet clearly visible, whereas the ends of the upper arch shaped vortex can be seen as the two local CP T-minima aroundz= 0.2m.