Point-of-Use Flow Control Design Project
Design Documentation Final Report
East Hill Surfers (EHS): Daniel Smith, Jacob Krall, Drew Lebowitz and Stephen Song
Abstract:
A flow control device is a component that holds vast potential for improving
point-of-use water purification systems in the developing world. The development of a
gravity-powered, cheap, yet effective device that achieved a low, yet constant, flow
would be extremely useful in developing more effective systems such as slow-sand filters
and chemical drip feeds. Although myriad possible devices exist for achieving this goal,
the floating-plug device is one of the simplest and most effective. The device utilizes a
float which is free to move vertically in a small intermediate tank. As the water level in
the small tank rises, a plug atop the rising float closes the water inlet, achieving a
constant head above the outlet. The steady-state device is able to produce a continuous
daily flow of water at about 20 liters, ideal for a single-family purification system.
This system has been designed, developed, and tested at Cornell University,
where it has been shown experimentally to consistently produce the desired flow rate.
This paper describes the process that led to the device’s selection, the theoretical and
practical considerations in its design, and the laboratory experimentation process that
verifies the device’s functionality. Finally, the report explores possibilities for
implementing this device in developing communities and possible improvements.
Design Project Goals and Context:
The current state of affairs in the global south shows that there is a pressing need
for an inexpensive, reliable, and low-maintenance source of clean water. Lack of
adequate municipal water, massive population growth, and the remoteness of rural
communities have made point-of-use treatment a viable longer-term solution. One
technology that is well-suited for point-of-use treatment is slow sand filtration. This
common filtering process offers a simple, low-cost solution for purifying water.
However, slow sand filters work best when the water flows at a constant rate. The
impetus for this project was the realization that the effectiveness of slow-sand filters
could be vastly increased by employing a flow control device as a filter regulator or
chemical feed dispenser. With this in mind, our team set out to design and build a selfpowered small-scale water flow device that was easy to maintain, inexpensive to
construct, and reliable.
A constant flow device for point-of-use treatment and chemical feeds must
achieve a very low flow rate. It was determined that for a household, the target daily
flow rate was 20 liters. A flow rate of this magnitude is so low that it can be measured in
countable drips. This presented unique challenges since it was quickly determined that
maintaining a low flow rate is more difficult than maintaining a high flow rate. One of
the main problems with maintaining a low flow rate device is clogging. Small orifices
and tubing are prone to clogging from turbid water. Because this device will be used as
part of a filtration process, it should be expected that there will be turbid water running
through it. Additionally, the device should not require any external power and needs to
withstand the corrosive properties of chemicals such as chlorine that would flow through
it. While a device that utilizes electricity may be very effective, it would not be suitable
for use in remote regions of the global south. These realizations proved to be important
in driving the design of a suitable device.
In order to motivate ideas, a web search was conducted to investigate prior
solutions to this problem. It was determined that there are many solutions, but everything
had its flaws. In addition to a web search, parts at a local hardware store were examined
to see what types of materials were readily available. With various ideas in mind, the
Team EHS brainstormed a list of designs, identified problems, and proposed solutions. A
design for a prototype device was eventually drawn up, constructed, and tested.
Device Decision Process:
Team EHS considered an array of preliminary design options before settling on
the float-controlled constant head mechanism presented in this report. In addition to the
basic performance requirements of being self-powered and providing constant flow, we
considered six additional key characteristics:
1. Ease of Use/Maintenance – Design must envision a vital device for a household
with limited access to replacement parts.
2. Size – Device must be as small as possible to minimize chance of breaking and
facilitate incorporation into existing devices
3. Universality – Device must be able to function with any mechanism of
manufactured or existing water purification system
4. Resistance to Clogging – Device should process water of any turbidity
5. Reliability – Device should be as simple as possible to limit failure mechanisms
6. Cost – Device production cost should be around $5.00 or less to be available to
families living on less than $1.00 / day.
The original brainstorming attempted to identify as diverse and innovative a solution as
possible. The process yielded a variety of ideas for flow control mechanisms. Most
devices employed a mechanism that would maintain constant head and in turn use this
constant head to force water out at our desired flow rate from a constant pressure. We
recognized several recurring categories of ideas, and grouped each idea into the
categories. Our six most viable ideas and their strengths and weaknesses are displayed in
Figure 1.
Along with our flow control devices, a control device consisting of only constant
outflow was considered. This device is intended to quantify the head variation that was
our original impetus for creating this design.
Selecting the Design
After evaluating our choices for our stated criteria, the floating plug-valve device
was selected as the optimal candidate. This solution is simple, cheap, and effective. One
notable aspect is that the device is entirely self-contained, making incorporation with an
existing flow system easy and unobtrusive. It can be placed either above or below a filter
or other purification mechanism, increasing its universality. Lastly, this device is easily
Floating
Inlet
Partial vacuum
inside tank
Ov
era
ll G
rad
e
na
lN
o te
s
A
B
C
C
B
Limited by small-pipe flow
B+
C
Feasible but illsuited to low flow
B
C
C
A
C
C
Overly complex and highly
prone to failure
Limited by smallpipe flow
A
B
B
B
A
B
High-wear gaskets prone to
B
frequent failure
Likely success but
requires minute
outlet
B
A
A
C
B
B
Requires precision valve;
Could be coupled with
Marriot Tube
B
C
B
A
A
A
Very easy to obtain parts;
Limited by size of unit
B+
A
A
B
B
A
Benefits of toilet valve yet
compact
A
Intermediate tank
Toilet Float
with auto-feedback Likely success
A
Valve
inlet
Intermediate tank Likely success with
Plugging
with auto-feedback
precise
A
Float Valve
inlet
components
Stand-Alone
None
Drip Valve Head is Variable
Ad
dit
io
Pre
d
Fu icted
nc
t
Su iona
cce l
ss
Self-adjusting inlet Eliminates majority
B
height
of head variation
Variable
Open-Channel
Torque
Flow
Pulley
SpringSelf-adjusting inlet
Supported
height
Bucket
Pinhole Air
Inlet
Ea
se
of
Us
e/M
ain
Siz
t.
e
Un
i
v
ers
Clo
alit
gg
y
ing
Re
sis
Re
tan
liab
ce
ility
Co
st
Device
Co
ns
tan
tH
ea
dD
ev
ice
constructed and modified, making it an ideal choice for optimization with our available
resources.
Generates
inconsistent flow
n/a
Control Device to verify FCD
necessity
Figure 1 - Device Decision Matrix
Theoretical Design Calculations, Size, and Constraints for the Float Valve:
The most important design consideration when using a float valve to control the
water level is the threshold pressure at which the valve will open. The objective of the
design is to keep the inlet into the FCD closed when the water is at the desired level and
open when it drops below that level, letting water in, and thus producing a constant water
level. Since the water level is constant, a fixed opening at the bottom of the FCD would
let water out at a constant rate according to Toricelli’s solution V=C√2gh. The maximum
pressure the valve will have to sustain while closed is the pressure created when the filter
or reservoir above it is full. If we call the maximum depth of the reservoir h, then the
pressure exerted on the valve is described by the hydrostatic equation: p = γh. Similarly,
the force is: F = A γh, where A is the area of the tube flowing into the FCD.
So now the objective is to plug the hole with a force greater than that produced by
the pressure head. There are a number of ways to accomplish this task of which the most
common is the toilet-style apparatus that uses a float attached to a lever arm to plug the
inlet when the water level raises the float (see Fig 2). We will look at this device briefly
as a means to understand why the float valve for the EHS FCD is more appropriate.
Inlet Flow
F = A γh
L1
Float
Pin
Buoyant Force = FB
L2
Figure 2 – Schematic for idealized lever arm system showing the pertinent forces at the inlet and
float and the lever arms about the pin associated with each one.
The force of the static pressure is greater than the force of the free jet that would come
out of the inlet, so we need to design for the static load. In this device the moment
created by the buoyant force about the pin must be greater than the moment created by
the pressure head and the weight of the float:
 M PIN  0  AhL1  FB L2  submergedL 2
However, for the small-scale of this design the use of a lever arm to amplify the
buoyant force of the float is unnecessary because of the miniscule area, and therefore
force, of the inlet flow. With the proper sizing, the float should be able to resist the
pressure head for a tank by its buoyancy alone with the inlet dropping vertically down
(see Fig. 3). The concept is identical to the float with a lever but the analysis and
construction is simpler. We assume that the maximum pressure the float will need to
resist is when the slow-sand filter or chemical feed reservoir is full at height H. We also
assume the float will effectively plug the inlet given the minimum size calculated.
Summing the buoyant force, pressure force, and weight of the float in the vertical
direction we have:
 F  0  FB  FP  W  submerged w   w HAi  tot f
For the minimum float size we set the total float volume Vtot equal to the submerged
volume Vsub to find an expression for the total float volume as a function of the head H,
inlet area Ai, and the specific gravity of the float:
 tot w   w HAi   tot f
  tot  HAi   tot S f
HAi
(1  S f )
To find the minimum float size needed we choose the maximum height to be H = 3m,
Ai= (π/4)Di2 where D = 2mm, and a reasonable estimate of Sf = 0.025 for low density
  tot 
styrofoam. Plugging in gives a theoretical minimum float size of Vtot = 9.67 cm3. This
approximation was taken into consideration during construction of the FCD.
The other value necessary to determine the size of the FCD is the height of water
needed to produce the desired flow rate of 20L/day. Based of an analysis of head losses
from commercially available small tube sizes, an inner diameter of around 2 mm was
found to be a good value for this application. To give the FCD a degree of versatility it
was designed for a maximum flow rate of twice the target flow rate: Qd = 40L/day. The
tubing needs to have length L = 9 cm to fit through the roller clamp and to adjust
comfortably. We also account for head loss at the entry of the tubing. Applying the
energy equation between the free surface in the FCD and the exit for water of
temperature 20°C we have:
2
( z1  z 2 )  V
 hL  K eV 2 2 g , K e  0.5,
2g
3V 2 32 LV

 1.25cm
4
D 2
Based on this analysis the head in the FCD need only be 1.25 cm. However, this height
will not overcome the surface tension effects that will resist the water entering the orifice
in the bottom of the FCD according to the equation:
1
h 
1
(67,000 2 )( r )
m
Therefore, if the depth of water in the FCD is 3 cm surface tension should not inhibit the
flow.
 h 
Analysis of the Prototype:
Based on the head and float size requirements, the FCD only needs to be tall
enough to accommodate a minimum of 3 cm of water and wide enough to fit a cylindrical
float with about a 3.5 cm diameter. However, the actual size of the FCD prototype was
determined by the availability and reliability of construction material as well as not be too
small to comfortable work with. The idea with the prototype was not to make it
absolutely as small as possible, but to demonstrate its functionality at a reasonably small
scale. Furthermore, it was constructed to a scale that was sure to be large enough to
produce the desire flow rate, overcome surface tension, and withstand a reasonable
amount of head so that the goal of showing its utility would not be compromised by
making the first attempt too small. The actual dimensions of the prototype are shown in
Figure 3.
2 cm
2.3 cm
1.5 cm
2 mm
0.5 cm
Rubber tip
4.4 cm
PVC
stem
11.0 cm
8.3 cm
Float
mass:
6 grams
6.5 cm
5.2 cm
Barb tubing
adapter
5.6 cm
9.1 cm
IV roller
clamp
Housing Dimensions:
ID = 7.85 cm
OD = 8.8 cm
IV tubing
(~10 drops/mL)
2 mm
Figure 3 - Cross-section of the cylindrical prototype FCD with dimensions. The housing material is
acrylic plastic and the float material is insulation foam. The IV roller valve is standard medical
supply.
Using the same analysis for maximum head the FCD can sustain as before, the prototype
can theoretically plug 53 m of head above it. Of course, much of the energy of the
buoyant force goes into compressing the rubber tip of the plug because the seal is not
formed perfectly upon contact. This number was made very large on purpose to ensure
that the device would work for sizable head. An experiment to determine the actual
maximum head the FCD can withstand would provide a functional relation between the
theoretical and actual head a float of a given size could sustain for a given aperture.
Realistically, the device does not need to completely plug the orifice at all to create
constant flow, but if it can make a water-tight seal for a given head it will certainly be
able to control flow from any depth less than or equal to that.
The flow rate is controlled by the IV roller clamp which constricts the area of the
effluent tubing from its maximum diameter down to a size that inhibits flow. The
maximum flow rate is determined by the energy equation in the reverse of the process
that was used to determine the minimum head required. The maximum head within the
FCD is 8.3 cm. The maximum flow rate will be Q=VmaxAvalve when Avalve=Amax, the
maximum area of the valve when the roller clamp is completely open and the diameter D
= 2 mm. Solving the energy equation for the velocity produced by an effective Δh = 8.3
+9.1 =17.4 cm (since the clamp is wide open the water in the vertical tube acts as more
head) we obtain a quadratic equation for V:
2
( z1  z 2 )  V
 hL  K eV 2 2 g , K e  0.5,
2g
3V 2 32 LV

 h  0
4
D 2
Solving yields a maximum exit velocity of V = 43.5 cm/s and a maximum flow rate of
5.9 times the target of 20L/day, which is about 118 L/day. The valve will be capable of
completely shutting off the water, so the minimum flow is Q=0. Clearly, the size of the
FCD is larger than it needs to be for the given design constraints, but it does provide a
wide range of flow rates which encompass the target flow rate.

Figure 4 - Assembled prototype.
Operator Instructions:
Team EHS made it a priority to ensure that the flow control device would be
simple to operate and maintain. To set it up, the device must be on an upright stand to
ensure that the float will move vertical with respect to the device. After mounting the
device, the source should be connected to the inlet. The inlet is a standard 1/4" tube
connector, so all that is required to link the device to the source is to push the tubing into
the connector until it is snug. After attaching the source, the IV valve on the bottom of
the device can be used to adjust the flow rate. Once the device fills up with water it will
begin to maintain constant flow. In the event that it is necessary to clean the device, there
are three screws which can be removed. Removal of these screws allows access to the
inside of the chamber so that it can be thoroughly cleaned. With the exception of the
bottom plate, every component of the device can be disassembled. This was done to
ensure easy replacement and cleaning of the components.
Prototype Testing and Results:
As a primary test of the performance of our device, we devised (after some trial
and error) an apparatus that would test the ability of our device to achieve constant flow.
A picture of the apparatus we used appears below (Fig 5). The narrow tube at the top
(diameter = 1 inch) was filled to a height such that the total head above the FCD valve
was 69 cm. A 7 kpa pressure sensor was connected above the FCD and measured the
decrease in head in the narrow tube at the top. The FCD allowed water to drip into a 10
cm diameter tube. Subsequently, the same apparatus was used to measure the ability of
the IV drip valve alone to achieve constant flow. The volume data was converted into
flow rate by using of Excel's slope function to average the slope over 50 seconds.
Figure 5 - Apparatus used to test performance FCD.
Flow Rate (mL/s)
Plots of the flow rate against time for the FCD and with the IV valve only appear below.
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
y = 5E-06x + 0.1645
R2 = 0.067
0
200
400
600
800
Time (seconds)
Figure 6 - Flow rate as a function time using the FCD.
1000
1200
1400
0.5
0.45
y = -0.0004x + 0.4544
R2 = 0.9545
Flow Rate (mL/s)
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
100
200
300
400
500
Time (seconds)
Figure 7 - Flow rate as a function of time using the IV drip valve only.
If constant flow existed, the plots above would be horizontal lines, and the
trendlines would have a slope of zero. Without the constant head float the flow rate (Fig.
7) decreases as a function of time, at a rate of 4*10-4 mL/s2. Using the FCD, the slope of
the graph is positive which corresponds to the flow rate increasing as head drops. This
makes no physical sense, so we attribute this small positive slope to data noise. In any
case, the downward slope using the IV only is two orders of magnitude higher than the
slope using the FCD, indicating that our device is much more effective in producing the
desired result of constant flow. It is important to note here that during the experiment our
FCD was running at a flow rate of approximately 0.17 mL/s which is equal to about 14.7
L/day. This is smaller than our target flow rate of 20 L/day. It is more difficult to
produce constant flow at small flow rates, so our results indicate strongly that our device
will be able to produce constant flow at the desired rate.
Another important characteristic to test is the resistance of our device to clogging.
We ran one test in which we showed that our device does not clog using dirty water
(about 180 NTU) on a time scale of six hours. Clearly, we must show more than this, and
we are currently running a longer study to determine if clogging is a problem on the scale
of several days using our device.
Recommended Modifications:
One issue with our FCD is that the prototype required significant labor to
produce. Thus, we are concerned about costs for our device. The cost per device would
be mitigated significantly if our device were mass-produced, but we would like to make
our device smaller so that it would require less material. We feel that the body of the
device could be made shorter and have a smaller diameter. This would also allow for a
smaller float diameter. Making these changes would reduce the costs of production of
our device.
In addition, the float could be made of a different material, such that it would not
become water logged over time. If clogging proves to be a problem for our device the
first approach to overcome the problem would be to widen the orifices. We will await
results from our clogging experiments before making such a decision.
Recommendations for Future Research:
While our flow control device has thus far proven to be successful at maintaining
constant flow, further testing must be conducted. Several repetitions of the experiment
cited above should be conducted to produce more data supporting the reliability of the
FCD to create constant flow. One of the important tests that has yet to be conducted
definitively is a clogging test using turbid water. Ideally, a test of this sort would be
conducted for several months to see if the accumulation of particles would lead to
eventual failure. However, in a preliminary, short term experiment on the order of six
hours in which water of approximately 180 NTU ran through the device no clogging or
deviations from constant flow were observed.
The device should also be tested to see what maximum head the float can
withstand. This will determine in what contexts the device can be successfully
implemented. An additional benefit to determining the maximum head will be the ability
to shrink the device down to its smallest possible size. A small device allows for more
universal application, ease of transport, and lower materials cost. Different materials
should also be investigated.
While the device may work well when it is operating with water, if it were to be
used as part of a chlorination process, there is the concern that some of the materials may
react with the chlorine. Therefore, non-corrosive materials should be utilized to ensure
universality. While there is still much research to be conducted, we are confident that the
design will prove to be successful in overcoming the challenges presented in this
problem.