Documente Academic
Documente Profesional
Documente Cultură
Understanding high lift generation and drag reduction for small bodies at low Reynolds
numbers(Re) is key in microrobotics design. To further our understanding of the kinematical
characteristics and aerodynamics around microgeometries at low Reynolds values, we conducted
a simulation through COMSOL to analyze low values aerodynamic characteristics of the NACA
0012 airfoil, a 2D cylindrical post, and a 3D sphere. Through many trials, the Reynolds number
was independently varied from 0.1 to 1000 to simulate laminar flow. For the NACA 0012 airfoil,
the angle of attack covered the range of 0, 3, 5, 9, and 12 degrees. Analysis of the data showed
that changes in Reynolds number effect the maximum drag coefficient and the drag force. In
reality, changes in Reynolds number affect all parameters to some extent. Unsteady flow could
not be observed for Re = 100 for an airfoil. We speculate that this is due to the airfoil being
designed to avoid unstable flows in the first place, so unless a high Reynolds number is applied,
we would not observe any unstable flows. H owever, we were able to observe unstable flow
around the 3D sphere at a Reynolds number of 100.
1
I. Introduction
When discussing the topic of flight, one usually considers large vehicles, such as
airplanes, that fly by generating high lift/drag differentials at high reynolds numbers (Re).
However, nature has created extremely small organisms, called microorganisms, that are capable
of flight at very low Re. Some of these organisms, such as fruit flies, have the ability to generate
the perfect amount of lift and drag despite low Re. While difficult, designing microrobots with
this ability is of great interest and use to many industries. For one, if engineers can design
microrobots that are capable of travel at low Re, such as the bloodstream, hospitals can inject
these robots into patients to deliver non-invasive treatment1.
To study the effects of laminar flow around very small geometries, we utilized
COMSOL, a multiphysics simulation tool. These geometries include the NACA 0012 airfoil
section, a 2D cylindrical post, and a 3D sphere. Specifically, we computed the drag force and
drag coefficient, along with the lift force and lift coefficient of these geometries, under low Re,
to study how different geometries induce different quantities of lift and drag. Furthermore, the
occurrence of unsteady behavior during laminar flow is of great interest and is further explored
in multiple time-dependent studies. By performing this study, we hope to further our
understanding of lift/drag generation of small geometries at low Re so that we may design
microdevices with flight characteristics.
II. Methods
We used COMSOL to simulate flow around different geometries at different Re. The
drag force and its coefficient on the airfoil and circular post were evaluated by computing the
force along the x-axis. The drag force and its coefficient on the sphere were evaluated by
computing the force along the y-axis. The axis was oriented so that the horizontal axis was the
y-axis and the vertical axis was the z-axis (figure 1).
We used a set of boundary conditions for all of our trials in the COMSOL simulations.
One of these conditions was the no slip boundary condition on the walls; this condition allowed
us to eliminate any edge effects that may have affected our results. Furthermore, the channels
were designed to be much larger than the bodies being tested meaning a much longer channel
and much wider width and height. The gage pressure at the outlet was set to zero to prevent any
adverse effects varying the flow results. The velocity profile was declared at the input, which
was always positive and normal to the face of the geometry to ensure steady and uniform fluid
flow throughout the channel. To achieve different Re, we calculated the corresponding velocity
for each desired value and varied the inlet velocity accordingly. The fluid, air, was declared to
have a constant density at 1.225 kg/m3 and viscosity of 1.48e10-3. With these boundary
conditions, we created simulations of time-dependent and laminar single phase fluid flow.
After obtaining a better understanding of flow behavior in two dimensions, we hoped to
further our knowledge to three dimensions by applying similar simulation conditions to a sphere.
In addition, we elected not to consider the effects at exactly 0 seconds because the fluid had not
made contact with the body. Furthemore, a large simulation time scale was used to reach steady
state. The times were varied for each simulation to ensure efficiency and reliable results.
1
Non-invasive compared to open surgery
2
Cylindrical Disk
A 2D simulation was done for a cylindrical disk, with a 1.0 cm diameter, in a channel
with a length of 200 cm and width of 100 cm. The time range used to achieve the steady flow
conditions was from 0 to 2400 seconds with a time step of 1 second. By having smaller time
steps we were able to more accurately identify the drag and lift forces imposed on the disk. Since
this is a 2D simulation, the characteristic length was used instead of the surface area. Using the
characteristic length implies a thickness of one unit length.
Sphere
A 3D time dependent simulation was done for a 1.0 cm diameter sphere to determine the
effect of the same Re used above, along with Re of 250 and 500, on the air flow around a sphere.
The channel was of length 140 cm, width 14 cm, and height 14 cm. Unlike the other geometries,
the inlet velocity field in the y-direction was set to be the desired velocity; this decision was
made after realizing that setting the normal inflow velocity generated unreasonable numbers in a
3D simulation. The simulations for the sphere were run from time 0 to 1000 seconds with a time
step of 10 seconds. To verify our simulation method, we compared our data with NASA’s
“coefficient of drag vs Re of a sphere” graph (figure 2).
III. Results
In order to further our understanding of fluid flow around complex geometries, we
compared the drag forces and coefficients of the sphere, circle, and the zero angle airfoil case for
multiple Re.
Sphere
First, we aimed to prove and verify the drag data collected by NASA. Overall, it can be
seen that for our graph and NASA’s graph are similar. Comparing Re of 0 to 5 to their respective
coefficients of drag, both graphs follow an inverse log function trend converging to zero. For Re
less than 1, the trend between the Stoke’s equation and collected data are similar. For Re values
greater than and equal to 10, both graphs followed a logarithmic trend. Because we ran an
idealized simulation, our results do not precisely match the values supplied by the Stoke’s
equation.
After validating the data, we performed a 3D time dependent simulation to prove and
further our understanding of the known trend of drag forces on a sphere. Additionally, the time
dependent solutions allowed us to determine how much time is needed for the fluid to reach
3
steady state. From figure 20a and 20b, it can be seen that it takes less time to reach steady state at
higher Re than the time it takes for lower quantities; for values in the range of 1 to 10, it takes
significantly longer to reach steady state than in values less than 1 or greater than 10. This
follows engineering intuition because at higher Re, the velocities are greater so it takes less time
for the velocity field to become uniform.
After running our simulations, we generated the streamlines and pressure profiles of the
various Re. Upon collecting and comparing images from figure 4a to 4k, it is clear that the flow
over the sphere does not become unsteady. Furthermore, for all of the values we tested, there
were no signs of turbulence; this made sense since turbulence typically begins to occur at Re of
approximately 2300. Instead, separation began to occur after hitting a Re of 1. The pressure
profiles proved to all be extremely similar in presentation but for higher Re, the greater velocities
resulted in higher pressure on the sphere's surface. Another note is that the streamlines along the
sphere at low Re, such as figure 4a, show much more curvature in the lines near the walls of the
channel than the streamlines at the same point in time for a higher Re as seen in figure 4k. This is
most likely due to the fluid above travelling slow enough to be affected by the curved fluid flow
below it.
Disk
For the disk, we saw that it follows a linear trend, a.k.a the Stoke’s Law, for small Re (0.2
to 1). This is shown in Figure 16. We tested this disk at slightly higher Re (50, 80 and 100) as
well and saw that the coefficient of drag no longer decreases in a linear fashion. It decreases
similarly to how the sphere graph does, albeit with a different function. We were also able to see
that there were some lift forces acting on the disk but since they were so small, we considered
them to be negligible. These lift forces are present only because fluid hits the bottom half of the
curvature. The bottom half is similar to that of an airfoil in that the area underneath it allows the
fluid to catch and push up the disk. However the same forces are counteracting on the top half as
drag due to the symmetric nature of the disk. Hence, we can neglect lift in our studies.
With increasing Re, we observe that the velocity streamline gathers closer to the circular
post and creates a curvature that is closer and closer to the shape of our post. At the critical Re of
4
100, it can be observed that an unstable flow occurs at the back of the circular post, shown in
Figure 18i.
In Figure 15, it is noted that the normal stress seems to be smaller than the shear stresses
at the circular post’s top and bottom vertices. This can be contributed to the fact that it is at a low
Re of 0.2. Therefore, the initial impact of the fluid at the face of the circular post is low but after
it moves around the surface, it experiences spatial acceleration and hence the shear stress at the
top and bottom vertices are higher.
5
allows the fluid to travel approximately horizontally at the end when it leaves the airfoil, hence
the top fluid would continue flowing horizontally and not collide with the bottom fluid.
6
Appendix
Figure 1: Axis oriented so that the drag force was computed as the force in the y-direction in 3D
7
Figure 3: Stress around a Circular Post at Re = 0.2
8
Figure 4c: Streamline of flow around sphere at Re = 0.95
10
Figure 4k: Streamline of flow around sphere at Re = 500
Figure 5: Cl /C
d vs. Airfoil angle of attack
11
Figure 7: Cd vs Re for NACA 0012 with 0 degree angle of attack
12
Figure 9: Streamlines around NACA 0012 at 12 deg AoA at Re=500
13
Figure 11a: Drag and Lift forces for varying angles of attack at Re=50
Figure 11b: Drag and Lift forces for varying angles of attack for Re=400
14
Figure 12: Cl /C
d vs Angle of Attack for Re=50 and Re=400
15
Figure 13b: 0 degrees: Airfoil Velocity Streamline at Re = 100
16
Figure 14b: 0 degrees: Airfoil Pressure Contour at Re = 500
17
Figure 15: Stress on circular post
18
Figure 17a: Circular Post: Pressure Contour at Re = 0.5
19
Figure 18a: Circular Post: Velocity Streamline at Re = 0.2
20
Figure 18c: Circular Post: Velocity Streamline at Re = 0.95
21
Figure 18e: Circular Post: Velocity Streamline at Re = 3
22
Figure 18g: Circular Post: Velocity Streamline at Re = 50
23
Figure 18i: Circular Post: Velocity Streamline at Re = 100
50 1.5 1.9
24
Figure 20a: Drag Force vs. Time at Reynolds number of 0.5
25
References
26