Anda di halaman 1dari 10

Numerical Modelling of Flow Around a

Turbine Blade
Advanced Modelling and Simulation (CFD)

Octavian Mihail
9763629
Introduction
The flow around a turbine-blade profile (T106) has been computed and analysed
using the Fluent software. This report studies the effects that mesh refinement,
discretization and inlet conditions have, on several important parameters: pressure
coefficient, Nusselt number, velocity, turbulent kinetic energy etc.
At first, both the coarse (8300 cells) and fine (16500 cells) meshes were
analysed and compared, at the same initial conditions (Re= 150,000, α = 37.7, and 3%
turbulence intensity), using the “Enhanced Wall Treatment”, to observe the possible
discretization effects. A third case was then studied, using the “Standard Wall
Functions” on a mesh with coarse near-wall grid spacing and the results were compared
with the ones obtained in the previous cases.
The flow around the blade is strongly dependent on the angle of the flow relative
to the chord of the blade. Therefore, an iterative analysis was finally carried out by
changing the inlet angle and its effects on the flow-field, pressure coefficient and
Nusselt number were discussed.

Results and Discussion


• Velocity vectors

Fig. 1: Coarse mesh velocity Fig. 2: Refined mesh velocity


vector vector

Fig. 3: Wall-function grid velocity


vector
It can be seen in the above figures that the maximum velocity of the flow is
reached on the edge of the upper blade. This is best represented when using the
“Standard Wall Function” (Figure 3), the maximum value being approximately 4.45 m/s.

• Pressure coefficient contour

Fig. 4: Coarse mesh pressure coefficient Fig. 5: Refined mesh pressure coefficient
contour contour

Fig. 6: Wall-function grid pressure


coefficient contour

When it comes to the pressure coefficient, the maximum values are reached on
the edge of the lower blade. However, these values vary slightly depending on the grid
density: 0.92 for the wall-function grid, 0.93 for the fine mesh and 1 for the coarse mesh.
These quantitative discrepancies between simulations can be better observed using x-y
plots, which are presented in the next section.
• X-Y Pressure Coefficient Graphs

Fig. 7: Pressure coefficient plot (Coarse mesh) Fig. 8: Pressure coefficient plot (Refined mesh)

Fig. 9: Pressure coefficient plot (Wall-function)


grid)

As previously stated, the pressure coefficient on the lower blade is close to unity
in all three cases, but it starts decreasing rapidly once the flow reaches the 0.7 m mark.
When it comes to the upper blade, however, the initial value of the Cp at the tip of the
blade is approximately -1.4 and it gradually decreases to a minimum value of -3.63, at
0.6 of the chord length. The pressure coefficient starts to raise after this point until the
solution converges.
Moreover, for the purpose of this analysis, experimental data, in form of pressure
coefficient values, was provided and plotted on top of the computed coefficient
distribution. When compared to this set of experimental data, the computed values of Cp
on the lower blade are accurate, as they follow the same path. On the other hand, the
values of the pressure coefficient on the upper blade are accurate only up until the 0.6
m point; after this point, the profiles follow the same path, but the experimental values
are slightly lower than the computed ones.
• X-Y Nusselt Number Graphs

Fig. 10: Nusselt Number plot (Coarse mesh) Fig. 11: Nusselt Number plot (Refined mesh)

Fig. 12: Nusselt Number plot (Wall-function grid)

In order to quantify the effects of heat transfer, the distribution of the Nusselt
number around the blade surface has been plotted for each of the three cases. By
comparing Figures 10 and 11 with Figure 12, the differences between the Enhanced
Wall Treatment and Standard Wall Functions can be observed as well.
In the first two cases, high Nusselt Number values can be observed on the
leading edge of the blades (~2300 for the lower blade, ~1400 for the upper blade), as a
result of the flow stagnation points located in this region. The values suddenly decrease
afterwards and remain almost constant as the flow moves along the blade. Finally, by
integrating the heat-transfer coefficient around the blade, the average Nusselt Number
was computed: 530.91 for the coarse mesh and 527.31 for the refined one.
When the Standard Wall Functions are used on the grid with coarse near wall
spacing, high values of the Nusselt Number around the leading edge of the blades are
detected again (~590 for the lower blade, ~660 for the upper blade). However, as
opposed to the previous cases, the values for the upper blade don’t immediately start
decreasing; they slightly fluctuate until the 0.6 m point and gradually decline from there.
On the other hand, the Nusselt Numbers on the lower blade constantly diminish as the
flow moves towards the trailing edge, but it starts increasing again at half of the chord
length. Again, the heat-transfer coefficient was integrated around the blade and the
average Nusselt Number was obtained: 499.42.

• X-Y y+ Graphs

Fig. 13: y+ distribution plot (Coarse mesh) Fig. 14: y+ distribution plot (Refined mesh)

Fig. 15: y+ distribution plot (Wall-function grid)

The distance between the near-wall grid nodes and the wall, y+, gives an
indication of whether or not the near-wall grid is properly fine.
When using the Enhanced Wall Treatment, the near-wall y+ values should be
close to unity for the mesh to fully resolve. When solving the coarse mesh, the
maximum value of y+ is 4.46 for the lower blade and 3.87 for the upper one, both
registered on the leading edges of the respective blades. These values immediately
drop, however, and come closer to unity as the flow moves towards the trailing edge.
The y+ distribution for the refined mesh is similar to the previous case, but the values
are even closer to unity: a maximum of 2.98 is recorded for the lower blade and 2.66 for
the upper one. Therefore, the y+ values are appropriate for the mesh to fully resolve in
both cases.
On the other hand, the Standard Wall Functions require y+ values greater than 20
to ensure that the near-wall cell is outside the viscous layer. Figure 15 shows that the
values recorded on the upper blade are always higher than 20, ranging from 35.33 to
57.4. Furthermore, the distance between the nodes and the wall on the leading edge of
the lower blade is 37.86. The values decrease afterwards but start rising again as the
flow moves past the 0.5 m point. Ultimately, the registered near-wall value of the y+
around the trailing edge is 20.63. Consequently, the distance between the near-wall grid
nodes and the wall is great enough to ensure that the near-wall cell is not inside the
viscous layer.

• Modified Inlet Conditions


One last study was made on the coarse mesh by changing the flow inlet angle
(25°, 30°, 35°, 40°, 45°, 50°). The effects on the Nusselt Number, the pressure coefficient
and flow-field were analysed by plotting the new values and by discussing the
outcomes.

Figure 16: Nusselt Number distribution (lower blade)


Fig. 17: Nusselt Number distribution (upper blade)

It can be seen in Figures 16 and 17 that the Nusselt Number values become
slightly lower as the inlet angle increases.

Fig. 18: Pressure coefficient distribution (lower blade)

Fig. 19: Pressure coefficient distribution (upper blade)


On the lower blade, the pressure coefficient’s value increases as the angle
increases (Figure 18). However, Figure 19 shows that even if the near-wall pressure
coefficient is higher on the upper blade as the angle decreases, it starts declining after a
certain point. Therefore, considering that the rate at which the coefficient declines is
higher for low angles, by the time the flow reaches the trailing edge, the values would
become lower for smaller angles.

Fig. 20: Velocity contour (25°) Fig. 21: Velocity contour (30°)

Fig. 22: Velocity contour (35°) Fig. 23: Velocity contour (40°)

Fig. 24: Velocity contour (45°)


Fig. 25: Velocity contour (50°)

Finally, Figures 20-25 show that as the inlet flow angle increases, the velocity
across the blade gradually decreases.
Conclusion
In conclusion, after studying the three cases, it can be stated that the computed
data is in close agreement to the experimental one. Moreover, the y + values are
appropriate for the mesh to be fully resolved. Finally, the Nusselt number distribution
showed the differences between the Enhanced Wall Treatment and the Standard Wall
Functions.

Anda mungkin juga menyukai