CHARACTERIZATION AND NONDIMENSIONAL ANALYSIS OF A

VARIABLE SPEED CENTRIFUGAL PUMP

A Thesis

presented to

the Faculty of the Graduate School

at the University of Missouri-Columbia

In Partial Fulfillment

of the Requirements for the Degree

Master of Science

Presented By

Michael Scheller

DECEMBER 2014

Graduate Advisor

Dr. A. Sherif El-Gizawy

i
The undersigned, appointed by the dean of the Graduate School, have examined the thesis
entitled

CHARACTERIZATION AND NONDIMENSIONAL ANALYSIS OF A VARIABLE SPEED CENTRIFUGAL

PUMP

presented by Michael Scheller,

a candidate for the degree of master of science

and hereby certify that, in their opinion, it is worthy of acceptance.

Professor Sherif El-Gizawy

Professor Hani Salim

Professor Yuyi Lin

Professor Billie Cunningham

Professor Roger Fales

 ACKNOWLEDGEMENTS

Firstly, I would like to thank my entire family for the support they’ve given me

throughout my entire graduate school process. I wasn’t sure they would be entire supportive of

my decision to attend graduate school instead of entering the work force. However, they stood

behind my decision to enter graduate school, and supported me with anything I needed while

completing my degree. I cannot thank them enough.

Next, I would like to thank all of the graduate and undergraduate students that have

helped me on this project.

- Shane Corl, a Doctoral Candidate for Dr. El-Gizawy, coordinated practically everything

that was done on this project. His countless hours programming and doing physical work

on the system made this project doable. Without him, this project would have gone

nowhere, and I am so thankful to have had his help.

- Bilal Hussain, an undergraduate student for Dr. El-Gizawy, helped me a great deal during

the experimental process for this project. We spent full days in our lab running

experiments and performing manual labor on the system, and without him, this project

would have been much more difficult. Thank you so much.

ii
 - Brian Graybill, Jake Harris, Andy Gunn, Tony Adamson, Annemarie Hoyer, Zhentao Xie,

Jay Shelby and Amer Krvavac. All of you have contributed something to this project, and

I really appreciate you helping in one manner or another. I am so happy to call you all

my friends. Thank you!

- My committee members, Dr. Yuyi Lin and Dr. Billie Cunningham, thank you for

volunteering your time in order to aid me in my pursuit of higher learning. I am grateful

for your thoughtful critique of my work, and I really do appreciate it.

Finally, I would like to thank Dr. El-Gizawy. I have worked with Dr. El-Gizawy for five years

spanning across my undergraduate and graduate programs, and I couldn’t have asked for a

better advisor. His resourcefulness when it comes to finding new projects is unparalleled. In

addition, the experience that I’ve gained through working with industry officials and

government employees is extremely valuable to me. I know that I would not have had these

opportunities without Dr. El-Gizawy, and I am very thankful that he sponsored my learning for

the past five years. Thank you so much.

iii
 TABLE OF CONTENTS

ACKNOWLEDGEMENTS…………………………………………………………………………………………………….………ii

LIST OF ILLUSTRATIONS……………………………………………………………………………………………………….….v

ABSTRACT……………………….…………………………………………….…………………………………………………………vii

CHAPTER ONE – INTRODUCTION TO CENTRIFUGAL PUMPS AND LITERATURE REVIEW… ...……..1

CHAPTER TWO – EXPERIMENTAL TEST RIG DESIGN…………………………………………….…………………12

CHAPTER THREE – STANDARD PUMP CHARACTERIZATION, RESULTS, AND ANALYSIS…….…..…27

CHAPTER FOUR – NONDIMENSIONAL PUMP CHARACTERIZATION, RESULTS, AND ANALYSIS..52

CHAPTER FIVE – OVERALL PROJECT ANALYSIS AND CONCLUSIONS………….………………………..….66

REFERENCES...........................................................................................................................…...70

APPENDIX A – SYSTEM PICTURES…………………………………..……………………..……………………………….71

APPENDIX B – MATLAB CODE…………………………………………………………………………………………………77

 LIST OF ILLUSTRATIONS

FIGURE………………………………………………………………………………………………………………..PAGE

1. Working Drawing of a Centrifugal Pump…………………………………………………..………….2

2. Image of a modern centrifugal pump impeller………………………………………………………3
3. Comparison of a modern impeller to a vintage impeller………….……………….……………4
4. Diagram showing impeller and volute………………………………….………………….…………..5
5. Standard Single Speed Pump Characterization………………………………………….………….7
6. Example Standard Variable Speed Pump Characterization………………………….…….….9
7. Initial Experimental Test Rig Design Proposal…………………………………………….……….13
8. Front View of Pump Canister………………………………………………………………………….……14
9. Top View of Pump Canister…………………………………………………………………….……………15
10. Completed Test Rig………………………………………………………………………………..……………16
11. System Static Pressure Sensors………………………………………………………..………………….18
12. In-Line Flow Sensor……………………………………………………………………..………………………19
13. RTD (Resistance Thermometer Detector) …………………………………..……………………….20
14. Accelerometers and Centrifugal Pump…………………………………………………………………21
15. Labview Interface……………………………………………………………………..………………………..24
16. Standard Single Speed Pump Characteriziation………………………….……………………….27
17. Initial Experimental Matrix………………………………………………………………………………….31
18. General Performance Plot Generated From First Experiments……..………………………33
19. Capacity-Efficiency Plot From Initial Experimental Matrix…………….…………………….34
20. Three Dimensional Plot of Capacity, Head, and Efficiency……………….…………………..38
21. Plot of Rotational Pump Speed vs. Pump Power Consumption…………………………….40
22. Plot of Capacity Versus Bulk Fluid Temperature…………………………….…………………….41
23. Plot of Capacity Versus Head, Using Valve Angle………………………………………………..44
24. Plot of Voltage and Efficiency for Various Valve Angles…………………….…….………….45
25. Plot of Capacity and Efficiency for Various Valve Angles…………………..…………………47
26. Three Dimensional Surface Plot of Standard Pump Characterization….……………….48
27. Plot of Pump Rotational Speed (RPM) Versus Pump Power Consumption…..…….…50
28. Chart of Specific Speed BEP Values……………………………………………………….…………….55
29. Plot of Specific Speed versus Efficiency (Constant Angle Lines) …………….…………….57
30. Plot of Specific Speed versus Efficiency (Constant RPM Lines) ……………….……..…….59
31. Chart Showing The Impellers Performance Versus Other Impeller Types…….……….61

v
32. Image Showing The Centrifugal Pump Impeller and Vane Shape…………..…....….……62
33. Three Dimensional Surface Plot of Non-Dimensional Pump Characterization……….63

vi
 ABSTRACT

When dealing with centrifugal pumps in the aerospace field, current industry standards

dictate the use of single speed drive pumps. The performance of these pumps is then

modulated using simply an electro-mechanical valve, which is controlled by an aircraft’s

onboard CPU. The French pump manufacturer, Intertechnique, has provided the University of

Missouri with a variable speed pump for the purposes of this project.

The primary focus of this project is to create control logic which can keep the pump

operating at its best efficiency point. This control software will be developed control the speed

of the pump, as well as a control valve attached to the pump. By varying the speed of the

pump, as well as the amount of flow being allowed by the valve, the pump should be operating

most efficiently just through the use of the software being developed by this project. In order

to create this software, however, the pump must be completely understood. That is, the pump

must be fully characterized. The completion of this characterization will allow us to achieve and

capacity and head within our allowable performance range. By doing this, we will be able to

minimize power consumption of the pump, while simultaneously maximizing the life of the

pump.

The purpose of this project is to show how this variable speed pump performs over a

wide range of testing parameters. Plots showing capacity (volumetric flow rate), pump head

(pressure), temperature, vibration and efficiency will be discussed. In addition, experiments

vii
have been analyzed non-dimensionally, so that these results may be applied to any pump with

the same impeller shape. The experimental results were then analyzed against existing data in

the field of centrifugal pump performance. The significance of the experimental results to the

field of mechanical engineering is then discussed, showing the relative efficiencies of the testing

done on this pump, and where those efficiencies fall in comparison to other centrifugal pumps

viii
 Chapter One – Introduction and Literature Review

Before going into detailed analysis of characterizing a centrifugal pump, it becomes

important to fully understand how a centrifugal pump works. In the field of mechanical

engineering, a centrifugal pump is typically classified as a piece of turbomachinery, which is a

term used to describe a machine that transfers energy between a motor and a fluid. The range

of fluids and motor speeds for a centrifugal pump can vary significantly. For instance, the most

common fluid used as a motive fluid in centrifugal pumps is water. Pumping water has many

uses, such as in sewage systems, coal fired and nuclear power plants, and car radiators. In all of

these cases, a centrifugal pump drives the motive flow of the fluid. Centrifugal pumps can also

be used to pump other common fluids, such as petroleum or petrochemicals. In all of these

cases, the flow demand by an engineer can be created by simply attaching a centrifugal pump

to a line of fluid, and giving the centrifugal pump a source of power, whether that be electrical

or gas generated.

1
 Figure One – Working Drawing of a Centrifugal Pump

Figure one shows a cross section of a centrifugal pump, and all of the important

components of such a pump. As with any pump, a centrifugal pump has an inlet, outlet,

impeller, and a shaft attached to the impeller. All four of these components are crucial to the

operation of a pump; removing any of these four basic components would result in the failure

of the pump.

2
 Figure Two – Image of a modern centrifugal pump impeller

Figure Two shows a pump impeller, which will be the focal point of this thesis. An

impeller is a rotating disk which is the primary inducer of flow within the pump. It typically has

many curved pieces of metal attached to it, which are typically referred to as “vanes”. The

shape of a vane has a very significant impact on the type and velocity of the flow induced by the

pump. An impeller typically gets its rotational energy from an electrical or gas powered source,

which will then cause spinning on the shaft that runs through the center of the impeller. The

method of flow generation for a centrifugal pump is through the conversion of the rotational

energy of the impeller, into the kinetic energy of the flow of the fluid. This transfer of energy is

normally described in engineering as ‘pushing’ or ‘lifting’ of the fluid, as the spinning vanes of

the impeller are literally inducing the flow themselves.

3
 The concept of a centrifugal pump has been around since the late 15 th century, however

it was not fully put into practice until an Italian engineer named Francesco di Giorgio Martini

developed the first prototype of a centrifugal pump. However, for the two hundred years of the

existence of a centrifugal pump, the impellers utilized straight vanes. These straight vanes

caused extremely turbulent flow, as well as incredible amounts of cavitation within the pump.

Figure Three – Comparison of a modern impeller to a vintage impeller

The late 19th century saw a very significant change in the shape of vanes in a centrifugal

pump. Figure Three indicates this change very well, as it shows the innards of a modern

4
centrifugal pump, versus an impeller from the late 17th century. The impeller on the right has

straight vanes, which caused many issues with the lifespan of the pump, as well as the

performance of the pump. However, the late 19th century, as well as the British engineer John

Appold, brought about the curved vane structure, which allowed for a vast increase in both of

the aforementioned categories.

The spinning of the impeller, however, does not decide the direction of the flow.

Referring back to Figure One, an indicator pointing to a volute is on the diagram. A volute is

simply a swept portion of solid material that physically directs the fluid in the desired direction.

In most cases, the volute directs the fluid in the direction of the outlet of the pump. For the

purposes of this project, our volutes follow a similar pattern, which can be seen in Figure Four.

Figure Four – Diagram showing impeller and volute

The final necessary part to a working centrifugal pump is the motor, which drives the

rotational motion of the impeller. When discussing motors that can be used to drive a

centrifugal pump, there are many different classes that the motor can fall into. The vast

5
majority of pump motors will be powered with either a combustible, such as gasoline, or

powered using electricity. Another way in which the motor has to be identified is whether the

speed of the motor can be varied in some fashion. When discussing motor variation, most of

the pumps are invariable in speed – that is, they are single speed pumps.

Since single speed pumps are the current industry standard for mass moving of fluids, it

is important to discuss single speed pump properties, as well as why they are normally chosen

for pumping demands. This is due to the fact that single speed pumps have been characterized

in depth for every impeller size, shape, as well as motor speed. The characterizing of these

pumps has been an ongoing engineering project since the early 20 th century. Therefore, there is

a massive amount of data which currently exists on nearly any type of pump conceived

possible. In addition, the control of a single speed centrifugal pump is significantly easier to deal

with than a variable speed pump.

Creating a control system for a single speed pump is very well understood, and can be

done very easily, given that the engineer knows the general performance characteristics of the

pump they are working with. In addition, the flow demand for a single speed pump can be

lowered to the correct amount by attaching a valve to the pump, and turning the valve to

throttle the flow to the correct amount. If the engineer wishes to create a greater amount of

flow, a single speed pump is also much more simple to work with than a variable speed pump.

A normal centrifugal pump will come with multiple impellers, all with different vane shapes, as

well as overall impeller diameters. These different impellers can be easily installed on any

existing single speed pump, given that the impeller is compatible with the existing motor. A

6
final factor which causes engineers to typically choose single speed centrifugal pumps over

variable speed centrifugal pumps is that, in general, they are significantly cheaper. These

reasons are all contributing factors as to why the current industry standard for pumping is to

use a single speed pump.

Figure Five – Standard Single Speed Pump Characterization

As mentioned above, the characterization for a single speed pump is generally much

more simple than when dealing with a multiple speed drive pump. This is because a pump

characterization curve, as seen in Figure Five, can be created with a very small amount of

experiments. When purchasing a commercial centrifugal pump, part of the documentation that

will come with the pump is a set of pump characterization curves. You will always receive a

number of characterization sets equal to the number of impellers that came with your pump.

7
For example, if you purchase a new pump with three impellers, it will come with a set of

characterization curves for each different impeller. This is due to the fact that different impeller

diameters, as well as different impeller shapes, will have a major impact on the performance of

a given pump. However, the general trends seen on these pump characterization curves will

always been the same. Therefore, when designing the experiments in order to characterize the

pump used in this research project, it became important to keep in mind exactly what

experiments would allow us to generate similar curves.

Pump performance curves are always structured in the same way, in order to maintain

consistency within the industry. A pump performance curve will always have capacity (flow

rate) on the x-axis, and any of a number of other variables on the y-axis. Typical variables that

can be plotted on the y-axis include head (pressure rise across the pump), pump efficiency,

power consumption of the pump, heat generation of the pump, as well as the net positive

suction head of the pump. As stated earlier, the shape of these variables should be generally

the same, regardless of the impeller shape or size being used. Knowing this fact is a good way

to check the accuracy of any pump testing, as a deviation from the general shape expected

indicates an error somewhere in the experimental process.

Due to the fact that the vast majority of pumps are single speed driven, there is only a

fair amount of literature currently available on variable speed centrifugal pumps. A variable

speed pump is characterized in a similar fashion to a single speed pump, however, instead of

different impeller shapes and diameters being used, pump speed is varied instead.

8
 Figure Six – Example Standard Variable Speed Pump Characterization

Figure Six shows an excellent example of a factory characterization for an existing

variable speed pump. There are a couple interesting things to note here. The first thing that an

engineer should notice is that the general setup for a variable speed pump performance graph

is the same as in a single speed graph. That is, capacity (flow rate) is on the x-axis, and head

(pressure rise) is on the y-axis. In this aspect, the characterization is more or less the same. The

typical variables that are plotted are the same as a single speed pump as well – Figure Six shows

head-flow curves for a variable speed pump. However, instead of one line per variable per plot,

9
there are multiple lines per variable per plot, with each line representing a different speed at

which the pump can operate. In general, ‘speed 1’ is the lowest speed at which the pump can

operate, while ‘speed n’ is the highest speed at which the pump can operate, where n is the

number of different speeds that the pump is being tested at. For the purposes of the

characterization being done on this project, the plots with lines representing different

rotational velocities can also be interpreted as three dimensional surface plots, with capacity on

the x-axis, head on the y-axis, and a third variable (efficiency, temperature, vibration

parameters) on the z-axis.

Current literature suggests that the three dimensional plotting methods discussed

above are unique to this project, as it is difficult to find anything similar to these types of plots

in literature. There are some contour plots representing these three dimensional plots in two

dimensional form, but using a three dimensional plot to characterize a pump would allow an

engineer to have a greater sense of accuracy when attempting to identify which type of flow

rate they would need for a desired efficiency output.

While there is a significant amount of literature on characterizing a variable speed

centrifugal pump, there is currently very little of the literature that has been non-

dimensionalized. Non-dimensionalizing data is a process which turns data that has some

measurement associated with it (length, time, velocity, etc.) into a measurement which has no

measurement associated with it, yet it still has a numerical value which represents something

about the measurement. Turning a measurement into a non-dimensional number is useful for

many reasons. The most obvious reason that an engineer would want to non-dimensionalize

10
data, is that it allows the engineer to apply data universally across similar types of projects. The

most famous example of a non-dimensionalized unit is the Reynolds Number, which is used to

describe fluid flow. The research in this project is much more useful when a non-dimensional

speed is applied to the rotational velocity of the pump. Instead of having revolutions per

minute as pump speed, it is much more useful to have the non-dimensional number “specific

speed” represent the pump speed. In characterizing data this way, this allows anyone with a

similar shaped impeller and vane to apply this data directly to that specific pump, regardless of

the actual size of the pump. Representing data in a non-dimensional fashion, as well as the

techniques that were used will be discussed at length in a later chapter.

While there is a very significant amount of work that has already been completed in

characterizing single speed pumps, there is significantly less data available in characterizing

variable speed pumps, simply due to the fact that they are not currently chosen in most

aerospace and mechanical engineering projects. Therefore, the characterization done in this

project will be used to show exactly how using a variable speed pump can use less energy, as

well as providing a much more accurate flow demand than a single speed pump. In addition,

the fact that the work on this project has been non-dimensionalized means that something

meaningful will be added to the field of engineering, as any engineer with a similar shaped

pump impeller will be able to apply the results from this project directly to another project.

11
 Chapter Two – Test Rig And Experimental Design

Before the design of our test rig could be completed, it became important to address a

few pressure questions. The French company Intertechnique provided very little information to

go with the centrifugal pump used on this project. The information that was provided regarded

the general size of the pump, in addition to some information regarding standard operating

conditions and procedures for the pump. This was certainly not enough to design a full test rig.

Instead, many important questions needed answered, such as what variables will need to be

measured, what types of sensors will need to be purchased, and who will do the technical work

in building system.

The important variables which will need to be measured include flow rate, pressure,

temperature, and vibration. In addition, various data acquisition devices and programs needed

to be identified well in advance of the actual design of the system, in order to insure that all of

our hardware would be fully compatible with each other. After identification of the variables

that were necessary to measure, selection of the actual hardware was the next task.

After selection of the sensors necessary for data collection, it became important to

identify the actual programs and devices that were going to be used to collect the data. For

data acquisition purposes for the experimental test rig, as well as data transferring, National

Instruments devices were chosen. National Instruments DAQ devices are used currently the

engineering standard for data collection and acquisition, so the choice was very simple.

12
 Figure Seven – Initial Experimental Test Rig Design Proposal

Figure Seven shows the initial test rig design, which was submitted to the project leads

for approval during May 2012. The first thing that should be examined is the top view of the

experimental test system. Examination of the far most left drum shows a rectangle in the

bottom of the tank. This rectangle represents the variable speed centrifugal pump, as the pump

being used in this study actually operates under submersible conditions. That is, the pump has

to be submersed for it to produce any flow whatsoever. This was a very important fact to keep

in mind when designing our experimental test rig system.

The first challenge that the engineers tried to address was leakages throughout the

system. The most obvious point of leakage will be where the pump is physically mounted to the

13
bottom of the left most tank. A very significant amount of time was spent trying to get the

system to seal at this point.

Figure Eight – Front View of Pump Canister

14
 Figure Nine – Top View of Pump Canister

The pump canister drawings seen in Figures Eight and Nine provide the details necessary

for installation of this pump into a system. Figure Nine shows that there are 12 equidistant

screws on the flange of the pump canister. Knowing this, 12 equidistant threaded holes were

made on the bottom of our left most 55 gallon drum. The accuracy of the distance and

diameter of these holes had to be extremely precise, as even a small error would make it

impossible for the pump to install correctly into this machined area.

The initial attempt to get the pump to seal in this area was unsuccessful, as oil slowly

leaked out of every screw hole. This initial approach was correct, however, and it was simply a

case of needing to add another anti-leak mechanism. The method was chose was to machine a

“ring” of metal, approximately 1 inch thick, that would be mounted on the bottom of the pump.

15
Then, instead of using short screws, longer bolts could be used. The additional threads on the

bolts allowed for the connection between the system and the pump to finally correctly seal. In

addition, this ring will allow the pump to be more stable, which will allow for more accurate

accelerometer readings, which will be discussed at a later point.

Figure Ten – Completed Test Rig

Figure Ten shows what the test rig looks like after work was completed on it. The total

construction time of the system was about one year from the beginning of the project, as the

construction proved to be much more difficult than expected. Figure Ten shows that the fluid

housing tanks are simply 55 gallon oil drums, which are set approximately 8 feet apart, and

16
connected using welding fittings on the outside of the tanks, and standard 1 inch internal

diameter PVC piping between the tanks. The physical placement of the sensors can be seen in

our diagram as well. The location of each sensor was carefully thought out before placing them

in our system. For example, our temperature sensors are placed in such a way that they

measure the inlet bulk temperature of the fluid, as well as the outlet temperature of the fluid

as soon as the fluid leaves the tank, for the most accurate reading of temperature rise across

the pump. The pressure sensors have been positioned in a similar manner, in order to get the

most accurate reading of pressure rise across the pump directly. If the pressure or temperature

sensors were placed further down stream, our data would be very susceptible to losses. For

example, the longer fluid travels through any section of piping, the natural skin friction

between the fluid and the inside of the pipe causes progressive pressure losses. Therefore,

sensor locations play a major role in the accuracy of the data that we collect off of the system,

which in turn will affect the accuracy of the characterizations performed on our pump.

Further examination of Figure Ten will confirm the diagram shown in Figure Seven. The

centrifugal pump is located in the bottom of the left hand tank, which will be discussed at

length later in this paper. The final test rig does have some deviations from the initial plan,

however. On Figure Ten, there are two vertical clear PVC pieces of piping attached to each 55

gallon drum. These are commonly referred to as viewing chambers, as they allow the engineers

to see how much fluid is located in each drum at any given time. The reason these viewing

chambers were added is to confirm the readings of the sensors which tell the engineers how

much fluid is located in each tank.

17
 Another modification to the system is the added lab power supply, which can also been

seen in Figure Ten. The pump uses a maximum of 32 volts, which initially lead to a power supply

sizing estimate of 50 volts would be suffice for this project. However, the research team

severely underestimated the amount of power that would be used from the various fluid

property sensors. Therefore, a much large power supply had to be chosen. Figure Ten shows

that the power supply is large not only in power output, but in size as well, as it takes up a

significant amount of space in the lab.

Figure Eleven – System Static Pressure Sensors

In addition to the temperature and pressure sensors needing be located properly, the

flow sensors also have to be placed in the correct locations in the system. In order for flow from

a pressure source, such as the pump in this research, to be fully developed, any flow sensing

element should be placed at least five pump diameters downstream from the pressure source.

18
Since this system uses 1 inch internal diameter PVC pipe, every flow sensor has to be at least

five inches downstream from the pump, in order to ensure flow data accuracy. [1] This is

indicated on the diagram very clearly, as both the flow sensor measuring the flow output of the

pump, and the flow sensor measuring the flow back into the main tank are at least 10 inches

away from the pressure source.

Figure Twelve – In-Line Flow Sensor

When choosing the temperature devices that would be selected for the purposes of

measuring the inlet and outlet temperature of the fluid, a number of attempts were needed to

get an accurate temperature reading. The first issue that became apparent was leakage

associated with taking a temperature measurement inside of a moving fluid. As with any

pressurized fluid system, leakages are extremely likely to occur. Due to the “oily” nature of the

19
fluid being used in this system, leakages are even more likely to occur than if water was the

motive fluid being used in this project. The first attempt to take a temperature measurement

involved the use of simple thermocouples. Installing the thermocouple became the issue at

hand, however, as drilling into PVC pipe in order to install a thermocouple makes the pipe

almost impossible to seal. After multiple failed attempts to seal a PVC pipe with a thermocouple

installed in it, the engineers decided to find a better option. Ultimately, it was decided to use a

Resistance Temperature Detector (RTD) to measure the temperature of the fluid. An RTD

comes with a threaded nipple attached to it, in addition to a threaded female adapter. From

there, it became very easy to drill a hole in the PVC, attach the female adapter to the pipe, and

screw in the RTD. This option was significantly more expensive that using a thermocouple,

however, it was the only option that successfully prevented leaks.

Figure Thirteen – RTD (Resistance Thermometer Detector)

20
 Another type of sensor was used on this project was an accelerometer. An

accelerometer measures the amount of acceleration, typically in “g-forces”, in a single axial

direction. In this project, an accelerometer was mounted on the sides of the pump in both the x

and y direction, in order to see if there are any anomalies in the vibration of the pump.

Figure Fourteen – Accelerometers and Centrifugal Pump

21
 Figure Fourteen shows how the accelerometers are oriented in the system, as well as

showing the centrifugal pump they are attached to. The accelerometers in this project are used

for two purposes. The first purpose is to measure how the pump itself is vibrating as the

impeller is spinning. Any cantilevered structure, such as the pump in this project, will have

different modes of vibration, and it is important to measure those modes properly, which is

what the accelerometers help measure. The second purpose of the accelerometers is measure

the revolutions per minute of the pump. The pump has no built in reading of how quickly the

pump is operating. When vibration sensing devices are used, the comparison frequency source

should be either line frequency, where stable, or a stable independent frequency and the speed

determined by observation. [2] Relating this statement to the system, the best way to measure

the RPM of the pump is to look at the vibration of the pump, and from there, use a Fast Fourier

Transform to determine the rate at which the impeller is spinning. This Fast Fourier Transform

analysis can be used to read which frequency has the most noise associated with it. The

frequency with the most noise is the frequency at which the pump is spinning. The use of this

Fast Fourier Transform analysis is a necessity in this project, as there isn’t another viable way to

determine the RPM of the pump.

Figure Fourteen also shows copper tubing wrapped in a helical shape inside of the left

most 55 gallon drum. This copper tubing is filled with water and attached to a heat exchanger

outside of the system. The reason this feature was added to the experimental test rig is

because the bulk fluid temperature needed to be the same at the start of every experiment.

However, running an experiment dumps a large amount of heat into the fluid, causing its bulk

temperature to rise very rapidly. The only way to cool the fluid back to room temperature in a

22
time efficient manner is to manually cool it off using the copper tubing heat exchanger. Before

this heat exchanger was installed in the system, the fluid would take least 10 minutes between

experiments for the fluid to cool down to the appropriate temperature. However,

approximately 2 minutes of running this heat exchanger is suffice for cooling the fluid down to

the appropriate temperature.

Returning to Figure Ten, there are a few modifications which were to the system made

after the system was deemed in working order. Examination of the bottom of the system shows

two plastic pallets, which were added after much of the testing had already been completed.

These plastic pallets are used to catch any fluid that may leak from the system, or spill when

adding additional fluid. In addition to these pallets, a metal stand was added underneath the

left most tank. This tank is the one that holds the centrifugal pump, and the stand was added

underneath it in order to allow easy access to the pump. The pump screws into the bottom of

the tank, however, it was impossible to actually access the pump once the system had fluid in

it. The left tank is usually filled with 55 gallons of 1010 mineral oil, commonly known as Brayco

460. The total weight of this fluid is approximately 400 pounds, making it impossible to actually

lift the tank in order to access the pump. This metal stand alleviated the problem, and proved

to be very helpful once it was installed.

The physical system construction proved to be the most difficult task for the purposes of

this project. However, the completed construction of the system did not mean that it was ready

to be tested, as a way to interface between the sensors and a computing system was still

needed. For the purposes of this project, the engineers decided to use the “Labview 2014” for

23
data collection purposes. This program allows data to be collected from an outside source of

hardware. In the case of this project, the hardware pieces are simply the various flow

characteristic sensors, in addition the variable speed centrifugal pump. The program can be

tuned to the exact needs of the engineer, and can be used either as a data collection resource,

or a way to actually control hardware. For the purposes of this project, Labview 2014 was used

in both manners. Initially, the program was used simply as a data collection tool. However, as

the research progressed, Labview 2014 was used as a way to control flow demand, as well as

the angle of valves in the system. Using Labview in this manner allowed for the creation of a

large number of flow types, which allowed for a great range of variables to be used when

designing the experiments.

Figure Fifteen – Labview Interface

24
 Figure Fifteen shows a screenshot of the Labview program. This portion of the program

is called the control panel, as many parts of the experimental test rig can be controlled from

this portion of the program. On this figure, you will see buttons which open or close the

primary control valve, control the angular velocity of the pump, control whether or not the

pump is on or off, and control whether or not data is currently being collected. Having all of the

experimental options located centrally allows the experimenters to change various flow and

valve conditions very quickly, allowing for a large number of unique experiments to be

completed in a relatively short amount of time. In addition to this portion of the program being

able to control various parts of the system, it is also used to read the data being outputted by

the sensors. Examination of Figure Fifteen will show multiple plots for temperature, pressure,

flowrate, and vibration data. All of this data is analyzed in real time, and plotted to the control

panel graphs in Labview 2014. Having all of the graphs centrally located allows the researchers

on this project to quickly determine if there is a problem with the system. In addition, having

the plots organized in this fashion allows the researchers to determine general flow trends of

the system, just from watching these graphs change during testing.

Though the construction of the experimental test rig was a long and difficult process,

the researchers on this project are extremely happy with the finished product of the system.

Overall, the system meets the strict requirements of quality set for us by the advisor. In

addition, the system allows us to collect very accurate data, due to the fact that the sensors are

located in very close proximity to the DAQ devices, and the computer collecting the data.

Setting up the system in this manner means that only a small amount of wiring was used, which

means that there is significantly less signal loss than there normally would be in an

25
experimental set up. In general, however, the system produces very accurate data in a simple

time saving manner. The overall design of the system makes running an experiment extremely

simple, as just about everything can be controlled from the local computer, which has allowed

the results gathered in this project to be very accurate.

26
Chapter Three – Standard Pump Characterization, Results and
Analysis

Characterizing a pump’s performance has been an area of scientific study since the

centrifugal pump was introduced into modern society. The earliest attempts at pump

characterization appear to be linked to the first attempts to create water and sewage systems

in major cities. This is logical, as engineers in this situation would need to understand how

much fluid can be moved in a certain amount of time. The easiest way to understand the

performance of the pumps in this situation would be to characterize the pump in a useful

manner. Performance analysis requires repeatable measurements of process parameters such

as temperature, pressure, flow, displacement, speed, power and time. [3]

Figure Sixteen – Standard Single Speed Pump Characteriziation

27
 Figure Sixteen shows an example of a standard pump characterization for a single speed

pump. Any centrifugal pump manufacturer will include a characterization similar to the one

seen in Figure Sixteen in the package with the pump itself. The layout of this characterization is

standard across the entire field of mechanical engineering, meaning that any purchased pump

will have the same variables plotted on the graph, such as pump efficiency and head capacity.

The first thing to notice on Figure Sixteen is that the X-axis is labeled “Flow Per Minute”. Many

manufacturers will refer to volumetric flow rate as “flow per minute”. However, for the

purposes of this paper, volumetric flow rate will be referred to as “capacity”. Capacity is simply

a measurement of how much fluid can be moved over a standard unit of time, either in seconds

or minutes. The standard measurement of capacity in the English system is gallons per minute,

usually referred to in centrifugal pump literature as “GPM”.

Next, the Y-axis should be examined. On Figure Sixteen, the Y-axis is labeled as “total

dynamic head”. Most engineering textbooks have the y-axis of a centrifugal pump

characterization graph labeled as simply “head”. Head is an engineering term used to describe

the amount of pressure being generated from the pump. The measurement of head is simply

the pressure at the outlet of the pump minus the pressure at the inlet of the pump. When

advertising the “size” of a centrifugal pump, any centrifugal pump manufacturer will “rate”

their pump to the maximum head and capacity. In doing this, any person that knows their flow

demand will be easily able to select a pump that fits their needs.

In order to perform the characterization for a single speed pump, a valve must be

attached to the outlet of the pump. From here, varying the valve position creates a change in

28
flow rate. For example, when the valve is fully open, the pump will operate at its maximum

capacity. However, closing the valve will change the amount of fluid being moved by the pump,

simply due to the obstruction of the fluid’s path. In addition to the valve being used to change

the capacity of the pump, changing how intrusive the obstruction is will change the head rise

across the pump as well. As the valve is closed, the pump will naturally see a drastic rise in

head, as fluid being pumped will become heavily pressurized as it moves through a closing

orifice. This pressurization of the fluid must be counterbalanced by the pump in order to

maintain flow in the direction it was already travelling. As such, the head rise across the pump

will increase during this phenomena, which can be seen as the “System Head Curve” plot on

Figure Sixteen.

The other variables plotted on a typical centrifugal pump characterization curve are

pump power consumption, and pump efficiency. Both of these variables are represented on

Figure Sixteen as “Motor Input” and “Pump Efficiency”, respectively. The general trends of the

plotted variables shown on the example pump characterization are another important thing to

note from this plot. Any time a pump is being characterized, the variables should follow the

same shapes as the plots shown in the example. For instance, the shape of the pump efficiency

curve should always be an upside down parabola. If the plot is not shaped in this manner, there

is an issue with the way the characterization was performed. Either there was a problem with

the experiments themselves, or a problem with the data that was collected off the pump. Either

way, there is clearly an issue that needs to be addressed if the plots do not manifest in a similar

manner to the one shown above. This fact was very helpful when performing the analysis on

29
the experimental centrifugal pump. If the data from the pump did not match the general

shapes shown in the example plot, a mistake was clearly made.

The pump being used in this study is, as mentioned previously, a variable speed

centrifugal pump. A variable speed pump is a pump that increases its rotational speed, in

revolutions per minute, as a greater input power is supplied to the pump. As discussed above, a

single speed pump is generally characterized by attaching a valve to the pump, and varying the

valve position. In doing this, a great range of capacities and head values can be generated. A

variable speed pump is characterized in the exact same manner. The difference is that pump

speed is a second independent variable that has been introduced into the characterization.

Therefore, in the case of variable speed pumps, the independent variables in the

characterization are control valve angle and the rotational speed of the pump.

There are two independent variables in the system, which causes a performance plot to

shift from a standard two-dimensional plot into a three-dimensional plot, with two

independent variables and one dependent variable. Generally, it is hard for an engineer to get

anything useful from a three-dimensional plot, so most three dimensional plots are typically

turned into a two-dimensional contour plot. A contour plot simply turns a three dimensional

plot into a two dimensional plot using level contour lines, which are placed at user defined

increments.

As discussed above, the two independent variables in the system are the valve position

and the rotational speed of the centrifugal pump. Logically, it was determined that varying

these two independent variables is an excellent spot to begin designing experiments. The

30
centrifugal pump operates on a rotational speed between six thousand and eleven thousand

revolutions per minute. It was decided for the purposes of this project that varying the

rotational speed of the pump every one thousand revolutions per minute would be an

acceptable resolution for this project. As far as the control valve is concerned, obviously any

valve has two extreme positions of fully open and fully closed. For the purposes of this project,

a position of zero degrees is considered to be fully closed, while a position of ninety degrees is

considered to be fully open. For the purposes of this project, it was decided that an appropriate

resolution for the experiments would be varying the valve angle in ten degree increments.

RPM
6000 7000 8000 9000 10000 11000
0 1 11 21 31 41 51
10 2 12 22 32 42 52
20 3 13 23 33 43 53
30 4 14 24 34 44 54
40 5 15 25 35 45 55
Angle
50 6 16 26 36 46 56
60 7 17 27 37 47 57
70 8 18 28 38 48 58
80 9 19 29 39 49 59
90 10 20 30 40 50 60

Figure Seventeen – Initial Experimental Matrix

Figure Seventeen shows the initial experimental matrix that was created, based on the

resolutions stated above. This matrix is very simple, and systematically goes through all of the

different combinations of experiments. This initial experimental matrix has a total of sixty

different experiments that were run, which is quite a large number of unique experiments.

31
These initial experiments were intended to give a good base line for what should be expected

from the pump performance wise.

After designing this experimental matrix, it became important to determine a method of

testing that would ensure accuracy of the data. When testing any type of fluid flow, any

variation in testing will create a very large error in the accuracy of the data collected. In the

past, fluid flow experimental data has been ruined due to changes in the environment around

the system. For example, a change in the weather can cause the air pressure in the room to

change up to as much as half of a pound per square inch. In order to negate this problem, an

extra pressure sensor was added to measure the ambient air pressure in the room. Then, the

difference between this ambient pressure and the pressure in the system is measured as the

“gauge” pressure of the system. In measuring “gauge” pressure, the risk of any environmental

parameters affecting the system is eliminated.

Another way in which the accuracy of the data was insured is that the experiments were

completed in a relatively short amount of time. The entire experimental matrix shown in Figure

Seventeen was completed in the span of a week. Completing the experiments in this small time

frame allowed us to establish a routine for the most efficient testing. For example, the

engineers were able to determine that the fluid needed approximately five minutes of cooling

time between experimental runs. During these five minutes, the settings of the speed of the

pump and the angle of the pump were set to the appropriate values. In addition, the data

collected on the previous experiment was examined for any anomalies. If a problem was

detected, the previous experiment was run again instead of moving onto the next experiment.

32
 Figure Eighteen – General Performance Plot Generated From First Experiments

Figure Eighteen is the first plot generated from the initial experimental test matrix. It

shows the performance of the centrifugal pump based on standard centrifugal pump

parameters, with capacity on the X-axis and head on the Y-axis. As mentioned previously in this

paper, the variation in the capacity of the system comes from the variation of the angle of the

control valve attached to the centrifugal pump. Referring back to Figure Sixteen, the user of this

chart can clearly see that the trend represented by the head-capacity plot is matched by the

trend seen in Figure Eighteen. As stated earlier, if the shape of the plot did not match the shape

of the plot represented in various literature, then a problem with the data has occurred and

would have to be rerun. However, this is not the case, and the data is indeed usable. A few

33
things should be examined on Figure Eighteen. The first is that a lower capacity, along any given

RPM value, always results in a higher head value. This is due to the fact that the lower capacity

is driven by the control valve. A small capacity means that the valve is choking down the

amount of flow being generated by the pump, which in turn creates a greater amount of stress

on the pump. In order to counteract this stress, the pump will produce a greater amount of

head. This phenomenon is well documented in literature, and will be seen throughout the

presentation of this research.

Figure Nineteen – Capacity-Efficiency Plot From Initial Experimental Matrix

Figure Nineteen is the capacity versus efficiency plot from the first round of testing.

Again, referring back to Figure Sixteen, the user of the graph can see that the general shaping

34
and trends of the experimentally generated plot match the trends expected by literature.

Although the shaping seen in this figure is what the expected shape, the actual results of the

experimental data are shifted downward from the expected results. Further examination of

Figure Sixteen will lead the user to see that a general expected value of a peak efficiency to be

around 50 percent. However, the centrifugal pump reaches a maximum of approximately 30

percent. This reduced efficiency is not a sign of experimental failure, however. Instead, it is

simply the result of a small capacity pump being used. The example plot in Figure Sixteen is

generated from a pump with a much higher capacity than the one being used in this project.

This phenomenon can be understood by examining the equation used to calculate pump

efficiency in this project. As with any efficiency, this pump’s efficiency is measured as a ratio of

the output power to the input power. In addition, pump efficiency can only be derived from the

power input required. [5]

where

and

The equations shown above are all based on the fact that any efficiency is always a ratio

of the output power to the input power. From this initial ratio, it becomes important to identify

exactly what represents the input and output for a centrifugal pump. Literature suggests that

35
the output of a pump is always given defined in terms of “work”, while the input to a pump is

always some form of horsepower, which is electrical horsepower for the purposes of this

project. Examination of the “pump work” equation will show the user that the two

measurements of “flow rate” and “head” are variables which are being directly measured in the

experimental test rig, which allows for easy calculation of the amount of work being done by

the pump at any given time. The final variable which must be determined in order to make the

calculation of pump work is the “specific gravity of fluid”. Specific gravity is a ratio of the

density of the fluid to the density of water. A specific gravity below 1 means that the fluid is less

dense than water, while a specific gravity above 1 means that the fluid is more dense than

water. The specific gravity for any fluid can be looked up in any fluid mechanics textbook, or by

contacting the manufacturer. Brayco 460, which is more common known as 1010 mineral oil, is

being used in this project, meaning that this fluid does not occur naturally. Therefore,

consultation with the manufacturer became necessary, which resulted in a specific gravity of

0.87 being used as the specific gravity value for this project.

The final equation which had to be satisfied in order to create a satisfactory value of

pump efficiency is the horsepower equation. The wattage input is whatever the users of the

test rig determine using the lab power supply. The pump operates on a range of 14 to 32 volts,

which can be converted to watts by multiplying the voltage value by the current value of the

system. Finally, the motor efficiency needs to be determined. Determining the motor efficiency

proved to be a difficult challenge for the research group. After running a couple of experiments

to determine this number, the researchers determined that a viable value for this variable

could not be determined. Instead, it became much easier to contact the manufacturer and ask

36
about the horsepower rating of the pump. As it turns out, the horsepower value for the pump is

exactly 1 horsepower, as determined through technical data provided to us from

Intertechnique.

Examination of these 3 equations can be used to understand why an efficiency curve has

a inverse parabolic shape. First, an efficiency curve has a parabolic shape due to the value of

motor efficiency. A motor always operates at its most efficient point in the center of its

operating range. This fact is why cars get a higher amount of miles per gallon in the 40 to 50

mile per hour range, and get significantly less miles per gallon above or below this range. The

shape of a motor’s efficiency is always parabolic in nature, as well. Being that the motor

efficiency is in the denominator of the pump efficiency equation, the shape of any pump’s

efficiency will always be an inverted parabola. This phenomenon can be seen in the example

pump characterization in Figure Sixteen, as well as the results generated in Figure Nineteen.

37
Figure Twenty – Three Dimensional Plot of Capacity, Head, and Efficiency

38
 Figure Twenty shows a three dimensional plot of the three most significant pump

performance characteristics, which are capacity, head, and efficiency. In most pump

characterizations, these variables are kept separate from each other, plotting only two of these

three variables against each other at a given time. In doing this, most pump characterizations

have three distinct plots which are looked at; capacity versus head, capacity versus efficiency,

and head versus efficiency. However, it would be even more useful to plot all of these variables

against each other at the same time. In doing this, a three dimensional plot is created,

something which seems to be unique to this project.

The benefits of presenting pump data in this fashion are numerous. However, properly

being able to read this graph requires a plotting software such as MATLAB, in order to correctly

find a desired location on the graph. A very distinct benefit of presenting results in this fashion

is that the pump’s maximum efficiency can be very easily determined when looking at the

results in this manner. Utilizing MATLAB or other software, a simple maximum command can be

run on the data generated from the experiments. This maximum command will find the global

maximum for pump efficiency. Once this global value has been determined, the values of

capacity and head which cause this maximum can then be linked to this point. Only the level of

fluid in the system, as well as the speed at which the pump is operating, would need to be

adjusted.

39
 Figure Twenty One – Plot of Rotational Pump Speed vs. Pump Power Consumption

Figure Twenty One shows a plot of the power consumption of the pump, based on

various capacities. This plot is extremely useful for an engineer for a number of reasons. For

example, a pump this size could be installed on a non-crucial system on a plane, such as waste

removal. If a situation were to occur that would require the fuel pumps to the plane’s engine to

provide more fuel, then it would be possible to lower the power consumption of this non-

crucial pump to a level which would allow the plane’s fuel pump to receive the power it needs

to maintain the operation desired by the pilots. An effective way to maximize energy cost

savings is to use the well known 80-20 rule. Applied to pumps, this means that 80% of power

consumption savings can be achieved from 20% of the pump systems. [4] Meaning, the top 20%

40
of pumps (in terms of power consumption) represent 80% of the cost reduction potential for

that facility. This should illustrate that small changes in power consumption for 1/5 of the

pumps currently in operation at a facility can mean major savings for that facility.

Figure Twenty One shows us that the power differential between 11000 RPM and 6000

RPM is, at a maximum, about 700 watts. This is amount of power which could be used in other

aircraft systems during times of high power demand from the aircraft, such as takeoff or

landing. In addition, if an emergency situation were to occur which would require the plane to

accelerate very quickly, this pump could be completely turned off by the aircrafts onboard

computer, in order to provide up to an additional 770 watts of power to other systems on the

aircraft.

Figure Twenty Two – Plot of Capacity Versus Bulk Fluid Temperature

41
 Figure Twenty Two is useful in determining heat generation of the variable speed

centrifugal pump. This plot shows capacity versus bulk fluid temperature of the pump, with the

thicker lines representing the temperature of the fluid at the inlet of the pump, and the thinner

lines representing the temperature of the fluid after the fluid has passed through the pump,

shortly after the outlet of the pump. In addition, it was noted that the bulk fluid temperature

does not increase throughout the length of a given experiment. Knowing this means that the

pump will not cause excessive change in fluid temperature for bulk amounts of fluid. This fact is

especially helpful when dealing with fluids that have a flammable temperature point,

commonly known as a “flash point”. The fluid on this project, Brayco 460, has a flash point of

133 degrees celcius. From Figure Twenty Two, it is possible to see that the fluid temperature on

this project does not exceed 23 degrees Celcius, which is well within the safe temperature

operating range for this project. This pump, however, can and will be used with any type of

aerospace lubricating fluid. This includes but is not limited to brake fluid and various coolants,

both of which generally have a flash point lower than that of Brayco 460. The results shown

indicate that unless the pump is operated for an extremely long amount of time continuously

(on the order of days), the pump will not generate a large enough amount of heat to cause the

operating fluid to reach a critical level.

Figures Eighteen through Twenty Two represent the relevant results generated from the

first experimental test matrix. The results from this first matrix are in the opinion very useful for

applications using this specific pump. However, it was decided that more data could be

generated which would give further insight into the performance of the variable speed

42
centrifugal pump. Since the first batch of data was plotted based on the rotational speed of the

pump, it was determined that it would be an excellent idea to have the next batch of data be

based on the angle of the control valve attached to the pump. From here, a batch of

experiments was designed to generate the data needed for this type of analysis. The

experiments involve simply setting the control valve to an angle between fully open, and 60

degrees closed. Closing the valve any further past 60 degrees closed results in a very low

amount of fluid actually passing through the orifice, which creates wildly inaccurate data. In

order to create data over the full range of angle values possible, it was decided to use valve

increments of every 10 degrees. In order to generate a change in data among every degree

increment, the LabView program was set up to vary the rotational speed of the pump from

6000 RPM to 11000 RPM over a time of two minutes. This experiment was run for every degree

increment described above.

43
 Figure Twenty Three – Plot of Capacity Versus Head, Using Valve Angle

Figure Twenty Three shows a standard capacity-head plot for the variable speed

centrifugal pump, using valve angle instead of pump rotational speed as seen previously. The

first thing that must be noted when using this chart is that the 90 degree increment indicates

the valve being “fully open”. In the same vein, the 30 degree increment indicates that the valve

is 60 degrees closed. Moving on to the actual data, this chart shows the engineers that as

capacity increases over a given time, the head generated by the pump increases as well. This

positive slope is maintained for every degree increment tested. However, the slope increases

drastically for the situations in which the valve is more closed than it is open. The phenomenon

of head increasing more quickly when the control valve is more closed is very easily explainable

44
and understood. With a smaller orifice to flow through, the pump must work harder to move

the same amount of fluid through a smaller hole than normal. In order for the pump to do this,

it must create higher amounts of head pressure. The converse of this statement is also true.

With a fully sized orifice, the head pressure required to move the same fluid is significantly less

than with a small orifice, which is illustrated quite clearly in Figure Twenty Three.

Unsurprisingly, this chart also clearly shows that the maximum capacity for a given valve angle

is much higher for situations in which the valve is fully open, as opposed to more closed.

Figure Twenty Four – Plot of Voltage and Efficiency for Various Valve Angles

45
 Figure Twenty Four is another standard pump characterization plot generated from the

second round of experiments. In this chart, pump voltage is plotted against efficiency, which

should give the user of the pump a good idea of the operating voltage which is associated with

efficient power consumption. The first trend that should be noted on this figure is that the

operating efficiency of the pump generally increases when the valve is more closed than it is

open. Remember, as stated during the analysis done on the previous figure, the line

representing 90 degrees is when the control valve is fully open. Therefore, the most efficient

control valve angle is actually more closed than it is open.

The next valve position that should be analyzed is the line representing 30 degrees on

this chart. Interestingly, the slope of this line is significantly (to the point of being noticeably

different) different from the lines representing the other control valve angles. Due to the fact

that the derivative of this line has larger absolutely value that the lines around it, this line spans

a greater amount of efficiency values than the lines around it. It is believed that this line

represents accurate data, and at the same time, leads to some conclusions as to why this data

is the way it is. The explanation arrived at is that the valve being 60 degrees closed creates a

very small orifice for the fluid to flow through. Generally, when fluid with laminar (smooth) flow

passes through a small orifice, that flow is then transitioned to turbulent flow. Turbulent flow is

fundamentally different from laminar flow, on a particle physics and mathematical level.

Turbulent flow is generally difficult to predict the behavior of, which should explain why the

plot for this specific valve angle plot is different than the ones around it.

46
 Figure Twenty Five – Plot of Capacity and Efficiency for Various Valve Angles

Figure Twenty Five shows another standard pump characterization plot, this time with

capacity on the x-axis and efficiency on the y-axis. The general trends shown on Figure Twenty

Fthe are again visible on Figure Twenty Five, with the fully open valve position (90 degrees)

generating the lowest pump efficiencies. Conversely, the control valve positions that are the

most closed (30 and 40 degrees) generate the highest pump efficiencies. Again, the derivatives

of the most closed valve positions have the highest absolute values for this chart. In addition, it

is possible to see that the 30 and 40 degree plots only span approximately 7 gallons per minute

of flow (in the x-direction). This is easily explainable, as a more closed control valve limits the

amount of flow through the valve orifice.

47
Figure Twenty Six – Three Dimensional Surface Plot of Standard Pump Characterization

48
 Figure Twenty Six is the culmination of the plots shown so far from the second round of

experiments. This plot shows a standard centrifugal pump characterization in three dimensional

form, plotting the three most relevant variables together all at the same time. As you can see

from the chart, capacity is on the x-axis, with head on the y-axis, and efficiency plotted on the z-

axis. As you can see, the chart has what appears to be “folds” on it. These folds are nothing

more than the various valve angles seen in the other figures up until this chart. The further left

“fold” represents a valve angle of 30 degrees, while the furthest right “fold” represents the

control valve in a fully open position. This chart would be extremely useful to any engineer

working on a centrifugal pump similar to this one, provided they had a reliable software

program that could read this chart. With MATLAB or another plotting aid, it would be easy to

determine the efficiency level that is required for a given project. From there, it would be

possible to look up the efficiency that is desired on Figure Twenty Six. Note that a given

efficiency likely occurs a large number of times on this Figure Twenty Six. Therefore, if the

experimental conditions dictated that an independent variable be a certain value, that value

would likely have the desired efficiency associated with it.

49
 Figure Twenty Seven – Plot of Pump Rotational Speed (RPM) Versus Pump Power
Consumption

The final plot that will be discussed in this chapter is Figure Twenty Seven, which shows

the rotational speed of the pump versus the amount of power the pump is consuming. The first

thing that the user of this chart may note is that the general trend, or shaping, of the different

angle lines on this plot is essentially the same. Each angle line has a positive first derivative,

which can be seen as the lines slope from the bottom left of the graph to the top right of the

graph. In addition, the graphs have a slight positive second derivative associated with them.

This can be seen in the slightly concave shaping of each line. Unsurprisingly, the pump

consumes more power when the rotational pump speed is higher for every case. The only

50
somewhat shocking trend that can be seen on this chart is that the pump utilizes less power

when the fluid is being pumped through a smaller orifice, which can be seen by the red line,

representing “30 degrees”, being on the lower end of power consumption for every pump

rotational speed. Initially, it was expected the pump to utilize more power when pumping into a

smaller orifice, as it seems like the pump would have to exert more work to move fluid through

a smaller space. However, this is not the case, as it can be seen that the pump actually

consumes more power when pumping the fluid through a completely open control valve.

The figures seen and discussed in this chapter provide the complete characterization for

the performance of this pump under any given circumstance. A normal single speed pump

characterization would be performed by the manufacturer before sale. The analysis done on

the variable speed pump, however, is significantly more complicated than a standard single

speed pump analysis, as a single speed pump analysis requires the creation of only a single

graph. The results discussed in this chapter show that a very large number of plots were

created in order to fully describe and characterize the performance of the variable speed

centrifugal pump. However, it was decided that non-dimensionalizing the results would bring

significantly more academic meaning to the research completed on this project.

51
 Chapter Four – Non-Dimensional Pump Characterization,
Results and Analysis

While any centrifugal pump will certainly go through extensive testing in order to arrive

at an acceptable pump characterization, very few of these pumps are characterized in a non-

dimensional fashion. Firstly, it becomes important to define non-dimensional characterization,

and next, it is important to understand the reasons and benefits from characterizing a pump

without dimensions.

When designing a centrifugal pump, a number of design variables need to be

determined. They are impeller rotational speed, impeller inlet or suction dimensions, impeller

outlet diameter, impeller blade number, impeller blade passage geometry, and impeller

position relative to the casing. [8] Surely, the shape and size of the impeller in a centrifugal

pump is extremely vital to determining the pump’s performance. However, these variables are

much more useful when determined in a non-dimensional format.

The development of the specific speed concept was a most important one in the history

of centrifugal pump development. At once, it became possible to utilize test and design data on

existing pumps to develop new designs of dimensionally similar pumps but of larger or smaller

size because the specific speed of a pump remains independent of size. [6] As the “non-

dimensional” clause suggests, non-dimensional characterization involves using variables that do

not have a unit associated with them. A non-dimensional variable is always created by using an

equation which causes all of the units to cancel each other out, leaving only a unitless number

52
as the result from the equation. These non-dimensional variables hold a special power

mathematically, as their lack of a unit allows for these numbers to be applied universally across

different applications. Firstly, for a counter-example of a non-dimensional variable, time is a

variable that has a standard dimension. It is easy to see that time is a dimensional variable, as

time always has a unit associated with it, which is usually “seconds” in the scientific community.

Unfortunately, there are no extremely common non-dimensional variables that are used

in every day life. The most famous non-dimensional variable is undoubtedly the Reynolds

number. The Reynolds number is a very famous fluid mechanics equation, which takes a

number of standard fluid mechanics variables into account. These variables are inserted into

the Reynolds number equation, and all of their dimensional qualities cancel each other out.

Running a calculation with the Reynolds number equation leaves a dimensionless quantity, as

suggested from the discussion thus far. However, just because this number does not have a

dimension associated with it, does not make this number useless. In fact, it is significantly more

useful than a dimensional quantity. A Reynolds number of below 2300 indicates that the flow

being studied is a smooth, laminar flow, while a Reynolds number of over 4000 indicates that

the flow is turbulent. This holds true for an astonishingly high number of fluid types, as well as

pipe materials and diameters. In addition to this number being able to explain the type of flow

expected from certain parameters, this number is also able to be used in scaling of projects.

Take for example, an aircraft wing. The flow over the wing will have a certain Reynolds number

associated with it. If the engineers suddenly decided to make the wing 10% bigger, the easiest

way to scale the wing while keeping the same flow performance is to use the Reynolds number.

Using the initial Reynolds number, a desired length of the wing could be set to 1.1 times the

53
original length of the wing, while keeping the desired Reynolds number the same. From here,

the Reynolds number equation could be rearranged in order to solve for the width or height of

the wing. From here, the width and height would be simply determined using basic algebra. In

this sense, the Reynolds number is extremely powerful, and without a doubt the most

commonly used dimensionless variable used in the field Mechanical Engineering.

While the Reynolds number is an excellent example of the power of a dimensionless

value, it is not very applicable to the research on a variable speed centrifugal pump. Instead,

centrifugal pump literature suggests that a value called “specific speed” is much more common

to research in this area. Specific speed is a value which is commonly used in characterizing the

operating speed of any piece of turbomachinery. A normal operating speed is typically defined

in terms of revolutions per minute (RPM), or angular velocity (radians per second). Obviously,

both of these quantities have dimensions associated with them, which greatly limits the ways

they can be used mathematically. However, the addition of specific speed as a variable into the

analysis allows for this research to be applied universally across all types of centrifugal pumps.

The catch with this type of non-dimensional number, however, is that the analysis performed

can only be applied to centrifugal pumps with the same shape of impeller. However, if a pump

does have the same shape of impeller as the one being used in this research, the use of specific

speed allows this research to be applied directly to the performance of that pump as well,

regardless of the size of the pump. The pump can be micro-sized, or very large, and the

performance of the pump should be the exact same when using specific speed as a unit of

analysis.

54
 √

The above equation is the unitless equation used to calculate specific speed, which is

denoted as in centrifugal pump literature. In this equation, n is pump rotational speed (in

radians per second), Q is flowrate (in US gallons per minute for this project), g is the

acceleration due to gravity (32.17 ft/s^2) and H is total head pressure (in pounds per square

inch). Using simple algebra, conversion of all of these variables into English Imperial units will

allow all of the units of each variable to cancel out in the final specific speed equation. As

suggested previously, the result of this equation with be a dimensionless number.

Figure Twenty Eight – Chart of Specific Speed BEP Values

Figure Twenty Eight shows a plot with the maximum range of specific speed values on

the number line above the images of various impeller shapes. A common type of analysis

performed using specific speed involves replacing H (head pressure) with H(BEP) (head pressure

at the best efficiency point). In doing this, a singular value of specific speed is generated, which

55
is then described as the specific speed BEP. This specific speed BEP value will then be forever

associated with that shape of impeller. Optimization of pump hydraulic geometry in terms of

the BEP specific speed has taken place empirically and analytically throughout the history of

pump development. [7] As stated earlier, it can be clearly seen that the maximum value of

specific speed is 15000, while the minimum is around 500. It is important to note that the

impellers shown on Figure Twenty Eight represent the impellers specific speed when operating

at its best efficiency point (BEP). Obviously, the non-dimensional analysis that will be performed

on this project will be analyzed over the centrifugal pump’s full range of specific speed values.

From here, it will be possible to determine the specific speed value which causes the pump to

operate most efficiently, which will then allow us to determine where the impeller type falls on

Figure Twenty Eight. This type of analysis is very important in determining the “area” in which

the pump operates. Once the area in which this pump operates has been determined, it will

allow other pump users to universally apply this research to other pumps that fall within the

same “area”.

56
 Figure Twenty Nine – Plot of Specific Speed versus Efficiency (Constant Angle Lines)

Figure Twenty Nine shows a non-dimensional plot of specific speed versus pump

efficiency. As you can see from the plot, the specific speed variable does not have a unit

associated with it. Because of this, any pump with the same shaped impeller would be able to

apply the data shown in Figure Twenty Nine directly to the performance of that pump. One

thing that should be noted on this figure is that the range of specific speed values for the

centrifugal pump being used in this project is between 6000 and 8500. Referencing Figure

Twenty Eight, it is possible to see that the maximum range of values for a pump’s specific speed

is anywhere from 500 to about 15000. It is possible from Figure Twenty Nine to see that the

more closed the control valve is, the higher the resultant efficiency is. The only anomaly is the

57
plot representing the “30 degree” valve position. This plot is interestingly below the 40 degree

valve position in terms of efficiency. It has been theorized the behavior of the pump for this

specific control valve angle is due to a transition from laminar (smooth) flow during the “40

degree” plot, to a turbulent flow during the “30 degree” plot. The turbulence of the fluid in this

30 degree plot will create increased friction between the fluid and the inside of the PVC piping

wall. In doing so, the efficiency at which the pump is operating will be lower than the

efficiencies seen on the “40 degree” plot.

58
 Figure Thirty – Plot of Specific Speed versus Efficiency (Constant RPM Lines)

Figure Thirty shows a plot of the same two variables seen in the previous figure, which

are specific speed being plotted versus efficiency. A quick comparison of the graphs shows that

the figures are indeed quite different when being plotted based on different trendlines. As you

can see, Figure Twenty Nine was plotted on a constant angle trendline basis. For Figure Thirty,

the same plot was generated, using constant RPM trendlines instead. The results shown in

Figure Thirty tell us a few key pieces of information about the variable speed centrifugal pump.

First, it is important to note that this type of analysis would generally be very difficult to do with

a single speed centrifugal pump. Examination of the specific speed equation should allow the

59
user to see that the “n” value represents pump rotational speed. In a single speed centrifugal

pump, the “n” value is completely fixed, which would make it impossible to create a matrix with

a range of specific speed values. However, the pump is variable speed, allowing for this “n”

value to be changed.

Moving on to the actual analysis of Figure Thirty, it is possible to see that the greatest

efficiencies occur around a specific speed value of 5000 for every RPM value. This result is a

very important step to completing the non-dimensional analysis of the experimental centrifugal

pump. The specific speed value of 5000 clearly holds a special place in fully understanding the

inner workings of the pump. It is important to remember, however, that the specific speed in

general tells engineers more about the performance of the shape of the impeller, and less

about the pump itself.

60
 Figure Thirty One – Chart Showing The Impellers Performance Versus Other Impeller Types

Since the impeller has a BEP (best efficiency point) Specific Speed value of 5000, it

becomes important to determine what flow region this corresponds to. Figure Thirty One

shows that the impeller falls in the region between “mixed flow” and pure “axial flow”. “Mixed

flow” generally refers to a type of flow that start out in the radial direction (outwards from the

center of the impeller) and ends in an axial direction, which is perpendicular to the direction of

flow as fluid comes off of the impeller. Conversely, “Axial Flow” refers to a type of flow that

heads in the direction of the axis of the impeller. This generally can only occur if the vanes are

pointing in the same direction of the axis of the impeller.

61
 Figure Thirty Two – Image Showing The Centrifugal Pump Impeller and Vane Shape

In an attempt to fully understand why the pump’s impeller falls in between the “mixed

flow” and “axial flow” regions on Figure Thirty One, Figure Thirty Two is called upon. The

important thing to examine on this figure is the shape of the vanes on the impeller. As stated

previously, it is important to remember that radial flow will occur if the vanes are straight

vertical (into the page). Conversely, axial flow will occur if the vanes lie in the same plane as the

face of the impeller (horizontally). Figure Thirty Two shows us that the vanes are almost

completely vertical. However, close examination will show the engineer that they indeed do

have a slight upward tilt. This realization allows the engineer to determine that this impeller

shape does indeed belong in the “Mixed Flow” region shown on Figure Thirty One. This result is

very exciting, as it helps us determine that the data pulled from the experiments, as well as the

analysis, has been accurate up to this point.

62
Figure Thirty Three – Three Dimensional Surface Plot of Non-Dimensional Pump
Characterization

63
 The final figure to be discussed in this chapter is Figure Thirty Three, which compiles the

results shown so far on non-dimensional analysis into a three dimensional plot. This plot is the

culmination of the non-dimensional analysis which has been performed on the centrifugal

pump, as the variables plotted in this chart are the 3 variables vital to fully characterizing a

pump. As this chart has been analyzed non-dimensionally, it becomes important to replace the

rotational speed of the pump (RPM) with Specific Speed, as it will allow the results of this

analysis to be applied to other centrifugal pumps with the same shape impeller. The “creases”

or “folds” shown on Figure Thirty three represent different valve positions, with the far left

crease representing the valve being 60 degrees closed, while the far right crease represents the

valve being fully open. It is important to keep in mind that this analysis would not be feasible on

a single speed centrifugal pump, as the variation of the specific speed variable would not be

possible on a pump with only one speed. Examination of Figure Thirty Three shows the user

highest efficiency operating point for this pump occurs at a specific speed value of 5000, as

stated previously, and a head value of approximately 25 PSIG.

Again, anyone with a pump that has the same shape of impeller, regardless of the size of

the pump, impeller, or vanes, could utilize the data presented in this chapter. The process of

utilizing this data with another pump would be extremely simple, as well. Since the results seen

in this chapter have been non-dimensionalized using the Specific Speed value, another pump

with the same shape impeller and vanes would need to be characterized non-dimensionally as

well. Once this has been done, however, the results seen here could directly applied to the new

pump, regardless of the actual size of the pump. Meaning, the pump could be on a micro or

macro scale, but the performance of the pump would be the same as long as the Specific Speed

64
values of the new pump matches the specific speed values discussed in this chapter. This

example illustrates the full power of utilizing a non-dimensional value, such as Specific Speed,

as data characterized in this manner can be applied to other research projects globally. The

classification of a variable speed centrifugal pump in a non-dimensional manner is novel to this

research project, specifically with three dimensional charts and contour plots being used.

65
 Chapter Five – Overall Project Analysis and Conclusions

As with any graduate research project, the overall goal of the project is always to add

something unique and useful to the chosen field of study. It is felt that the work presented in

this paper furthers the understanding of variable speed centrifugal pumps in a measurable way.

Firstly, the construction of the physical system for experimental testing shows that the students

on this project possess a level of technical ability rarely showcased by graduate students. The

work done on building this system, while not novel to the field of mechanical engineering, is not

to be underestimated. A very large amount of time was invested in preparing the system for

data collection, and along the way, many technical were learned.

The study of the inner workings of a centrifugal pump is another key element to the

research completed on this project as well. Through the work completed on the physical

construction of the system, centrifugal pumps as a mechanical device were studied in a very

detailed manner, as test rig and experimental design would be heavily influenced through

knowledge of how a centrifugal pump works. Information such as vane shape and size, impeller

shape and size, volute shape, pitch, and size played a great role in determining how the system

was physically constructed, as well as in determining the battery of experiments. The way this

knowledge is presented is common across all centrifugal pumps, whether they are single speed

or variable speed.

In addition to the specifications of single speed pumps, a great deal of information was

required to be learned a about variable speed pumps. While it turns out that the inner workings

of a variable speed pump are very similar to that of a single speed pump, that does not mean

66
that working with a variable speed pump would be equally as convenient. In fact, the addition

of a variable speed motor to a centrifugal pump made this project significantly more

challenging than it would have been otherwise. Mapping voltages to angular operating

velocities of the pump was certainly one of the most challenging aspects of this whole project.

Through experimentation, it turned out that there was no way to initially determine the speed

at which the pump was operating. The solution to this problem was very complex, as it became

necessary to measure the vibration of the pump. In using a fast fourier transform (FFT) on the

vibration data generated from the pump, it became possible to measure different resonant

frequencies that occur during pump operation. From here, it became possible to isolate the

resonant frequency responsible for the spinning motion of the impeller. Once this frequency

was determined, the amplitude of this frequency was then mapped to the speed of the

impeller, allowing a determination of exactly what speed the pump was operating at. This

breakthrough allowed for the beginnings of designing experiments based off pump speed,

which is one of the primary variables in which pumps are characterized with.

The experiments used to collect data were designed in a manner that allowed the most

common centrifugal pump variables to be the independent variables in the experiments. The

variables on this project that were the most controllable were determined to be pump

rotational speed, most commonly measured in revolutions per minute, and head, most

commonly measured in inches of fluid. As it turns out, these two variables are the variables

which play the largest roles in pump characterization. In fact, when purchasing an off the shelf

pump, the way of determining if the pump will meet specific flow demands is by examining the

head-flow curve. This type of plot is the most commonly cited figure in the field of centrifugal

67
pumps, as it will tell you the operating speeds of the pump, as well as the pressures that it

generates at these speeds. For a single speed pump, the head-flow curves generated are

presented in a two dimensional plot. However, a variable speed pump produces head-flow

curves in a three dimensional plot.

Over 120 experiments were run in order to fully map the performance characteristics of

this specific pump, and it was important to describe the pump in a way which allows this

research to be applied to other projects. In order to do this, the help of a non-dimensional

variable called specific speed was enlisted. Non-dimensional numbers allow equations and data

to be applied universally across projects, as long as certain conditions are met. The condition

that needs to be met in order to use the non-dimensional data presented in this paper is that

the shape of the vanes need to be the same in another pump. If another pump has similar

shaped vanes, then the research presented in this paper can be applied directly to the

performance of that pump. The most powerful part of a non-dimensional number is that it

allows the data to be applied to another pump regardless of the size of the pump. The pump

could be very large or very small, and the data would still be directly applicable to the

performance of the pump. In addition, three dimensional plots of pump performance were

created using this specific speed value. Presenting the research completed on this project in a

three dimensional manner, through the use of non-dimensional variables, is novel to this

project. The creation of the plots in this manner will hopefully influence other engineers around

the world to present their research in this manner, as it is a more robust way of presenting data

than traditional methods. A three dimensional plot using non-dimensional independent

variables will allow the user of such a plot to quickly determine the exact flow demand needed.

68
From here, the user of a plot could determine various other dependent variable values, such as

efficiency, temperature, or power consumption at a given flow demand. These graphs are

extremely powerful, and present the research done on this project in a clear and concise

manner.

69
 References

[1] Centrifugal Pumps (Newtonian Liquids): A Guide to Performance Evaluation. New York: American
Institute of Chemical Engineers, 2002. Print.

[2] Standards. New York: Hydraulic Institute, 1965. Print.

[3] Beebe, Raymond S. Predictive Maintenance of Pumps Using Condition Monitoring. Kidlington, Oxford,
UK: Elsevier Advanced Technology, 2004. Print.

[4] Warring, R. H. Pumps: Selection, Systems, and Applications. Houston, TX: Gulf Pub., 1984. Print.

[5] Bloch, Heinz P., and Allan R. Budris. Pump User's Handbook: Life Extension. Lilburn, GA: Fairmont,
2006. Print.

[6] Karassik, Igor J. Centrifugal Pump Clinic. New York: Dekker, 1989. Print.

[7] Karassik, Igor J. Pump Handbook. New York: McGraw-Hill, 2008. Print.

[8] Turton, R. K. Rotodynamic Pump Design. Cambridge: Cambridge UP, 1994. Print.

[9] Vachez, Mr. Jean-Francois. "Intertechnique Correspondence." Interview.

[10] Herbstreit, Michael. "Boeing Correspondence." Interview.

70
Appendix A – Variable Speed Centrifugal Pump Images,
Solidworks Renders and Test Rig Images

71
72
73
74
75
76
 Appendix B – MATLAB Code
Note : Raw data is needed for MATLAB programs to produce figures.

Constant RPM Experiment Analysis Program

function data_compile_crpm

rpm = [6,7,8,9,10,11];

rpm2 = rpm*10^3;
cal_d_p_m = 5.9994;
cal_d_p_a = -0.0385;
cal_m_p_m = 6.0018;
cal_m_p_a = -0.0425;
cal_p_p_m = 9.987;
cal_p_p_a = -0.089;
cal_a_p_m = 6.0072;
cal_a_p_a = -0.024;
cal_f_m = 6.4076;
cal_f_a = -1.2904;
cal_v_m = 4.981469;
cal_v_a = 0;
cal_c_m = 5.9427;
cal_c_a = 0.0432;
v_m = 4.98;
i_m = 5.9427;
i_a = 0.0432;
eff_cal = 0.435;
cal_tin_m = 20.35; cal_tin_a = 1.249;
cal_tout_m = 20.40; cal_tout_a = 1.347;
colors = ['r','g','k','b','c','m'];

for ii = 1:6
filename = [num2str(rpm(ii)),'000rpm.tdf'];
output = tdfread(filename);
dtp = output.Drain_Tank_Pressure*cal_d_p_m+cal_d_p_a;
mtp = output.Main_Tank_Pressure*cal_m_p_m+cal_m_p_a;
pp = output.Pump_Pressure*cal_p_p_m+cal_p_p_a;
ap = output.Ambient_Pressure*cal_a_p_m+cal_a_p_a;
flow = output.Flowrate*cal_f_m+cal_f_a;
t_in = output.Inlet_Temperature*cal_tin_m-cal_tin_a;
t_out = output.Outlet_Temperature*cal_tout_m-cal_tout_a;
tin(ii,1:430) = t_in(1:430);
tout(ii,1:430) = t_out(1:430);

77
 drain_p(ii,1:430) = dtp(1:430);
tank_p(ii,1:430) = mtp(1:430);
pressure_rise(ii,1:430) = pp(1:430)-mtp(1:430);
actual_v(ii,1:430) = output.Voltage(1:430)*v_m;
actual_i(ii,1:430) = output.Current(1:430)*i_m+i_a;
watts(ii,1:430) = actual_v(ii,1:430).*actual_i(ii,1:430);
gpm(ii,1:430) = flow(1:430);
efficency(ii,1:430) =
(0.435*gpm(ii,1:430).*pressure_rise(ii,1:430))./watts(ii,1:430);
ss(ii,1:430) =
rpm2(ii)*(gpm(ii,1:430).^.5)./(pressure_rise(ii,1:430).^0.75);
rpm4(ii,1:430) = linspace(6000,11000,430);

figure(1);plot(gpm(ii,1:430),pressure_rise(ii,1:430),'Color',colors(ii),'Line
Width',2);
xlabel('Capacity (Gallons Per Minute)');ylabel('Head (Pounds Per
Square Inch)');
legend('6000 RPM','7000 RPM','8000 RPM','9000 RPM','10000 RPM','11000
RPM')
title('Plot of Capacity and Head for Various Pump RPMs');hold on

figure(2);plot(gpm(ii,1:430),efficency(ii,1:430),'Color',colors(ii),'LineWidt
h',2);hold on
xlabel('Capacity (Gallons Per Minute)');ylabel('Efficency');
legend('6000 RPM','7000 RPM','8000 RPM','9000 RPM','10000 RPM','11000
RPM')
title('Plot of Capacity Versus Efficency for Various Pump RPMs');hold
on

figure(3);plot(ss(ii,1:430),efficency(ii,1:430),'Color',colors(ii),'LineWidth
',2);hold on
xlabel('Specific Speed (Unitless)');ylabel('Efficency');
legend('6000 RPM','7000 RPM','8000 RPM','9000 RPM','10000 RPM','11000
RPM')
title('Plot of Capacity Versus Efficency for Various Pump RPMs');hold
on

figure(6);plot(gpm(ii,1:430),tin(ii,1:430),'Color',colors(ii),'LineWidth',1);
hold on
xlabel('Capacity (GPM)');ylabel('Bulk Fluid Temperature (Degrees
Celcius)');
title('Plot of Inlet and Outlet Temperature Versus Capacity');hold on

plot(gpm(ii,1:430),tout(ii,1:430),'Color',colors(ii),'LineWidth',3);hold on;
legend('6000 RPM','6000 RPM','7000 RPM','7000 RPM','8000 RPM',...
'8000 RPM','9000 RPM','9000 RPM','10000 RPM','10000 RPM','11000
RPM','11000 RPM')

figure(9);plot(gpm(ii,1:430),watts(ii,1:430),'Color',colors(ii),'LineWidth',2
);hold on
xlabel('Capacity (GPM)');ylabel('Power Consumption (Watts)');

78
 legend('6000 RPM','7000 RPM','8000 RPM','9000 RPM','10000 RPM','11000
RPM')
title('Plot of Capacity Versus Efficency for Various Valve
Positions');hold on
end

figure(4)
surf(gpm,pressure_rise,efficency)
xlabel('Capacity (Gallons Per Minute)');ylabel('Head (Pounds Per Square
Inch)');
zlabel('Efficency');title('Surface Plot of Pump Performance');
figure(5)
surf(ss,pressure_rise,efficency)
xlabel('Specific Speed (Unitless)');ylabel('Head (Pounds Per Square Inch)');
zlabel('Efficency');title('Non-Dimensional Surface Plot of Pump
Performance');

end

Constant Position Analysis Program

function data_compile_cpos
close all;
rpm = [6,7,8,9,10,11];
deg = [30,40,50,60,70,80,90];
rpm2 = rpm*10^3;
cal_d_p_m = 5.9994;
cal_d_p_a = -0.0385;
cal_m_p_m = 6.0018;
cal_m_p_a = -0.0425;
cal_p_p_m = 9.987;
cal_p_p_a = -0.089;
cal_a_p_m = 6.0072;
cal_a_p_a = -0.024;
cal_f_m = 6.4076;
cal_f_a = -1.2904;
cal_v_m = 4.981469;
cal_v_a = 0;
cal_c_m = 5.9427;
cal_c_a = 0.0432;
v_m = 4.98;
i_m = 5.9427;
i_a = 0.0432;
eff_cal = 0.435;
colors = ['r','g','k','b','c','m','y'];
vol_m = 1103.3; vol_a = 1735.2;
for ii = 1:7
filename = [num2str(deg(ii)),' deg.tdf'];
output = tdfread(filename);
dtp = output.Drain_Tank_Pressure*cal_d_p_m+cal_d_p_a;

79
 mtp = output.Main_Tank_Pressure*cal_m_p_m+cal_m_p_a;
pp = output.Pump_Pressure*cal_p_p_m+cal_p_p_a;
ap = output.Ambient_Pressure*cal_a_p_m+cal_a_p_a;
flow = output.Flowrate*cal_f_m+cal_f_a;
drain_p(ii,200:648) = dtp(200:648);
tank_p(ii,200:648) = mtp(200:648);
pressure_rise(ii,200:648) = pp(200:648)-mtp(200:648);
actual_v(ii,200:648) = output.Voltage(200:648)*v_m;
actual_i(ii,200:648) = output.Current(200:648)*i_m+i_a;
watts(ii,200:648) = actual_v(ii,200:648).*actual_i(ii,200:648);
gpm(ii,200:648) = flow(200:648);
rpm3=(output.Voltage*vol_m+vol_a)';
rpm4(ii,200:648)=rpm3(200:648);
efficency(ii,200:648) =
(0.435*gpm(ii,200:648).*pressure_rise(ii,200:648))./watts(ii,200:648);
efficency(efficency >= 1|efficency <= 0)=0;
ss(ii,200:648) =
rpm4(ii,200:648).*(gpm(ii,200:648).^.5)./(pressure_rise(ii,200:648).^0.75);
ss(imag(ss)~=0)=0;

figure(1);plot(gpm(ii,200:648),pressure_rise(ii,200:648),'Color',colors(ii),'
LineWidth',2);
xlabel('Capacity (Gallons Per Minute)');ylabel('Head (Pounds Per
Square Inch)');
legend('30 Degrees','40 Degrees','50 Degrees','60 Degrees','70
Degrees','80 Degrees','90 Degrees')
title('Plot of Capacity and Head for Various Valve Angles');hold on

figure(2);plot(actual_v(ii,200:648),efficency(ii,200:648),'Color',colors(ii),
'LineWidth',2);hold on
xlabel('Pump Voltage');ylabel('Efficency');
legend('30 Degrees','40 Degrees','50 Degrees','60 Degrees','70
Degrees','80 Degrees','90 Degrees')
title('Plot of Voltage Versus Flow Rate for Various Valve
Angles');hold on

figure(3);plot(gpm(ii,200:648),efficency(ii,200:648),'Color',colors(ii),'Line
Width',2);hold on
xlabel('Capacity (Gallons Per Minute)');ylabel('Efficency');
legend('30 Degrees','40 Degrees','50 Degrees','60 Degrees','70
Degrees','80 Degrees','90 Degrees')
title('Plot of Capacity Versus Efficency for Various Pump RPMs');hold
on

figure(7);plot(rpm4(ii,200:648),efficency(ii,200:648),'Color',colors(ii),'Lin
eWidth',2);hold on
xlabel('Specific Speed');ylabel('Efficency');
legend('30 Degrees','40 Degrees','50 Degrees','60 Degrees','70
Degrees','80 Degrees','90 Degrees')
title('Plot of Capacity Versus Efficency for Various Valve
Positions');hold on

80
figure(9);plot(rpm4(ii,200:648),watts(ii,200:648),'Color',colors(ii),'LineWid
th',2);hold on
xlabel('Rotational Pump Speed (RPM)');ylabel('Power Consumption
(Watts)');
legend('30 Degrees','40 Degrees','50 Degrees','60 Degrees','70
Degrees','80 Degrees','90 Degrees')
title('Plot of Capacity Versus Efficency for Various Valve
Positions');hold on

end

figure(4)
surf(gpm(1:7,250:end),pressure_rise(1:7,250:end),efficency(1:7,250:end))
xlabel('Capacity (Gallons Per Minute)');ylabel('Head (Pounds Per Square
Inch)');
zlabel('Efficency');title('Surface Plot of Pump Performance');
figure(5)
surf(ss(1:7,250:end),pressure_rise(1:7,250:end),efficency(1:7,250:end))
xlabel('Specific Speed (Unitless)');ylabel('Head (Pounds Per Square Inch)');
zlabel('Efficency');title('Non-Dimensional Surface Plot of Pump
Performance');

end

81