11 Open Channel
11 Open Channel
11 Open Channel
Page 14.1013.1
Nomenclature
The Bernoulli flow experiment has been used at Rochester Institute of Technology for many
years. It is a staple experiment performed by senior students in the Thermofluids Laboratory
course. In the past, students have expressed frustration in obtaining meaningful results from the
experiment. Some past students have called it “an exercise in futility,” and others have failed to
use the raw data in meaningful ways. Other students have even suggested changing the design of
the channel to a circular cross section. This investigation was performed to elucidate an
interesting phenomenon observed when analyzing the raw data. Moreover, this investigation will
help future students better understand the experiment, and it will be used in an advanced fluid
mechanics course for engineering technology students.
The experimental procedure is typically performed by first obtaining a steady fluid level in the
inlet tank, and then various flow rates can be obtained by adjusting an outlet pipe. After
obtaining a steady flow rate, students typically record the water height in 11 manometer taps
equally spaced along the length of the channel. Next, the fluid is allowed to collect in a basin
whereby the volumetric flow rate is recorded using a clear sight glass with a scale and a
stopwatch. This is typically done several times, and the average volumetric flow rate is then
calculated. When analyzing the raw data, students typically plot the change in pressure head vs.
distance for the various turbulent flow rates. Students also plot the change in kinetic energy vs.
distance.
This investigation beganafter evaluating the total energy, ideal head and calculated head losses.
Calculated head losses decrease after the channel throat and in turbulent flow become negative.
This is not expected, because the fluid cannot convert kinetic energy into pressure energy
without appreciable positive head losses. Several experiments were then performed along with
attempting numerical simulations to investigate and elucidate this phenomenon. Next, the
experiments and their results will be discussed. The process of determining the root of the
strange results and the current status of the project will also be outlined.
Initial Experiment
The first step in this investigation was to run the Bernoulli flow experiment “as-is” and attempt
to replicate the results of past students. Figure 1 shows the P6231 Bernoulli apparatus along with
the test bench. The P6100 hydraulics bench consists of: P6103 constant head inlet tank, P6104
variable head outlet tank, variable speed centrifugal pump, and volumetric measurement tank.
Initially, a total of eight different flow rates were run, and each consisted of five volumetric flow
rate trials. This was done in an effort to eliminate any measurement errors that could occur while
reading the volume sight glass. The manometer heights were recorded for each flow rate using
the 1/16th inch scale on the Bernoulli apparatus. Figure 2 shows a section view of the apparatus
with the numbered manometer taps. The fluid enters on the left side of the channel from the
Page 14.1013.3
P6103 inlet tank, then travels from taps 1 to 11 and finally exits into the P6104 outlet tank.
P6104
1 2 3 4 5 6 7 8 9 10 11
P6100
P6231
P6103
Fig. 1 P6231 Bernoulli apparatus and P6100 Fig. 2 Section view of P6231 Bernoulli
hydraulics bench apparatus (1)
After collecting the raw data, a series of initial calculations were performed, and several data
products were produced. Initially, the results in the turbulent flow regime are as expected. Figure
3 shows the varying head along the length of the channel. As the fluid enters and moves through
the converging section, its kinetic energy increases as a function of channel height. This increase
in kinetic energy must be compensated for by a decrease in pressure energy as defined by the
simplified Bernoulli’s equation, equation 1. Figure 3 shows the change in pressure head along
the channel for various flow rates.
P V2 Pi
Equation 1: + =Κ where = PH i = H i i = 1,2,...11
γ 2⋅ g γ
Page 14.1013.4
After graphing the change in pressure energy, the total head and ideal head were found at each
manometer and plotted. The cross-sectional areas at each manometer tap were calculated using
the channel heights shown in figure 4, a channel thickness of 6.35mm, and equation 2. Figure 5
shows the total head and its overall decreasing trend over the channel. Total head is the sum of
the pressure head and velocity head at each tap (see equations 3 and 4), thus it would be expected
that an overall decrease in total head would occur. Ideal pressure head is shown in figure 6 and is
equal to the algebraic difference of the total starting head and the velocity head at each
manometer as shown in equation 5. The ideal head represents a total recovery of pressure energy;
thus it is symmetric about the throat of the channel.
Equation 2: Ai = hi ⋅ t
Fig. 4 Channel Heights at manometer taps
Q Vi 2 Vi 2
Equation 3: Vi = Equation 4: TH = PH + Equation 5: I H i = TH1 −
Ai i i 2 ⋅ g 2⋅ g
Page 14.1013.5
The experimental head losses were found using equations 6 and 7, cumulative and non-
cumulative, as shown above. Figure 7 shows the cumulative head losses in the channel, and the
trend appears to be normal from taps 1 through 6. The head losses increase as the fluid increases
in velocity and decrease as the fluids speed is reduced in the diverging half of the channel.
However, because figure 7 is cumulative, a decrease in head losses should be represented by a
smaller positive slope. This is not what figure 7 shows after manometer 6, and it can be
postulated that negative head losses have been calculated. A better understanding of the negative
head losses can be seen in figure 8. The head losses in each flow rate sharply decrease and
become negative after the throat.
Fig. 8 Non-cumulative head losses
Oddly, the losses then return to zero and do not change from taps 8 and 9. Furthermore, these
graphs indicate normal behavior in the converging half of the channel and abnormal fluid
behavior in the throat and diverging half. Negative head losses suggest that energy is being
introduced into the system or that there is another phenomenon happening that is not yet
Page 14.1013.6
understood.
Controlling the Experimental Environment
After cleaning both the apparatus and test bench, the Bernoulli apparatus was checked to ensure
the inlet and outlet were leveled. Figure 10 shows an image of the apparatus before it was
adjusted. This small incline could have added potential energy to the fluid, and it was not
detectable to the naked eye. Two-millimeter-thick shims were added to the diverging end of the
channel to achieve a level reading.
Debris blocking
tap inlet hole
Fig. 9 Debris in Bernoulli apparatus
Page 14.1013.7
In an attempt to achieve laminar flow rates with the Cussons test bench, the inlet tank
was modified by removing the upper 25 mm portion of the overflow pipe. Additionally, a ball
valve was added to the adjustable discharge pipe. This allowed for a much lower range of flow
rates to be achieved, and the manometer readings could be easily read at the 25 mm inlet tank
level. After running a number of initial trials, three runs were performed with the same procedure
as the previous experiments, the subsequent raw data was collected, calculations were
performed, and the results were analyzed. Again, pressure head was plotted against manometer
tap and can be seen in figure 11. The overall loss in energy seems to be much less than in the
turbulent cases, and this is expected for lower flow rates. It is important to note that the
manometer scale is not accurate for flow rates lower than 0.9 LPM and is the limiting factor for
the range of flow rates. The 0.92 LPM flow rate almost appears to be a horizontal line when
plotted on the same scale as the turbulent runs. It should be noted that this run, 0.92 LPM, was
the only one having Reynolds numbers at or below 2300 in the entire channel, including the
throat (2375).
Page 14.1013.8
Fig. 12 Cumulative head losses (Laminar) Fig. 13 Non-cumulative head losses (Laminar)
As shown in figure 12, the flow rates of 1.74 and 1.35 LPM follow the same trend as the
turbulent flow rates; however, the 0.92 LPM flow rate shows a different trend after the throat.
Head losses in the 0.92 LMP run seem to not become negative after the throat, but this could be
caused by the inaccuracy of the measurement scale. Nonetheless, the negative head losses at tap
7 do seem to become closer to positive as the flow rate decreases, as shown in figure 13,
particularly at the lowest flow rate. Overall the laminar trials do not yield results that can be used
to explain the negative head loss phenomenon, but the trend indicates that the flow regime might
have an impact.
Flow Visualizations
To better understand the behavior of the flow in the channel, several flow visualizations were
performed in various flow rates. Black food coloring was diluted with an equal proportion of
water and then injected into the inlet of the converging portion of the channel. A reservoir of ink
was connected to the injector pipe with a clear tube, and initial calculations were performed to
determine the optimal height for the reservoir. To clearly observe the nature of the flow, it was
critical to match the ink to the fluid’s flow rate.
Recirculation
Zone
Fig. 14 Laminar flow visualization
Fig. 15 Recirculation zone in diverging channel
Page 14.1013.9
Small variations in the ink height resulted in large changes in the ink flow rate, so the optimal
flow rate was found through trial and error. In the assumed laminar flow rates, an undisturbed
layer of ink can be seen over the entire channel length (figure 14). This observation was used to
confirm that the calculated Reynolds numbers were correct along with the assumed flow
regimes. One major discovery was made in the diverging half of the channel: during all flow
rates, a clear recirculation zone can be observed (figure 15). This suggests that the fluid velocity
is not slowing at the expected rate. The area above the recirculation zone is occupied by a
distinct stream of fluid that seems to have the same flow area as in the throat of the nozzle. This
discovery has some major implications in the way that the data is analyzed. To calculate the fluid
velocity head, its kinetic energy is found as a function of its velocity. From the continuity
equation, the fluid velocity is a function of the flow rate and the channel flow area. Thus, if the
fluid is not expanding in the diverging half of the channel, its flow area remains close to that of
the throat. This correction will serve to alter the predicted head losses and provide a more
accurate result.
To better understand the head losses in the system, the rates of energy conversion were plotted.
Figures 16 and 17 show the differing rates of energy conversion in the converging and diverging
sections of the channel. In the figures, pressure head was found using equation 1, and velocity
head was found using equation 8. As defined by Bernoulli’s equation, the rate of energy transfer
from kinetic to pressure energy should be equal in both the diverging and converging sections of
the channel. In the converging section, the magnitude of the slope is greater than in the diverging
section. This suggests that the converging section is converting kinetic energy to pressure energy
with more losses. The diverging section has a slope that suggests fewer losses occur.
Vi2
Equation 8: VH i =
2⋅ g
Page 14.1013.10
Fig. 16 Energy transfer taps 1-5 Fig. 17 Energy transfer taps 7-11
0.25 4 ⋅ V i ⋅ Ri ⋅ ρ
Equation 9: fi = Equation 10: N Ri =
⎡ ⎛
2
η
1 5.47 ⎞ ⎤
⎢log ⎜ + 0 .9 ⎟ ⎥
⎢⎣ ⎜⎝ 3.7 ⋅ ( Ri / ε ) N Ri ⎟⎠ ⎥⎦
Equation 11: Ri =
[Δhi + h6 ] ⋅ t Equation 12: Δhi = tan(θ ) ⋅ ΔLi −6
2 ⋅ (hi + t )
⎡⎛ ΔL ⎞ ⎛ Vi +1 2 ⎞⎤
= HLTi + f i +1 ⎢⎜⎜ ⎟⎟ ⋅ ⎜ ⎟
⎜ 2 ⋅ g ⎟⎥
Equation 13: HLTCi +1
R ⋅
⎣⎢⎝ i +1 ⎠ ⎝
4 ⎠⎦⎥
Figure 18 is a comparison of the experimental and theoretical head losses for the maximum flow
rate. The head losses were found using equations 9 through 13 and are shown above. The change
in length between manometer taps ( L) was measured to be 25 mm for all manometers, and a
wall roughness ( ) of 3.0 x 10-7 m was used for plastic. Likewise, 4.29 degrees was used for both
the converging and diverging angle of channel incline ( ). From taps 1 through 4, the losses
appear to match almost flawlessly. In contrast, the head losses from taps 5 through 11 do not
match as closely.
Fig. 18 Comparison of head losses
Next, using observations made during the flow visualization, two adjustments were made to the
theoretical losses calculation. First, as previously mentioned, the fluid in the diverging section of
the channel only experiences one-half of the wall friction. This is because a recirculation zone
exists and the fluid is moving in a confined stream. Secondly, the flow area in the diverging
section of the channel can be adjusted to not include the recirculation zone. The corrected
formula for cumulative head losses (equation 14) can be seen below and was only used for taps 7
through11. By estimation, an adjusted theta of 2.0 degrees is used to calculate the actual flow
area in the diverging portion of the channel (see equation 12). After making these adjustments,
Page 14.1013.11
the theoretical head losses after the throat correlate well with experimental values as shown in
figure 19. It can now be definitively concluded that the recirculation zone serves to reduce wall
friction and flow area in the diverging section of the channel. Furthermore, the throat of the
channel can now be the focus for the rest of the investigation.
1 ⎡⎛ ΔL ⎞ ⎛ Vi +1 2 ⎞⎤
= HLTi + ⋅ f i +1 ⋅ ⎢⎜⎜ ⎟⎟ ⋅ ⎜ ⎟
⎜ 2 ⋅ g ⎟⎥
Equation 14: HLTCi +1
2 R ⋅
⎢⎣⎝ i +1 ⎠ ⎝
4 ⎠⎥⎦
Fig. 19 Adjusted cumulative head losses
Numerical Simulations
Many 2D numerical simulations were attempted using FLUENT software in the laminar flow
regime. Different boundary conditions, mesh refinements, and discretization methods were used.
Currently more numerical simulations need to be done to accurately model what is happening in
the channel. In some cases, the recirculation zone was observed, but the numerical solution may
have contained errors indicated by the residuals plots. Moreover, a better understanding of the
fluids behavior at the throat of the channel will come with more complex numerical simulations.
Furthermore, simple 2D models may not explain the fluid behavior at the throat of the channel.
Conclusion
Several important discoveries were made during this investigation: Negative head losses in
turbulent flow, a recirculation zone in the diverging channel section, and the throat of the
channel were determined to be the cause of the strange results. Some difficulties were also
encountered during the investigation process. Achieving laminar flow was difficult due to the
accuracy of the manometer measuring scale. Nevertheless, a flow rate of 0.92 LPM was
concluded to be laminar from calculated Reynolds numbers and showed positive head losses in
the diverging half of the channel. However, this data may have been affected by the inaccuracy
Page 14.1013.12
of the measurement scale; thus it cannot be used to make any definite explanations about the
source of strange head loss results. On the contrary, after using various correction factors for
reduced wall friction and a decreased flow area, theoretical head losses correlated well with
experimental losses with the exception of the throat. Overall, the fluid’s behavior is highly
complex in the throat and diverging section of the channel, and further experimental
investigation, along with numerical simulations, are required to fully understand it. To
summarize, numerical simulations can be an equal partner with theory and experimental data, but
replicating the fluid flow correctly was found to be challenging. Lastly, with more numerical
simulations and better-developed boundary conditions, the numerical results could help to
explain the nature of the fluid flow at the throat.
References
Page 14.1013.13