Optimization of Venturi Performance
Brandon Baetz, Philip Guerieri, Travis Krell, Theodoros Roustopoulos
Phase 4
Team Precision AirConvey
0
Table of Contents
Prototype 2
Test Plan 2
Plan Used 2
Problems and Future Improvements 3
Analysis and Interpretation 4
Methods 4
Results/Conclusion 5
Theory Review 7
Design Iteration 10
Empirical Testing for Venturi Selection Calculator 11
Market Introduction 12
Patent Consideration 12
Appendix 13
Large Prototype Photo 13
Stress Analysis 13
Error Analysis 14
References 15
1
Prototype
For a full description of the modular prototype design, see Phase 3. For the sake of
understanding some of the project problems discussed in this report, recall that the prototype
needed t be fabricated before the design was completely optimized and finalized. The design
team encountered a limit on the theoretical engineering tools available (like computational fluid
dynamics) and required an empirical method to gather more information to make design
iterations. The problem therein is that the prototype needed to allow as many design iteration
options as possible while still being static enough to serve as the final validation prototype by the
end of the project. To this respect, the iteration options were limited by the design of the
prototype modules which occurred before any empirical data and insight was available.
Overall the prototype worked well for which it was designed. The modular prototype
made it simple to exchange the convergent and divergent modules while the tests were being
conducted. The design of the material inlet tube also allowed the team to slide the aspiration
tube back and forth to easily change the constriction size. This made the tests take less time,
which increased the number of tests that could be run. The fit of the material inlet tube was ideal
but did pose some difficulties. The fit needed to be tight enough to minimize leaking but loose
enough for the user to slide the tube by hand. In the end there was enough leaking that the joint
needed to be taped before each test however if the fit was any tighter to address this, the team
would have been unable to slide the tube by hand. In the future, tightening the fit and applying a
lubricant may increase the performance of the inlet tube assembly.
There were also a few problems with the modular prototype. Since the parts were made
individually by hand, they did not always fit together perfectly. Gaps were present where
different modules interfaced which had to be taped to prevent air from escaping. The geometry
was also not perfect since the modules were made by hand. For instance, careful examination of
the constriction shows that it is not perfectly constant around the circumference of the Venturi
because the convergent sections were not always perfectly circular. These problems would not
occur in a Venturi that was fabricated as a solid piece for a customer, but does add a degree of
conservativeness to the results. The modular prototype made it easy to test the different variables
in the new Venturi, which was the best way to optimize the design. Pictures of the modular
prototype were taken during testing and can be seen in the appendix.
Test Plan
Plan Used
In order to validate the new optimized Venturi conveyor, it was necessary to keep the test
rig identical to the rig used for all other design iteration validations and most importantly,
identical to the rig used to analyze the old PAC model. The proper assembly of the optimized
design can be referenced in the Modular Venturi Assembly section of Phase 3 and is available as
2
an engineering drawing provided separately from this report. In addition to the Venturi
Assembly, a detailed description of the rest of the test rig is available in the Optimization Testing
section of Phase 3. As mentioned in Phase 3, the fan was ran at a frequency of 30 Hz and
choked down to accomplish the desired of 5500 FPM (approximately 2.3 inches of water) at the
material inlet. This desired velocity pressure was satisfied when the manometer pressure
readings peaked at about 2.7 inches of water (an entire velocity profile measurement was
necessary to confirm that the desired average velocity was achieved).
To carry out a validation test, a specific procedure was devised and followed consistently
to ensure a basis of comparison between different tests:
1. Assemble the test rig, Venturi being tested, and engage the blower.
2. Search the Venturi and rig carefully for leaks and tape to seal before proceding.
3. Vary the blower frequency and/or butterfly valve until the average inlet velocity
pressure is approximately 2.3 inches of water (occurs about when peak velocity
pressure is 2.7 inches of water).
4. Using the hand-held double pitot-tube and digital manometer, measure the
minimum and maximum velocity pressure at three to five points evenly spaced
between the pipe wall and pipe center to establish a velocity half-profile for each
of the three measurement stations (material inlet tube, blower inlet tube, and
venturi outlet tube). Take every measurement at least five times to achieve
statistical confidence. E.g. for the three inch inlet pipe, measure at say zero inches
up against the side wall of the pipe, 0.5 inches into the pipe, 1 inch, and finally 1.5
inches into the pipe. If after at least two full trials, the profile shape in a section is
remaining constant, the point in the profile of average velocity can be calculated
and used as the singular measurement point instead of measuring a profile.
5. Again with the hand-held pitot tube and manometer, measure the minimum and
maximum static pressure at each of the three stations at least five times for
statistical confidence.
Problems and Future Improvements
When analyzing airflow in the Venturi conveyor system, the nature of the airflow is very
important. It had been previously expected that the flow in the system would largely be laminar
(the Reynolds Number is less than about 2040 and there is no lateral mixing of the flow). After
the first few rounds of testing, the team quickly made the conclusion that the flow in all sections
of piping was highly turbulent. Using the calculation for the Reynolds Number ( *V*D/ ). The
known constants for air at 20 C were: = 3.6*10^-5 lb/in^3 and dynamic viscosity = 9.97*10^-7
lb/(in*s). The values of velocity were based off of velocity pressure and the diameter of the pipe
was based on the location of where a velocity pressure was measured. The minimum velocity
found in the blower pipe of our system was calculated as 2953 FPM (with D = 5 inches). The
Reynolds number of this airflow is 106,627. Within the inlet pipe (D = 3 inches), the minimum
3
velocity was 5032 FPM. This equated to a Reynolds number of 109,018. Downstream in the
outlet (D = 6 inches), the minimum calculated velocity was 3560 FPM. The Reynolds number
for airflow in this pipe is 154,254. The values of the Reynolds Numbers were far within the
turbulent regime and it is the opinion of the design team that this will always be the case for an
effective Venturi conveyor.
Due to the turbulent flow in all pipes of the system, the profiles observed varied with
small system or design changes and did not correlated to conventional shapes (the turbulent flow
would need significant entry length in a straight pipe to develop to a conventional velocity
profile). This made data collection difficult because some measurement readings for velocity
pressure were higher closer to the wall than at the center of the flow (counterintuitive
considering the no-slip condition of flow in a pipe at the pipe walls). As one would expect,
velocity should be greatest at the center due the lowest friction loss from the wall of the piping.
One impact of turbulence in the system is on a team’s the ability to repeat and accurately collect
data. The turbulent flow led to fluctuations in the dynamic pressure readings (lesser in the static
pressure readings but still evident). Taking the minimum and maximum for five different trials
helped counteract this fluctuation however fluctuations with longer time periods are likely and
would not have been normalized with this method.
To address the issues that arise from the highly turbulent flow within the system, a
measurement procedure could be devised to use a computer setup to take readings over longer
periods of time to obtain more steady averages. This would require measurement instruments
that can interface with a computer (unlike the pitot tubes and manometers). Turbine flow meters
or ultrasonic flow meters would not only interface with a computer but would also address the
problem that the pitot tubes can only take pressure readings at points. Turbine and Ultrasonic
flow meters can measure the average of the velocity over the pipe diameter and a computer could
average these readings over time. For the static pressures, simple pressure transducers could be
mounted on the pipe walls and interface with the same measurement computer. While all these
improvements would have a dramatic effect on the resulting data accuracy and time necessary to
run validation (or analysis) tests on a system, they are highly expensive compared to the pitot
tubes and manometers used for this project. It would be up to PAC to determine in the future if
the added pay-off of higher accuracy and less testing time would be worth the up-front costs of
designing, purchasing, and assembling such a setup.
Analysis and Interpretation
Methods
In order to interpret the results of the validation tests and understand how they translate to
the key design wants and needs, the data must be analyzed to calculate key performance metrics
and add analytical context that is useful for general understanding and investigation and may
prove vital for selection worksheet population later. A procedure has been established to reduce
4
the recorded data to descriptive information and key metrics (Microsoft Excel files used to
computationally perform the analysis will also be provided to PAC):
1. Calculate the statistical average of the maximum and minimum values for each different
measurement taken.
2. For any velocity pressure measurements that were taken over a profile across the pipe
diameter, calculate the average velocity pressure at the station by mirroring the
measurements across the pipe centerline and taking the statistical average.
3. Use formulas from PAC’s Static Calculator utility or other engineering desk reference to
calculate velocities and flow rates at each of the three stations based on the velocity
pressure and respective pipe diameter.
4. Calculate the efficiency factor used for the efficiency metric by dividing the inlet velocity
pressure by the blower total pressure (the sum of the blower static pressure and blower
velocity pressure measured).
5. Calculate the ratio of outlet to inlet velocity used for the effectiveness metric.
Results/Conclusion
Initially, the Box-Behnken statistical optimization method was used to recommend a
modular Venturi combination, which significantly reduced the number of initial tests required.
After collecting data for all modular combinations necessary for the method, PAC’s current
model was tested as a baseline for comparison of the performance of the different design options.
Once the Box-Behnken method theoretically recommended a certain combination to be the
optimal design, it was tested to validate the results (it had not been one of the combinations that
needed to be tested initially for the statistical method). The results of the optimized Venturi
configuration according to the Box-Behnken method seemed to be 5-10% more efficient than the
old PAC model. However, the new configuration was also compared to other design
configuration options, specifically designs with different material inlet lengths (see the Theory
Review section to understand why this is the most sensitive design variable). When the
performance was measured at different material inlet positions, the results did not support the
choice of optimal position according the Box-Behnken statistical method.
It was also necessary to test whether the Box-Behnken method effectively captured the
optimal convergent and divergent module selections; therefore, the design was iterated to include
larger convergent and divergent modules (following the principles discussed in the Theory
Review section) and the validation testing was repeated. Similar to the case of the material inlet
position, the larger modules performed with higher efficiencies than the configuration
recommended by the Box-Behnken method. This once again disproved the findings utilizing
Box-Behnken as the primary optimization method (it was determined that the quadratic fitting
used within the method was insufficient to accurately model the complex relationship between
the design variables and the Venturi performance). The optimal inlet length was determined by
manually adjusting the inlet length and running the validation test. After optimizing all
components of the modular Venturi and inlet length manually, an optimized design was selected
5
and seemed to show an efficiency increase of 0.10 compared to the analysis of PAC’s original
Venturi model (new model efficiency factor equaled 0.29 and the old model was thought to have
an efficiency of 0.19). No further iterations were deemed necessary to optimize the Venturi due
to the significant increase in efficiency.
Using the optimal convergent and divergent modules and inlet length, one final rigorous
validation test was performed on December 9th to simply confirm the results of the same test
performed previously when the team was optimizing the inlet length. As another check of firm
confidence before concluding the project, the old PAC model was to be retested as well even
though it had been evaluated on two occasions previously. It was during this retest that an
experimental setup mistake was discovered in the original benchmark tests of the old PAC
model. During the original benchmark test (completed about one month prior to project
completion date), a six-inch pipe was used to connect the fan to the blower inlet. At the end of
this pipe, a six-inch flange was used to attach the pipe to the five inch flange on the blower inlet
of the old Venturi model. This interface of between the six inch flange and 5 inch flanged inlet
created an abrupt pipe diameter change with a half-inch lip inside the pipe normal to the
direction of the flow.
Figure 1: Flange mistake on original old PAC model benchmark tests.
This error resulted in a significant amount of erroneous efficiency loss in the first
benchmark tests of the old PAC model. The efficiency was calculated to be 0.19 with the flange
mistake. The increased static pressure in the blower inlet pipe from airflow impacting the lip at
the flange mis-match dramatically decreased the calculated efficiency of the primary benchmark
(see the Theory Review section for the efficiency equation) and the basis of comparison used for
all previous design iteration decisions. After correcting the issue with the flange and using a
five-inch blower pipe to connect to the Venturi, the old model was tested again to re-evaluate the
efficiency benchmark. The efficiency of PAC’s old model was found to actually be 0.29. This
was noticeably higher than our current optimized design. Had the benchmark been corrected
sooner, the team is confident that further design iterations could have further increased the
6
efficiency of the new model above the benchmark (ideas and insight into possible iterations
discussed in the transition plan).
At this point the design team re-evaluated the design iteration decisions made previously
based on a flawed benchmark however the time available significantly limited the design change
options available (no further modules could be fabricated). No design changes with the modules
available could further increase the efficiency of the new model. The final design: a fourteen-
degree convergent section and five-degree divergent section yielded an optimal efficiency for an
inlet length of 1.75 inches. The efficiency for this combination was 0.29, the same as PAC’s
current design. The velocity ratio (outlet velocity/inlet velocity) was 0.68, nearly identical to
PAC’s (0.69). If the flange mistake had been noticed sooner with more time available, further
iterations would have allowed the team to redesign the body module to be less constrictive and
cause less of a pressure drop prior to the constriction (further discussion of this design iteration
suggestion is discussed in the Transition Plan section).
Model Efficiency Output/Input Cost to
Factor Velocity Manufacture
Old PAC (with Flange Mistake) 0.19 +/- 0.73 +/- 0.03 $336
0.01
Old PAC (newly confirmed on Dec. 0.29 +/- 0.68 +/- 0.03 $336
9) 0.01
New Design 0.29 +/- 0.68 +/- 0.03 $485
0.01
Table 1: Final Results Summary
Theory Review
One of the secondary objectives of this project was to apply fluid dynamical principles to
gather further understanding of the theory of operation behind the Venturi conveyor.
Throughout this theory review, it is important to keep in mind the equation used to define
the efficiency of the Venturi conveyor:
Two equations in fluid dynamics were used most extensively throughout the project:
Bernoulli’s equation and the equation of mass conservation.
Bernoulli’s Equation
7
Mass Conservation
Bernoulli’s equation is a statement of conservation of energy for fluid flow. In the case of
the Venturi system’s internal flow, this equation is not perfectly accurate as flow is not perfectly
incompressible but comprises an excellent approximation suitable for the purposes of these
analyses.
Conservation of mass simply explains that mass inflow must equal mass outflow.
Assuming that density remains constant, one can simplify the equation to show that net flow rate
in must equal net flow rate out.
Figure 2: Venturi Theory Schematic
To fully understand the Venturi system, the figure above shows a schematic of the
theoretical internal flow. The equations above can be used to explain the Venturi effect: as flow
is constricted, the velocity must be increased to conserve mass flow. According to the Bernoulli
equation, as velocity increases, the static pressure must decrease. It is this large drop in static
pressure at the constriction that creates the pressure differential from atmosphere required to
induce flow in the material inlet pipe.
The key to engineering an optimally efficient Venturi conveyor is to minimize the loss
term in the Bernoulli equation. The greater the losses in the system, the less velocity change that
ends up translating to pressure change to create suction in the constriction. If the losses approach
zero (for proof of concept, not practically possible), the entire increase in velocity pressure
(1/2[ ]*V^2 term in the Bernoulli equation) will be conserved as decrease in static pressure to
induce flow.
8
It is also beneficial to think through the effects the three primary design variables have on
the fluid flow within the Venturi. Two of these variables were the divergence angle and
convergence angle. The divergence angle only controls the separation of flow as it enters the
section. If the angle is too large, the momentum of the highly turbulent flow exiting the
constriction will carry it forward faster than it will expand laterally to keep up with the
increasing diameter. The result is that the flow only follows half of the pipe wall and separates
from the other half with turbulent eddy’s forming to fill the space. The effect can be visualized
with a figured borrowed from Kibicho and Sayers’ paper, “Experimental Measurements of the
Mean Flow Field in Wide-Angled Diffusers: A Data Bank Contribution”:
Figure 3: Axial Velocity Profile measured in a 50° Diffuser (Kibicho 337).
Notice that the flow velocity is concentrated on the upward wall of the diffuser. In practice one
would also notice that along the bottom wall, turbulent eddy’s with some back-flow would form.
To avoid the high losses associated with this effect, the half-angle of a diffuser (like the
divergent module used in the Venturi prototype) should be within 5°-7° i.e. the lower the angle is
designed to be, the smoother the transition, and the lower the effective losses will be. The
convergent section is bounded by the same principle but has been found to be much less dramatic
because there is no threat of flow separation. The recommended range of convergent half-angles
is 14°-20° to avoid a dramatic increase in losses at higher angles.
The material inlet length was found to have the greatest effect on the efficiency and
performance of the Venturi conveyor. This is due to the fact that the position of the inlet tube
actually controls the size of the constriction in the shape of an annulus around the perimeter of
the end of the inlet tube inside the body of the Venturi. The cross-sectional area available to the
blower air flow is actually the smallest at this point, not at the interface of the convergent and
divergent sections. As the inlet is moved in or out, this constriction shrinks and grows
respectively. The effects comprise a classical engineering trade-off: As the constriction is shrunk,
the Venturi effect is increased which increases efficiency (smaller constriction, higher velocity,
lower pressure, greater induced suction). However, a smaller constriction also requires more
static pressure applied from the blower to maintain flow which decreases efficiency. As a result,
the position has to be carefully optimized to achieve the greatest Venturi efficiency.
9
Transition Plan/Path Forward
Design Iteration
As evident in the previous discussion of the final validation test results, the latest design
of the new Venturi conveyor fails to exhibit an increased efficiency over the previous model;
however, most recently the design team identified a likely flaw in the model that already exhibits
an efficiency equal to the benchmark. Therefore iterating the design to correct the flaw is very
likely to improve the efficiency beyond the benchmark.
Reducing the pressure losses seen by the blower air due to viscous and frictional forces
would reduce the required blower pressure to induce the inlet velocity. In doing so, the efficiency
is increased (since deflections of the flow create friction, this is the same principle that motivated
the 45 degree blower inlet). Losses in the pipe and Venturi scale directly with flow velocity so as
the blower air is constricted and the velocity increases, so do the losses. Therefore, it is
beneficial to avoid constricting the blower airflow until absolutely necessary to create the
Venturi effect (i.e. within the convergent section).
Figure 4: Venturi Body Photo
Note that in the current design where the material inlet tube and blower inlet tube
converge to form the conveyor body, the blower inlet tube diameter remains constant even with
the material inlet tube occupying cross-sectional area within the body. The blower air is
constricted as soon as the material inlet tube enters the body even though it does not need to be
constricted until the convergent section to induce suction.
10
Figure 5: Possible further design iteration to correct body flaw.
By increasing the volume of the body around the material inlet tube, the area available for
blower flow to occupy can be kept closer to constant before the convergent section, decreasing
the velocity in that section and reducing the losses. The expectation is that this would ultimately
increase efficiency. A prototype body module should be fabricated with a diverging conical
blower inlet tube whose diameter increases from five inches to seven inches at no more than a
seven-degree divergent angle and body is comprised of a seven-inch diameter conical section
that leads to the convergent module. Convergent modules would also need to be modified to
converge from seven inches to three inches with a fourteen-degree convergent angle. The
validation test could then be repeated to investigate the final effect on efficiency.
It may also be beneficial to work with the manufacturer to decrease the permissible gauge
of the steel used for the Venturi. The current 16-gauge steel can handle up to 26,000 inches of
water before being in danger of yielding. If the Venturi can be welded together with smaller
gauge steel (the thinner steel does pose difficulties for the manufacturer to make smooth welds),
it will surely still be able to withstand the maximum static pressures and significant cost
reductions in shipping, handling, and production costs could be possible.
It is necessary to establish a final design that maximizes the efficiency during operation
and minimizes shipping, handling, production, and installation costs to PAC. At least one more
design should be prototyped and validated (possibly more if possible improvements are still
evident) before proceeding.
Empirical Testing for Venturi Selection Calculator
In order to effectively provide customers with Venturi-conveyed systems that boast
increased efficiency, PAC engineers require a method to identify the fan required to supply
adequate flow through systems of various static pressure losses upstream and downstream of the
Venturi.
The nature of the effect of static pressure loss in the system on required blower energy
(i.e. flow rate and static pressure) was briefly investigated qualitatively in laboratory however a
full quantitative investigation was not possible within the time constraint created when the flange
mistake was discovered during final validation. The most cost-effective method of understanding
11
and quantifying the particular relationship will be to empirically test the Venturi with different
static pressure losses attached to the system to practically measure the required blower flow rate
and static pressure to achieve required velocities at the inlet and outlet.
The team has provided an example of a Venturi selection calculator that could be
populated with such empirical data and an interpolation function derived by using Excel or
another utility to apply a trend line (most likely a second to fourth order polynomial fit). Testing
should use the static pressure on the inlet and outlet as the two independent variables and the
required blower flow rate, required blower static pressure, inlet flow rate, and outlet flow rate as
dependent variables. For proof of concept, hypothetical data was used to populate the provided
Venturi selection calculator however this data has little experimental basis and should not be
used for actual Venturi selection.
Market Introduction
The primary market value added as a result of the new Venturi design is due to its unique
configuration and appearance compared to other models on the market. The team is also
confident that with one further design iteration, the final model will boast increased efficiency
compared to the industry standard (i.e. PAC’s old model). Marketing that highlights those two
primary points, the unique appearance and increased performance for less cost, can affiliate the
PAC name, already known for some of the best cutters on the market, with unique innovation
that translates directly to cost savings for their customers in the Venturi market. Standing out
among competitors and directly saving customers money can increase PAC’s Venturi market
share and increase sales ultimately increasing PAC profits.
Patent Consideration
The Venturi design even before further iterations is unique in the industry of large-scale
blower based Venturi conveyors. The market value discussed previously only exists if the design
remains unique to PAC’s product offerings. If competitors were to offer similar models, PAC’s
market share would likely return to its level prior to this project. Therefore, it may prove
beneficial to the long term profits to consider applying for a U.S. Patent to protect the investment
made and to be made in completing this Venturi conveyor overhaul.
12
Appendix
Large Prototype Photo
Stress Analysis
The stresses from the static pressure inside the venturi were calculated to ensure that the
venturi would last for infinite life. The venturi is made from 16 gauge steel. The two critical
locations on the venturi are the constriction and the blower inlet. The pressure is lowest in the
constriction, and since this is lower than atmosphere the venturi is in compression. The blower
inlet is in tension, because this is where the pressure is highest in the venturi. The venturi can be
13
thought of as a pressure vessel, so the hoop stress on the venturi can be found with the equation
=Pr/t. The yield strength of steel is 40000 psi, which is 1108330 inches of water. The
thickness of 16 gauge steel is .06 inches. That means the maximum static pressure at the blower
inlet is 960 psi, which is equal to 26600 inches of water. The minimum static pressure at the
constriction is 1600 psi below atmosphere, which is -44333 inches of water. The maximum and
minimum values that were found during testing were not close to these boundaries, so it can be
concluded that the stresses in the venturi are not relevant. The fatigue stress of steel is about
20000 psi, which is 554165 inches of water. The static pressures are much lower than the fatigue
strength, so the venturi will last for infinite life.
Error Analysis
Included in separate Excel data file.
14
References
Kibicho, Karanja, and Anothny Sayers. "Experimental Measurements of the Mean Flow Field in
Wide-Angled Diffusers: A Data Bank Contribution." World Academy of Science,
Engineering, and Technology 43 (2008): 332-37. Print.
15
