You are on page 1of 19

Technical

Note Concerning a Simple Flow

Visualization Device using Soap Films for 2D Fluid

Dynamics Experiments

Theron Pray

Collaborators: Evan McAyeal and DJ Turner

Abstract: The device outlined in this paper is a cheap, easy to construct apparatus
for creating consistently flowing soap films. Optical thin-film interference creates
readily visible patterns in the film that reveal underlying flow characteristics. The
film represents an experimental realization of 2D fluid dynamics and preliminary
qualitative results show agreement with theory. Future work should be done to
increase reliability and allow for quantitative analysis.


Introduction:

Soap films have been used for experiments in fluid dynamics for decades [1,

2, 3, 4, 5], and their spherical counterparts, bubbles, have captivated non-scientific

audiences with their colorful patterns for even longer. It is those same colors that

make soap films an excellent candidate for experiments in fluid dynamics. It has

been shown by Rutgers et. Al, that the optical interference patterns that are

naturally present in thin soap films provide an excellent method for visualizing the

flow in a variety of cases [1]. Soap films provide a laboratory analogue to the

theoretical two dimensional fluid. 2D fluids are useful tools in tackling the

behavior of fluids because often 3D fluid behavior is impossible to analytically solve

for. One can obtain results for a 2D fluid and extrapolate findings to the 3D realm,

for instance, approximating a flag flapping in the breeze with a string flapping in a

2D flow [6].

Methods and Results:

The basic challenge is to create a stable soap film with the proper rate of flow

and thickness such that optical interference patterns are visible. This is not a trivial

task; soap films are inherently unstable and, like most fluids, behave chaotically

such that the slight perturbations can lead to drastically different and unpredictable

flow characteristics. The ideal flow is uniform in composition, velocity, and free

from edge effects.


Apparatus:

Our design was based on the design of Rutgers et. Al. [1], and a schematic can

be seen in Fig. 1.

B
C


Figure 1 - Shows a schematic of our apparatus next to our actual apparatus. The top reservoir (A) feeds
soap solution to the nozzle (B) which runs down the guide threads (D) to form a film (E). The guide
threads are pulled apart by expansion lines (F) attached to a frame (C). The bottom reservoir catches the
soap at the end (G). The top bucket and frame are shown suspended from the ceiling on the right, and
the expanded (very thin) guide threads can be seen in the center.

The apparatus consists of two reservoirs vertically separated by about 2 meters.

The top reservoir (A) consisted of a bucket suspended from the ceiling. In the

bottom of this bucket we cut a hole and affixed (using a water-tight seal) a piece of

flexible plastic tubing about a foot long. At the end of this tube we installed the

nozzle (B) outfitted with a knob to control the rate of flow. Elastic guide threads (D)
were affixed inside the nozzle and stretched from the top reservoir to the bottom.

The nozzle was loosely fixed to the frame (C). The frame was constructed out of

wood and also suspended from the ceiling. The guide threads were tied to 4 inelastic

expansion lines (F). The expansion lines consisted of fishing line tied to simple wire

hooks. These lines passed through holes drilled into the wood frame and could be

hooked onto tacks pushed into the wood frame to expand the flow. The flow must

always start with these expansion lines unhooked and the two guide threads sitting

adjacent in the center of the apparatus. Once saturated with soap, the flow can be

expanded and a film will form. The guide threads were attached to a C-stand placed

in the lower bucket.

The basic design is pretty simple, however there are a few key features that

deserve a more in-depth discussion: the nozzle, the guide threads, and the frame.

The nozzle we chose (see Fig. 2) was a thin plastic nozzle 6 cm long that could be cut

to a desired output diameter. In between the nozzle and the plastic tube we placed

an on/off switch that had a knob to control the rate of flow. This switch fit snugly

inside the tube. The guide threads ran up through the output of the nozzle and were

attached to the plastic tube using a screw-tightened metal bracket. There was very

little weight on the nozzle itself since the elastic was not pulling down on the nozzle.

We slid the plastic nozzle up until it covered the output of the switch and loosely

attached it to the tube by wrapping copious amounts of Teflon plumbers tape

around it and the switch, which also waterproofed the assembly.



Figure 2 - Shows the nozzle, flow rate control, metal bracket holding the elastic lines, and the tube
leading from the top reservoir. Note the blue wire loop into which the nozzle would be inserted, in order
to hold the nozzle in place during a run.

The guide threads we ended up favoring were cut from elastic bands 3 cm

wide and 3m long (fully stretched). We cut these down their length with a razor

blade into a pair of 1mm wide threads that had rough edges left over from the cut.

Despite these non-ideal rough edges we did not view significant edge effects and we

were successful in obtaining stable films. Before deciding on elastic guide threads

we experimented with the use of flexible but non-elastic plastic insulated wires, as

recommended by the literature [1]. We found that it was near impossible to remove

all kinks from the wire and that these kinks led to undesirable 3D undulations in the

flow. Furthermore, if we attached them to the bottom reservoir while in their fully

extended state, when it came time to create a film we pushed them together and

ended up with a lot of slack near the bottom reservoir. This made it very difficult to

form films. Both of these problems were solved by using taut elastic lines. The
geometry of the guide threads greatly affected the flow obtained, and so obtaining

the proper geometry was our primary concern when designing the frame.

The frame was constructed by screwing thin boards together, and was about

1.2 m tall, 60 cm wide, and 5 cm thick. In designing the frame we took into account a

number of considerations. First, we wanted a way to easily experiment with

different guide thread geometries. The geometry of the threads greatly affects the

behavior of the flow. The initial expansion of the flow should be gradual enough so

as to avoid side wakes (see Fig. 3). The middle section, which is where any tests

should be carried out, should have parallel sides. The final section can have a

steeper slope than the expansion section because we are not interested in

maintaining uniformity at this point, but it should still give enough room to the

middle section so that when the film thickens as the soap decelerates towards the

end the middle section is unaffected. To this end, we chose to drill holes through the

vertically oriented boards at 40 cm, 60 cm, and at 110 cm from the top. We tied

fishing line to the elastic at the expansion points and then passed the fishing line

through these holes and then attached hooks (see Fig. 4). We could choose how fast

we wanted the flow to expand by selecting either the 40 cm or 60 cm holes, but

found the 60 cm to be better in all cases. We could select the expanded width by

moving the tacks either closer or further from these holes, and we found a good

width to be 16 cm.

Figure 3 - Shows a non-ideal flow. The speed of the flow is too fast and thus too thick to see interference.
3D standing waves can be seen on the right hand side that can be eliminated by slowing the flow. A side
wake is visible on the left, and you can tell popping is imminent due to the black hole in the center of
the whirlpool that corresponds to very thin film about to break.


Figure 4 - Shows the lower hole through which the expansion line passes. The line can be pulled taut and
the hook placed on the tack to expand the flow. Also note the holes drilled in the upper right of the
photo. The frame hung from ropes tied through these holes.

The horizontal boards of the frame had holes drilled into them on either end so that

they could be suspended from ropes (Fig. 4). When vertical flow was desired, one

could simply suspend the frame from the top two ropes and use C-clamps placed on

the bottom of the frame to reduce swinging. However, if so desired one could set the
entire apparatus at an angle by tying the bottom ropes to the ceiling as well (see Fig.

5). In theory, this allows for direct control over the speed of the film as it falls due to

gravity (flow slows as you approach horizontal). In practice, the speed of our flow

was more dependent on how far we opened the flow rate controller near the nozzle.

This method of flow control was crude and led to large variation in flow rate

between runs. Regardless of this source of error did obtain velocity measurements

for different angles, and they are presented in Fig. 6.


Figure 5 - Shows the apparatus set at an angle of 19, by suspending it from both upper and lower ropes.
The lower reservoir must be elevated to ensure the film does not contact the lower section of the frame,
however the upper reservoir need not be moved.

Figure 6 - A plot of the speed of our film as we change the angle from vertical (90) towards horizontal.
We could not maintain films past 19 as it would bow under gravity until it broke. Average speed was 2.5
0.5 m/s.

Techniques:
Due to the unstable nature of these films much of our time was spent

learning about the quirks of the apparatus. Largely by the process of trial and error

we learned what led to breaks and what led to longer soap film lifetimes. Our

techniques still leave much room to be improved upon, but we were successful in

obtaining acceptable qualitative results using the following procedures.

In order to generate a soap film, you must first ensure the flow rate control is

closed (perpendicular to nozzle) and fill the upper reservoir with soap solution. We

used a 5% Dawn:95% water solution. Next, the expansion lines should all be

unhooked and the guide threads should be no further than a few cm apart along

their entire length. The guide threads should not make contact with the wood frame
at the top or bottom, and should be centered. A simple wire loop wrapped around

the nozzle and loosely secured to the top of the frame was effective in centering the

nozzle (see Fig. 2). At this point you can turn on the flow and you should see the

soap begin to saturate the guide threads. Once the soap has reached the bottom the

threads should naturally stick together, or a small but thick film should form

between them. If it does not, you can coax it into forming by pushing the threads

together where they do not touch. Once the film reaches from the nozzle to the

termination point in the lower reservoir, you can pull the expansion threads and

hook them onto their respective tacks to expand the flow. You should expand the

top two first, and then the bottom two. Even as we became better at this process,

about two out of three times the film would break before we could use it. The best

films we obtained survived for 2 minutes, and the average lifetime was about 30

seconds. Because of this, data had to be collected quickly, but 30 seconds was

usually ample time for two people to create a film, insert an object of interest, and

take pictures.

In order to view the wake of an object, it must be inserted without breaking

the surface tension of the film. We had success using smooth plastic and metal

cylinders and other shapes. First coat the object in solution and either hold the

object by hand or mount the object to a C-stand. Then place the object in between

the guide threads while they are collapsed together in the center and the film

between them is thick. You can then expand the film in the usual manner, and the

film should flow around the object. Because of the inconsistency of the films the

process can be frustrating, but there are a few techniques to reduce failures.
The most important of these techniques involves the flow rate controller.

When turning on the flow it is important to not open it entirely. Although the device

was designed to be used in an entirely open or entirely closed state, we found that if

the knob was turned to entirely open, the resulting flow was always too fast and

turbulent. This turbulence would warp the film out of the plane of the guide threads

and standing waves would form. Aside from being unusable for our purposes, this

kind of flow inevitably broke upon expansion. If warping of this kind is visible in the

film, you should slowly close the flow rate control until it disappears. If, on the other

hand, the film consistently pops upon expansion, no warping is visible, and

interference contours are readily visible then your flow rate is probably too slow

and you should open the control further.

If the film continues to be unstable, look for the cause of the pop in the

interference patterns. The film is thinnest in areas where white and black contours

swirl together. These kinds of contours are usually present just before a pop. They

usually arise when the fastest flow does not reach across the entire channel. Often

you will see a thick main channel in the center where contours are close together

(sometimes so close that they cannot be resolved and look white), and highly visible

slowly moving whirlpools form on the sides (see Fig. 7a). These whirlpools do not

get replenished with fresh liquid, so as the water evaporates they get thinner until

they break the surface tension of the film. The best way to fix this problem is to

decrease the width of the channel, or to turn down the rate of flow until the side

wakes are minimal.


You will probably find that if you generate a film that has visible contours in

the center, the flow is inconsistent and the film is very short lived (5-10 seconds), as

in Fig. 7b. However, if you generate a film that is long lived and has consistent flow

in the center of the channel, it will be too thick to see contours, as in Fig. 7a. The best

way to obtain a film that has visible contours, and is consistently flowing with an

object in it for long enough to capture pictures is by creating one of the reliable but

unusable thick flows an then turning off the flow rate control entirely. After turning

the flow off, the film will become much thinner and slower flowing, as can be seen in

Fig. 7c. Contours will show up within about a second and for about five seconds after

cutoff you will have a consistent and visible flow around your object. Take pictures

of the wake during this period. After that period, the flow will become inconsistent

and randomly directed until it pops shortly thereafter.


Figure 7 (a, b, c) - Shows a flow that is too fast (a), too slow (b), and an ideal flow obtained using the
"on/off" method. Note the lack of contours in the central flow and side wakes in (a), the inconsistent flow
but clearly visible contours of (b), and the consistent and visible contours of (c).

Viewing Interference Patterns:

Once you have obtained a useable film, you must be able to see the

interference patterns. If you are not using a camera and only want to view the
contours with your eyes, the overhead fluorescent lights are sufficient. The contours

are most visible by standing about a foot away from the film and off to one side, and

then looking down through it. This is a good method to employ during practice runs

or when multiple people want to view it, as it can be viewed from both sides.

However, by using a bright, directed light source, a white sheet, and a black

sheet, the contours can be brought into striking clarity from one angle, ideal for

recording the flow on a camera. This setup can be seen in Fig. 8. The lights and black

backdrop are placed on one side of the film, and the camera and white sheet should

be placed on the other. The lights should be illuminating the film and directed

towards the white screen. Bright, full spectrum lights (like the tungsten lights used

in cinematography) were ideal. The camera should be looking through the film such

that the black backdrop is directly behind the film from the cameras viewpoint. The

lights themselves should not be in the cameras sight, as this will overexpose the

shot. Finally, the film should be reflecting the white sheet back into the camera. The

black backdrop ensures that the light from the interference does not mix with light

from objects in the background. By viewing the contours against the reflection of the

white sheet they become much brighter. There will probably be shadows cast by the

frame on the white sheet. Try to angle the lights such that these shadows do not

enter the reflection in the film or else the contours will be less visible in the camera

as they pass over the reflection of the shadow.



Figure 8 Shows the ideal setup as seen from above. The camera and white screen are on one side of the
film and the light and black screen on the other. The light shines onto the white screen while the camera
views the film with the black screen in the background. The reflection of the white screen on the film
should be seen in the camera.

The camera we used was a Casio Exilim EX-F1 high-speed camera. The

camera was mounted on a tripod so that the quick motion of the flow could be

captured without blur. We had originally intended to shoot high-speed video (300

FPS) to capture the flows as they evolved in time, but found that we did not have

sufficient light at such high frame rate to obtain useful video. Instead we used the

action photography mode that takes a number of HD stills at a user set number of

frames per second. By holding down the take picture button this generated 60 stills

from which we could choose the best shot. We set the white balance to Tungsten

and turned up the exposure to maximum and found that this lead to clear, bright

contours.


Results:

While our setup was not consistent enough to yield quantitative results, by

employing the techniques discussed above we did obtain qualitative results of the

flow past various objects, shown in Figures 9 and 10.


Figure 9 (a, b) - Shows two wake regimes for the same cylinder. Flow is fast in (a), and produces a
smoothly flowing wake characteristic of the high Reynolds number regime. In (b) flow is slower and the
distinctive Krmn vortex street is visible, corresponding to a lower Reynolds number.


Figure 10 (a, b) - Shows the wakes of an equilateral triangle with point down (a), a square (b). Note the
two wakes behind either face of the triangle in (a). Vortex shedding occurs at the bottom and in between
these wakes. In (b) the wake was more chaotic and vortices formed directly behind the bottom surface.
Future work:

Despite the qualitative success, our techniques and apparatus could be

greatly improved upon to yield more useful results. These adaptations and

alterations to our apparatus would open avenues to many interesting future studies.

Ideally, the apparatus should produce a thinner film with more readily visible

contours, a much longer lifetime (others have reported films lasting for hours [1, 2,

3]), and in which the flow is consistent in velocity over the test section, without edge

effects.

Other groups have reported flow speeds of between 0.5 m/s and 1 m/s [1, 2,

3]. Our average speed of 2.5 0.5 m/s was significantly faster than theirs and was

most likely due to a thicker film. A thicker film will reach a higher terminal velocity

due to less drag from the surrounding air. This lack of a truly thin film also

manifested itself in the fact that for our most long lived flows contours were rarely

visible except in the side wakes. The cause of this problem cannot be linked to any

one part, but rather must be solved through a number of changes.

First, a more accurate flow rate control should be used before any other

changes are made. The ability to precisely choose your flow rate will lead to a more

consistent flow overall and decrease the amount of time-consuming trial and error

greatly. Ideally it should be easy to change mid-flow, should have some marking

system such that you can choose the same flow rate every time, and should not

produce any excess turbulence (although the correct nozzle can reduce this).

Once this has been obtained the channel width should probably be

decreased. Others have reported great success using channel widths of only 5 cm [1,
2]. A thinner channel combined with finer flow control will allow for lower flow

rates and corresponding thinner films to survive while constantly flowing. Once the

channel has been made smaller, edge effects will most likely be present due to the

rough edges of the elastic bands we used. While other groups [1, 2, 3] preferred to

use rigid wire or fishing line, elastic guide threads still seem to be better for our

purposes. However, the elastic bands used in our apparatus could be replaced by a

smoother, thinner string, such as nylon. By combining these changes the basic

apparatus can be improved on to create more consistent flows and pave the way for

quantitative studies.

Other groups have used a different soap composition to great effect. Kellay,

Wu, and Goldberg [3] report that by using a 4% Dawn, 5% glycerol, 91% water

composition the film lasted much longer. It is well known that glycerol increases the

lifetime of soap films. However, in a more recent paper Rutgers, Wu, and Daniel [1]

reported that the addition of glycerol increased the viscosity of the fluid such that it

did not expand to the full width of the channel and side wakes formed. We did not

experiment with different soap concentrations, but this could prove fruitful for

future study as the short lifetime of the films ended up being one of the most time

consuming problems with the apparatus.

Finally, our apparatus was constructed such that panes of plastic could be

clamped to either side of the frame at the four corners of the frame using large C-

clamps. Other groups used vacuum chambers to isolate the film from the chaotic and

inconsistent flow of surrounding air [1], but the panes of plastic could provide

isolation from fast moving air currents without the need for a complicated vacuum
system [1, 2]. The thinner flow will most likely be more easily affected by air

currents, so attaching the panes of plastic (while making it more difficult to access

the film itself) may be necessary to achieve a consistent flow.

Conclusion:

The apparatus detailed here was cheap, relatively easy to make, and has the

ability to generate striking visualizations of 2D flows. Although repeatedly creating

films for just one data point can be frustrating, the results obtained are beautiful and

useful. There is still much room for improvement, which would open up the exciting

possibility of further quantitative studies. This technique represents the ideal

undergraduate fluid dynamics experiment, but has proven its potential for use at

any level.

Acknowledgements:

I would like to acknowledge the help of Mark Zach for providing materials

and assisting in the construction of the apparatus. We would also like to thank Tom

Baraniak for lending us the high speed camera and other materials. Melissa Eblen-

Zayas and Marty Baylor assisted us in various ways throughout the process. Finally,

we would like to acknowledge the Carleton College Physics Department for

providing funding and lab space.











Sources:

1. Rutgers, M. A., X. L. Wu, and W. B. Daniel. "Conducting Fluid Dynamics
Experiments with Vertically Falling Soap Films." Review of Scientific Instruments
72.7 (2001): 3025. Web.

2. Vorobieff, Peter, and Robert E. Ecke. "Fluid Instabilities and Wakes in a Soap-
film Tunnel." American Journal of Physics 67.5 (1999): 394-99. Web.

3. Kellay, H., X-l. Wu, and W. I. Goldburg. "Experiments with Turbulent Soap
Films." Physical Review Letters 74.20 (1995): 3975-978. Web.

4. Bruneau, C. H., O. Greffier, and H. Kellay. "Numerical Study of Grid Turbulence


in Two Dimensions and Comparison with Experiments on Turbulent Soap Films."
Physical Review E 60.2 (1999): R1162-1165. Web.

5. Couder, Y., J. M. Chomaz, and M. Rabaud. "On the Hydrodyanmics of Soap


Films." Physica D.37 (1989): 384-405. Web.

6. Zhang, Jun, Stephen Childress, Albert Libchaber, and Michael Shelley. "Flexible
Filaments in a Flowing Soap Film as a Model for One-dimensional Flags in a
Two-dimensional Wind." Nature 408 (200): 835-39. Web.

You might also like