display - Edge - Rochester Institute of Technology

Multidisciplinary Senior Design Conference
Kate Gleason College of Engineering
Rochester Institute of Technology
Rochester, New York 14623
Project Number: P15453
DYNAMIC JOURNAL BEARING LOADING SYSTEM
Christina Amendola
Mechanical Engineering
Michael Bush
Mechanical Engineering
Kevin Burnett
Mechanical Engineering
Anthony DePina
Mechanical Engineering
Molly Mariea
Mechanical Engineering
ABSTRACT
A journal bearing is a supporting sleeve which allows the formation of a lubrication film,
creating a low friction surface in which a shaft can freely rotate. Journal bearings are
utilized in modern rotating equipment solutions, in the oil, gas, power and transportation
industries worldwide. Additionally, journal bearings contribute to the longevity and
efficiency of rotational systems. Therefore, it is important to understand failure modes.
Currently, seed-of-fault testing is conducted on full scale equipment to gain an
understanding of the effects of damage and contamination on the degradation of journal
bearings. Data collected during the testing provides opportunities for fault detection and
system design improvements.
Rochester Institute of Technology (RIT), supported by Dresser-Rand, researches
performance and instrumentation of the ESH-1 reciprocating compressor. To conduct the
seed-of-fault testing a deliberately damaged journal bearing must be perpetually replaced
to identify various failure conditions. This replacement can take over six hours to
accomplish. To reduce compressor down time a series of capstone projects have been
commissioned by Jason R. Kolodziej, Assistant Professor of Mechanical Engineering at
RIT to design a stand-alone test rig capable of simulating compressor environments. The
alpha prototype accomplished a working journal bearing similarity test rig capable of
static loading conditions. The beta prototype, currently in development, will fit the alpha
prototype with dynamic loading capabilities using electromechanical actuators (EMA’s)
donated by Moog, to more accurately replicate compressor characteristics. The beta
prototype will also focus on implementation of sophisticated data acquisition equipment
for the study of film thickness, flow conditions, fluid properties, and vibrations.
Furthermore, the methodologies of design, testing and evaluation will be explored in this
paper.
Copyright © 2015 Rochester Institute of Technology
Proceedings of the Multidisciplinary Senior Design Conference
Page 2
TABLE OF CONTENTS
ABSTRACT...................................................................................................................................................................1
INTRODUCTION.........................................................................................................................................................3
PROCESS .....................................................................................................................................................................3
Needs & Specifications .................................................................................................................................................3
Concept Selection .........................................................................................................................................................4
Mounting Design...........................................................................................................................................................6
Vibration Analysis.........................................................................................................................................................6
Oil Flow Problem Solving.............................................................................................................................................8
RESULTS & DISCUSSION ........................................................................................................ ................................9
CONCLUSION & RECOMMENDATIONS ..............................................................................................................9
REFERENCES............................................................................................................................................................10
ACKNOWLEDGMENTS ............................................................................................................. .............................10
LIST OF TABLES AND FIGURES
Figure 1: Alpha Test Rig .............................................................................................................................................3
Figure 2: Pareto Analysis.............................................................................................................................................4
Figure 3: Functional Decomposition ...........................................................................................................................4
Figure 4: Morphological Chart ....................................................................................................................................5
Figure 5: Component Brainstorming............................................................................................................................5
Figure 6: Pugh Chart.....................................................................................................................................................5
Figure 7: Allowable Cycle Before Maintenance..........................................................................................................6
Figure 8: X-Direction Mounts......................................................................................................................................6
Figure 9: Y- Direction Mounts.....................................................................................................................................6
Figure 10: Sample of LabVIEW Interface...................................................................................................................7
Figure 11: Steps to Problem Solving...........................................................................................................................8
Figure 12: Final Progress of Given Components.........................................................................................................9
Figure 13: Load at 0.5 and 1 Hz.................................................................................................................................11
Figure 14: Load at 1 and 12 Hz...................................................................................................................................9
Table 1: Requirements.................................................................................................................................................4
Table 2: Frequency Relationships...............................................................................................................................7
Table 3: Excerpt from Table 4.89....................................................................................................................... ........7
Project P15453
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 3
INTRODUCTION
Dresser-Rand designs, manufactures, and services a wide variety of products for use in the
oil, gas, process, and power industries. In 2011, Dresser-Rand donated an ESH-1 reciprocating
compressor to RIT for graduate and undergraduate research. Currently, much of the research is
based in the areas of measurement, controls and extended life/fault testing. A journal bearing
test rig was designed and fabricated by an RIT Multidisciplinary Senior Design team during the
2013-2014 academic year to simulate Dresser-Rand’s ESH-1 reciprocating compressor. The
prototype became the alpha version of the test rig. This project serves as a continuation of the
2013-2014 project, improving the alpha version to the beta version.
The reciprocating compressor test rig allows for rapid
testing of the crank journal bearing found in the full scale
ESH-1 compressor. Comparatively, disassembling the full
scale compressor can take many hours while disassembling
the test rig can take approximately a sixth of the time. The
rapid disassembly allows for testing and monitoring of the
crank journal bearing while reducing time between tests. The
alpha version is fitted with static loading due to time and
budgetary constraints. However, this does not accurately
simulate the environment found in the compressor. The goal
of this project is to successfully adapt the alpha version with
dynamic loading while meeting bearing loads of
approximately 1900 lb. at 6 Hertz. Both the dynamic loading
components and the base test rig will be fitted with data
acquisition sensors for condition monitoring. The resulting
design will not only add both dynamic loading as well as data
acquisition without sacrificing ease of use or safety.
PROCESS
Needs and Specifications
In order to properly simulate the loading profile, there are many characteristics that
needed to be met. To ensure the project provided an efficient and usable solution, we mapped the
given customer requirements to engineering requirements. This identified a path to solution
while focusing efforts to the most important aspects of the problem. Furthermore, the
engineering requirements were measured against the customer requirements and judged for their
assistance in meeting the final goal. That is, properly simulating the load profile of a
reciprocating air compressor while not increasing the time required to change the bearing. The
raw scores provided for each requirement were then analyzed and organized using the pareto
analysis to prioritize requirements. Additionally, in order to document the testing and calibration
procedures, the team created standard operating procedures. These procedures provide step by
step instructions on how to perform the required calibration and testing procedures in simple
terms and with detailed pictures. Requirement examples can be seen below.
Copyright © 2015 Rochester Institute of Technology
Proceedings of the Multidisciplinary Senior Design Conference
Page 4
Table 1: Requirements
Figure 2: Pareto Analysis
Concept Selection
In efforts to find the most effective solution the team generated multiple concepts at each
level of design. Starting at the systems level, a functional decomposition was developed to lay
out all of the key functions the test rig. This allowed the team to focus on the subsystems that
ranked highest on the Pareto chart while understanding their connections to the other parts of the
system. The critical subsystems are shown in orange (to the right) in figure 2
Figure 3: Functional Decomposition
Further decomposing the subsystems into individual components followed similar
brainstorming process. Morphological charts were used to visually display possible options that
were then turned into potential unique solutions. This led to the identification of the benefits and
risks of each potential solution. Furthermore, minimizing risks and combining multiple ideas
lead to concept selection.
Project P15453
Proceedings of the Multi-Disciplinary Senior Design Conference
Figure 4: Morphological Chart
Page 5
Figure 5: Component Brainstorming
Concepts were then evaluated using a Pugh Chart, which is a comparative chart that displays
each concept based on categories specific to the engineering requirements and design
considerations. Furthermore, combining certain aspects of the concepts produced new concepts;
these combinations were added to the Pugh Chart and analyzed. This became an iterative process
allowing us to narrow our concepts down to four. One using piezoelectric actuators, the second
using hydraulic actuation, the third using pneumatic actuation, and the final using an
electromechanical actuator (EMA). Each concept was examined further until the EMA was
chosen due to its accuracy and cost. While piezo had excellent micro motion abilities it was
limited in its ability to displace large distances while also being very costly. Hydraulic was able
meet force and speed actuation but required many additional components while also being very
costly. Finally, Pneumatic actuation struggled to meet the required force without a large quantity
of air on demand.
Figure 6: Pugh Chart
The EMA design consists of three main components; the actuator, the power supply and
the controller. EMA actuation is controlled with a displacement feedback loop from the pitch of
a ball screw and its rotational position. In order to control the actuator using force feedback a
secondary control loop must be created to adjust position based on applied and required force to
adequately replicate the desired load profile. Another design challenge is false brinelling, this
occurs from lubrication being pushed out of a loaded region during small oscillatory movements
Copyright © 2015 Rochester Institute of Technology
Proceedings of the Multidisciplinary Senior Design Conference
Page 6
that occur with micro displacements. One way to mitigate this is to perform periodic macro
displacements with one or more full screw revolutions to to re-distribute lubrication within the
rolling elements. To ensure testing won’t exceed manufacturer's recommended limits of microcycles the following calculation was completed.
Figure 7: Allowable Cycles before Maintenance
Mounting Design
When deciding how to mount the EMA’s various design aspects were taken into consideration.
These include both budget and structural features. Additionally to simulate a sinusoidal loading
profile actuation has to occur in two directions x any y. Parallel to the test rig surface was
determined to be the x-direction while parallel to the test rig legs is the y-direction. Mounting in
the x-direction simply required holding the EMA to the table while placing the shaft in the center
of the bearing housing. The specifics of the x-direction mounting can be seen below in Figure 8.
However, mounting in the y-direction was more difficult because the EMA has to sit
perpendicular to the surface. Thus the EMA had to either be mounted above the bearing block or
hang from below the table surface. For simplicity, mounting in the y-direction follows the same
design as mounting in the x. The specifics of the x-direction mounting can be seen below in
Figure 9.
Finally, both mounting
configurations use a Moog
front mount bracket that
supports all of the EMA’s
weight. Furthermore, the
ultimate considerations for
the mounting design were
stress
and
deflection.
Because we were actuating
at such low displacements
the deflections had to be
small enough to overcome
with our control. Lastly, the
total stress had to be low
enough to ensure the endurance limit of our components
allowed for infinite life of steel.
Vibration Analysis
To analyze the vibrations of our test rig, the focus was directed towards the components which
created a frequency: the motor and the shaft. While the shaft is known to spin at 360rpm (6Hz)
the motor drive rpm was unknown. Using simple pulley ratios, the rpm and frequency of the
motor drive was determined to be 1285.7rpm and 21.43Hz respectfully. With the frequencies of
both major components on the table, Leissa’s vibrations of plates analysis was then performed to
Project P15453
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 7
determine the vibration of the plate itself, using the ”Simply Supported” scenario at the four
corners of the plate as the worst-case scenario [4]. Taken
from NASA’s technical report on leissa vibration of
plates, table 4.89 is used to find ƛ, which gives values of
the desired wavelengths determined by the length ratio.
Frequency was found using the following relationship and
input parameters
Because our ratio is between two of the parameters found
in Table 4.89, we know the frequencies found with these
two values will be the limits of our actual vibration.
Typically, the frequencies will be considered safe to avoid
resonance between them if they are outside their values
multiplied by 4. The actual value will come out to
approximately 100Hz, but cannot be directly interpolated
since the relationship of the data isn’t linear.
Data Acquisition
Although the primary
focus of this project is the ability
of dynamic actuation, data
acquisition capabilities account
for many of the secondary
objectives. In terms of condition
monitoring and seed of fault
research, the ability to obtain
accurate data at the time of
failure is crucial. Knowing the
applied load to the system, is
important to both the actuation
system, to provide feedback to
the EMAs, and to the data
acquisition
system,
to
understand at what load the
bearing
fails.
Transducer
Techniques load cells capable of
2000 lbs. of force will be implemented between the EMAs and the bearing housing. The load
cells work with strain gauges that output millivolt signals corresponding to the experienced load.
These signals are then interpreted by signal conditioners that convert it to a zero to ten volt scale
that is then interpreted using National Instruments’ LabVIEW programming methods.
Copyright © 2015 Rochester Institute of Technology
Proceedings of the Multidisciplinary Senior Design Conference
Page 8
We are using load cells and signal conditioners donated by Bill Nowak, for the first
iteration, which have limited sampling capabilities of 60 samples a second. Testing will
determine if this is sufficient. Determining the position of journal in respect to the journal
bearing allows the systems eccentricity and film thickness to be determined. The diametric
clearance of 90 microns represents the total displacement the system can undergo, therefore
linear variable differential transformers (LVDT) were chosen due to their ability to measure such
small displacements. Similar to the load cell, the signals are converted using a signal conditioner
and processed using LabVIEW. The VI also interprets data from an Omega pressure transducer
located at the oil inlet of the bearing housing along with vibration, angular position and
temperature data. The vibrations of the system are analyzed by a Kistler accelerometer and
National Instruments (NI) usb-4431 capable of 102.4kS/s. A Photocraft quadrature encoder is
used to analyze the angular position which is sent to the actuation controller to apply the proper
load profile and to the data acquisition system to give rotational velocity (RPM) data. Finally the
Omega thermocouples (TC) acquire temperature data from the journal bearing, oil outlet and oil
reservoir and are imported to VI using an NI TC module.
Oil Flow Problem Solving
As part of our customer requirements we were tasked
with acquiring the oil flow rate through the journal
bearing. The system that was used in the alpha
prototype was unsuccessful due to significantly smaller
flow rates than anticipated. To understand the cause of
this reduced flow we utilized our problem solving
process. First we identified that the issue was
perpetuating from the bearing housing by checking oil
flow at each major component. We then analyzed the oil
flow path through the bearing housing and identified
potential restriction. With the assistance of Dr.
Kolodziej we developed a theoretical model for the oil
flow, Qp, under no load and no rotation using the
equation given by Martin and Lee [6].
Where D is the bearing diameter, C is the radial clearance between the journal and
bearing, Pf is the groove supply pressure, is the dynamic viscosity, is the journal eccentricity
ratio, L is the overall bearing length, and a is the groove width. From this equation we were able
to identify the parameters with the most influence and check our assumed values with
experimentation. Such as measuring the pressure drop induced by the journal bearing’s orifice
like feed hole which was 20 psi. The measured radial clearance between the journal and journal
bearing was about 30 percent smaller than specified by the alpha team’s design. After the
analyses the theoretical flow rate was 0.0022 gpm when the journal eccentricity ratio is 1 and a
value of 0.0009 gpm when the journal eccentricity ratio is 0. Thus, our measured value of 0.0016
gpm correlates well. From these findings, we are able to verify the low flow rate as adequate and
that the alpha prototypes flow meter was oversized. Next we plan to test flow rates with journal
rotation and dynamic loading both analytically and experimentally.
Project P15453
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 9
RESULTS AND DISCUSSION
Figure 12: Final Progress of Given Requirements
Overall the team successfully fitted the alpha test rig with dynamic loading capabilities
and most action items were completed. Figure 11 above details the progress of each requirement.
From senior design one, the loading system was completed with electromechanical actuators.
From senior design two, a flow analysis was completed to better understand and diagnose the
lubrication system, a vibration analysis was completed, a seal removal tool was created to reduce
bearing replacement time and finally, a LabVIEW interface was created for bearing environment
monitoring. The LabVIEW interface works with placed sensors. After preliminary data
collection, the test rig will be ready for seed of fault testing and bearing analysis at reduced speed
and frequency of original goal, that is the loading profile of an ESH-1 reciprocating compressor.
Figure 12: Load at 0.5 and 1 Hz Figure 13: Load at 1 and 12 Hz
CONCLUSIONS AND RECOMMENDATIONS
First, the beta prototype is now fit to apply dynamic loading to the journal bearing.
Currently, the speed of which the actuation can be applied is restricted by software and backlash
in the rod end joint and journal bearing clearance. Our team recommends the implementation of
a machine vice style pre-load system to eliminate the backlash. More details on this concept has
can be found on the teams EDGE website.
Copyright © 2015 Rochester Institute of Technology
Proceedings of the Multidisciplinary Senior Design Conference
Page 10
Second, data acquisition for loading is limited by the speed of the load cell meters.
Currently, the load cell setup can acquire at 60 Samples/sec. The system is also setup to record
the load calculation of the EMA at 2.6 kHz. Our team recommends exploring more capable load
cell meters or task a team with development of their own system.
Next, the alpha prototype oil system was found to be far over-sized for the amount of
flow actually passing through the bearing housing. To properly measure this flow rate, which is
on the order of drops per minute, our team recommends an investigation or project based around
ultra-sonic flow sensors. The systems oil pump elicits more audible noise the more it’s used and
the oil pump should be replaced in the near future. The system would also benefit from a fully
enclosed
oil
reservoir
with
a
level
sensor.
Finally, vibrations from the oil pump were reduced over fifty percent. However, there is
still room for improvements. The interface between the table top and frame could be
reconfigured to reduce vibrations. Our team would recommend adding noise cancellation
components to this interface to continue the noise reduction of the oil pump.
REFERENCES
[1] Budynas, Richard G., J. Keith. Nisbett, and Joseph Edward. Shigley. Shigley's Mechanical
Engineering Design
[2] Fox, Robert W., Alan T. McDonald, and Philip J. Pritchard. Introduction to “Fluid
Mechanics”. Hoboken, NJ: Wiley, 2008. Print.
[3] Holzenkamp, Markus. “Modeling and Condition Monitoring of Fully Floating Reciprocating
Compressor Main Bearings Using Data Driven Classification.” Rochester Institute of
Technology, 2013.
[4] Leissa, A. W., 1969, “Vibration of Plates,” NASA Report SP-160
[5] Manring, Noah. “Hydraulic Control Systems”. Hoboken, NJ: John Wiley, 2005. Print.
[6] Martin, F. A., and Lee, C.S., 1983, “Feed-Pressure Flow in Plain Journal Bearings,” ASLE
Transactions, 26, pp. 381-392.
[7] Palm, William J. System Dynamics. Boston, MA: McGraw-Hill, 2010. Print.
[8] Parker Hannifin Corporation. Parker Pneumatic. Cataloug PDE2600PNUK, 2013. Print.
[9] "Piezo Nano Positioning." Physik Instrumente (PI) Gmb H and Co. KG, Web.
[10] Rippel, Harry C.,1960. “Cast Bronze Bearing Design Manual,” Cast Bronze Bearing
Institute Inc.
ACKNOWLEDGMENTS
Team P15453 would like to express our most sincere gratitude to Dr. Jason Kolodziej, William
Nowak, Joe Dyer, Jim Kowalski, Scott Delmonte, Steven Luchessi, and Dr. Stephen Boedo for
their guidance throughout the design and build phase.
We would also like to acknowledge the RIT Machine shop staff: Robert Kraynik, Jan Maneti,
Dave Hathaway and Ryan Crittenden for their continuous assistance and support.
Project P15453
`