Functional analysis and modelling of a dynamic seal
71
Functional analysis and modelling of a dynamic seal component for a reciprocating gas compressor CHANDRAMOULI SURYANARAYANAN Master of Science Thesis TRITA-ITM-EX 2018:685 KTH Industrial Engineering and Management Machine Design SE-100 44 STOCKHOLM
Functional analysis and modelling of a dynamic seal
Functional analysis and modelling of a dynamic seal component for a
reciprocating gas
compressor
Master of Science Thesis TRITA-ITM-EX 2018:685 KTH Industrial
Engineering and Management
Machine Design SE-100 44 STOCKHOLM
i
Master of Science Thesis TRITA-ITM-EX 2018:685
Functional analysis and modelling of a dynamic seal component for a
reciprocating gas compressor
Chandramouli Suryanarayanan
Approved 2018-10-03
Contact person Andreas Söderberg
Abstract The rod packing seal in a reciprocating compressor plays a
vital role to reduce the leakage of highly pressurized gas along
the piston rod. Reciprocating compressor in the natural gas
transmission and process industry is one of the crucial
applications where the Nextseal technology can be put to greater
use for the reduction of green-house gas emissions. Nextseal
technology revolves around the notion of pressure balancing of
seals or rings.
The purpose of this master thesis is to understand how the design
of the novel rod-packing seals/rings behaves in correlation with
the concept of pressure balancing. The work also includes
developing a model, both CFD and analytical to obtain a
relationship between fluid (viscous oil) pressure and displacement
of the dynamic component. Functional analysis of the dynamic system
is conducted by numerical simulation using MATLAB. Major parameters
influencing the dynamic behaviour are identified at the
beginning.
The scope for CFD model is defined and the developed method is used
to obtain correlation between hydraulic fluid pressure and
displacement of the dynamic component. A derived analytical model
is solved, and the results are compared and validated. The
validated correlation is employed to solve the dynamic system
numerically and the results are analyzed. From this numerical
method the effect of friction force and geometry of the dynamic
component on pressure difference and displacement of the dynamic
component is well analyzed and discussed in this thesis. From the
influence of friction force on pressure difference study, a linear
relation is observed. Also, by changing the geometry (chamfer
length and angle of the dynamic component) of the dynamic
component, it can be observed that design configuration with 60°
chamfer angle gives smaller pressure difference value compared to
the original design. Thus, the model developed can be used to
obtain results for pressure difference, displacement, and to study
the effects of friction, geometry and mass.
Keywords: Rod packing dynamic seals, Computational Fluid Dynamics,
Pressure balance, Numerical analysis
iii
Chandramouli Suryanarayanan
Godkänt 2018-10-03
Sammanfattning Stångförpackningstätningen i en fram- och återgående
kompressor spelar en viktig roll för att minska läckaget av gas
under högt tryck längs kolvstången. Kolvkompressorer är viktiga
komponeter för naturgasöverföring- och inom processindustrin.
Teknik utvecklad av Nextseal kan potentiellt användas för att
tryckbalansera tätningar eller ringar och därmed minska utsläpen
växthusgas.
Syftet med detta examensarbete är att skapa kunskap kring hur
utformningen av de nya stavförpackningstätningarna/ringarna
påverkar tryckbalanseringen. Arbetet innefattar också
modellutveckling, både av CFD-modeller och analytiska modeller för
att analysera relationen mellan vätsketryck (viskös olja) och
förskjutning av den dynamiska komponenten. Funktionsanalys av det
dynamiska systemet utförs med numeriska simuleringar och
experiment. Viktiga parametrar som påverkar det dynamiska beteendet
identifieras i början.
Den utvecklade metoden och CFD-modellen används för att studera
korrelationen mellan hydraulvätsketryck och förskjutning av den
dynamiska komponenten. En härledd analytisk modell löses numeriskt
och resultaten jämförs och valideras. Den validerade modellen
används för att simulera det dynamiska systemet numeriskt och
resultaten analyseras. Effekten av friktionskraft och geometri hos
den dynamiska komponenten på tryckskillnad och förskjutning av den
dynamiska komponenten analyseras och diskuteras i avhandlingen. Det
visas att friktionskraften har en linjär effekt på
tryckdifferensen.
Genom att ändra geometrin (fasens längd och vinkeln på den
dynamiska komponenten) observeras att konstruktionskonfigurationen
med 60° fasvinkel ger en lägre tryckskillnad jämfört med den
ursprungliga konstruktionen. Således kan den utvecklade modellen
användas för att studera relationerna mellan tryckskillnad och
förskjutning och för att studera effekterna av friktion, geometri
och tröghetsmassa.
Nyckelord: dynamiska tätningar, CFD, tryckbalans, numerisk
analys
v
Foreword
This thesis work was carried out over a period of eight months at
Nextseal AB, Stockholm and KTH Royal Institute of Technology at the
department of Machine Design. It has been a great experience and
immense pleasure working with a novel technology and I would like
to thank Nextseal AB and KTH, both the organization for providing
various supports throughout my thesis work.
I would first like to thank my thesis supervisor Andreas Söderberg,
CEO of Nextseal AB for providing me this opportunity to work with
an interesting and challenging technology and also provided all the
support to keep going forward with my thesis work. I would also
like to thank Bengt Adolfsson, board member and senior consulting
member of the firm, for providing valuable inputs for many problems
that I faced with my experimental test setup. I would like to thank
Tomas Östberg. He was very helpful with manufacturing of different
parts required for the test-rig and also gave valuable suggestions
related to design for manufacturing. I cannot thank them enough. I
wish to express my sincere gratitude to my academic supervisor
Stefan Björklund, who has always been available whenever I faced
technical problems and has provided with valuable perspectives and
solutions. He has continuously guided me to the right track. I
would also like to thank Ulf Sellgren, my examiner at KTH, for
providing timely support with the thesis and giving his insights
whenever needed.
Finally, I would like to convey my sincere regards to my parents,
sister and friends for continuous motivation and support throughout
my years of education and through the process of writing this
thesis.
Chandramouli Suryanarayanan Stockholm, Sweden
2-D Two Dimensional
FE Finite Element
PTFE Polytetrafluoroethylene
Abstract i
Sammanfattning iii
Foreword v
Nomenclature vii
1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 1 1.2 Purpose . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4
Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Frame of Reference 7 2.1 Rod Packing Technology . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 7 2.1.1 Packing Cups . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.1.2 Packing
rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . 8 2.1.3 Flange . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 9 2.2 Packing ring Designs . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 9 2.2.1 Single Acting Rings . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.2 Double
Acting Rings . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . 11 2.2.3 Oil Wiper Rings . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . 12 2.3 Viscous Fluid Flows . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.1 Flow Regimes . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . 14 2.3.2 Flow through
channels . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 15 2.4 Governing Equations . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 16 2.4.1 Navier-Stokes Equation . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . 16 2.4.2
Hagen-Poiseuille flow . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . 17 2.5 Analytical Background . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 18
3 Methodology 19 3.1 Parameter Identification . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Experimental Testing . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . 21 3.2.1 Test-Rig .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . 21 3.2.2 Design and Construction . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 23 3.2.3 Test Objective . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . 24 3.2.4
Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . 25 3.2.5 Data Acquisition . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . 26 3.2.6 Drives and Control . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . 27
ix
3.3 Computational Analysis of the Dynamic System . . . . . . . . .
. . . . . . . . . . . . . . . 28 3.3.1 Scope and Method Development
for CFD . . . . . . . . . . . . . . . . . . . . . . . . 28 3.3.2
Fluid Model . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 30 3.3.3 Meshing . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . 30 3.3.4 Setting up Physics . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 31 3.3.5 Solving and Post-Processing. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . 32 3.3.6 Analytical model .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . 33 3.3.7 Geometry Design configurations . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4 Results
and Discussions 37 4.1 CFD Simulation of liquid domain . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1.1 Validation with analytical model . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . 38 4.2 Dynamic Response analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . 39 4.2.1 Time-Domain Analysis . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . 39 4.2.2
Performance analysis . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . 41 4.3 Experimental Result . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 45
5 Conclusions 47
Introduction
The second largest source of energy in the United States is natural
gas. The gas from both onshore and offshore sources, after refining
process, is transported to supply households and industries all
over the country where it is consumed for heat and electricity
generation. The United States natural gas industry has undergone
change of high magnitude and pace for the past decade. Natural gas
production in US increased 33 percent between 2005 and 2013 [1].
Gas demand for power generation has grown from 15.8 billion cubic
feet per day (Bcf/d) in 2005 to 22.2 Bcf/d in 2013 [1]. Natural gas
pipeline network has been developed to transport natural gas, both
interstate and intrastate, from production and storage areas to
distribution systems and end users. As of 2007 there are 210
pipeline systems with 305,000 miles of high- pressure transmission
[2]. Additionally, 4000 miles of interstate transmission pipelines
are constructed from 2008-2013. The transmission network embraces
more than 1,400 compressor stations to perpetuate high pressure in
the pipelines. Methane is a potent green- house gas. Even with a
short atmospheric lifetime of 10-12 years, methane is considered to
be 50 times more effective than C2 at trapping heat in the
atmosphere. This chapter covers essential background of this
project, project scope, formulation of research question,
methodology pursued and delimitations.
1.1 Background
The mitigation of methane gas emission is of major concern, as the
Environmental Protection Agency (EPA) has stringent policies [3].
There are many ongoing researches on various compressor components
for better performance by various compressor OEM’s and several
other contributing players in oil and gas industry. Nextseal AB, a
Swedish based company has an interesting solution for rod packing
seals that could revolutionize this issue.
Reciprocating compressors are commonly and widely used compressors
to build pressure to transport gas in a pipe. For transporting
natural gas, there is a compressor station for every 100
kilometers. There are leakages from pipelines, joints and various
compressor components, like valve, rod seals, and other fittings.
From Transportation, storage and distribution, EPA has stated that
most methane emission is from reciprocating compressors that
account for over 1 to 6 standard cubic meters per hour (scm/h) [4]
for large and high- pressure compressors. The largest contribution
is from rod packing rings/seals in the compressor.
Current technology as shown in figure 1.1, uses sets of
specifically-cut, dry-ring seals held in place with springs and
cups. There is a trade-off between leakage reduction and friction
in today’s technology. In the reciprocating compressors as piston
moves back and forth, the pressure differential across the packing
rings/seals creates a twisting effect that causes natural gas to
leak into the casing. Ring/seal twisting also causes increased
friction and wear to the sealing rings and piston rod. The gas
leaking from rod packing case is vented to atmosphere.
Figure 1.1 Reciprocating compressor with existing rod packing seal
technology Source: Image from U.S EPA 2006a
Nextseal technology revolves around the notion of pressure
balancing of seals. A schematic representation of pressure
balancing concept is shown in figure 1.2 below. It takes the
concept of liquid sealing and combines it with a novel, patented
arrangement for pressure balancing across a seal arrangement
(Patent No: US 7,757,599 B2 [5]). The pressure balancing has two
appealing characteristics. First, it stops the gas from leaking by
letting the seal act as a separator of gas and liquid with the same
pressure. Second, it reduces the friction caused by the seal
running against the opposing surface. This results in both less
leakage and less friction which will be a huge advantage over
current solution for dynamic seals. This has several applications
like waste heat recovery system- steam expanders, compressors in
food industry, reciprocating compressors for gas transmission and
so on.
Figure 1.2 Schematic representation of pressure balancing concept
Source: Nextseal
3
1.2 Purpose The main purpose of the thesis is to understand how the
design of the rod-packing seals/rings behaves in correlation with
the proof of concept (pressure balancing) through experimental and
numerical analysis. Hereafter, rod-packing seals/rings will be
referred as dynamic component in this report. Behaviour of the
dynamic component means displacement and velocity based on the
pressure acting on both sides of the dynamic component and how both
the pressure curves behave. The test-rig almost reproduces the
operating conditions of the real time application (reciprocating
compressors of natural gas transmission). The objective also
includes studying the effects of friction force and geometry
changes on the dynamic response of the system.
Tools used to achieve the results are ANSYS Fluent and MATLAB. CFD
tool is used to develop the relationship between fluid pressure and
displacement of the dynamic component, which then can be used to
analyze the dynamics and functioning of the system. Defining the
scope of the thesis led to the formulation of following three
research questions.
1. Can the behaviour of the dynamic component be understood through
numerical analysis?
2. What is the relation between friction force and pressure
difference over the dynamic component?
3. Does the geometry of the dynamic component influence the
functioning of the system?
1.3 Delimitations
• Analysis of the sealing ring design will not be of focus as
standard packing rings from leading market will be used.
• Product design influencing production cost is not included in
this project work.
• Tests are undertaken for one configuration of fluid flow channel
and design of dynamic component
• Parameters like dynamic component geometry, friction force that
have an effect on the system functioning will be studied using the
analytical model but does not include the effects of component
mass, pipe geometry and flow rate.
1.4 Methodology The methodology acquired to achieve the above
mentioned objectives are provided in the form of flow-chart as
shown in figure 1.3 below. A short description on each context is
published in this sub-topic.
Figure 1.3: Methodology flowchart
Literature study -
A comprehensive literature study is conducted to understand the
basics of fluid dynamics & mechanics, mathematical equations,
fluid-structure interactions, planning tests and how to implement
CFD tool. It also includes learning about the existing rod packing
seal technology in the market and the area of research being
conducted in this field. Manuals of control drives (Donfoss
frequency converter, Bosch Rexroth drive) are also learned as an
outcome to set-up and operate them effectively.
Defining the Scope -
The scopes as defined in section 1.2 are vital, as they should be
developed based on time constraint and knowledge limitations.
Accordingly, research questions are formulated.
Identify system parameters -
The behaviour of the dynamic component will be studied and is
affected by various parameters.
5
• Flow rate of hydraulic fluid • Mass of the dynamic component •
Friction acting on the dynamic component • Speed of the piston rod
• Tube length of hydraulic fluid flow • Geometry of the dynamic
component
Modify test-rig –
The test rig contains a motor, a modified Stirling single cylinder
marine engine, test frame, a lubrication oil pump, and a
Bosch-Rexroth hydraulic pump. A double-acting piston cylinder is
designed and manufactured for testing. Additional components like
control-volume adjuster, adapters, and cooling jackets are designed
and manufactured. The testing reproduces the operating conditions
as that of in a real time application.
Set-up control drives and instrumentation –
To set up control drives is an integral part of testing for remote
access to operate the drives (hydraulic pump, motor). Setting
appropriate parameter values for example, motor speed, pump
flow-rate to the drives using the designated software is essential.
The test-rig is instrumented at the desired location and data are
collected using the data acquisition system.
Dynamic testing –
• The dynamic component is tested with the modified test rig •
Assembling the double acting piston cylinder with appropriate
instrumentation and
attachments • Setting up instrumentation and drives for remote
access • Conducting tests and recording the outcomes
Simplified FE model and CFD analysis –
The dynamic component to be tested experimentally can be studied
and analysed using any of the CFD tool commercially available. In
this project ANSYS Fluent, is utilized for computational fluid
dynamics. The fluid model of the dynamic system is developed using
SolidEdge and is imported to ANSYS workbench. Meshing of an
axisymmetric 2-D fluid model is performed with ANSYS mesh modeler.
Pre and post processing are conducted with Fluent.
Analytical model –
An analytical model that represents the CFD model is formulated and
its results are validated. The equation of motion which represents
the dynamic system of the model (dynamic component) is derived.
MATLAB numerical solver for ordinary differential equations is used
to solve the differential equation. Result from validated
analytical/CFD model will be used to solve the mathematical
model.
Analyse the results –
• Validating the results obtained from CFD model with the
analytical model. • Analyzing the dynamics and functioning of the
component with computational
methods. • Varying the system parameters and studying its effect on
the dynamic behaviour of the
model.
Conclusion and future work –
A brief discussion on various findings from this project will be
reported. It will also describe the validity of the CFD and
analytical model. Proposing possible improvements as future works
based on the results obtained.
7
Chapter 2
Frame of reference This chapter delineates the necessary details
about the existing technology for basic understanding about rod
packing rings and the background of technical knowledge obtained.
The chapter provides information about the rod packing technology
and ring design and continues with technical description on fluid
dynamics. The governing equation behind numerical fluid dynamic
analysis, namely Navier-Stokes equation for fluid motion is
discussed.
2.1 Rod packing technology In reciprocating process gas
compressors, one of the most critical technologies is rod pack
seals. Based on the application, and the operating condition of the
reciprocating compressors, the design and size of rod packing seals
varies accordingly. When compressing gas at varying pressure
(suction pressure to discharge pressure) using a reciprocating type
piston compressor, there will be leakage along the piston rod due
to the clearance between the cylinder and the rod [6]. The main
function of the rod seals is to reduce the leakage rate. Rod seals
are also called mechanical packings and their nomenclature is shown
in figure 2.1. Mechanical packings comprise of three major
components.
1. Packing case / cups 2. Packing rings 3. flange
Figure 2.1 Nomenclature of mechanical packing, Source: Image taken
from CPI- Mechanical packing ‘design and theory of operation’
[6].
The design of mechanical packings depends on the following
parameters [7]
• Lubricated or non-lubricated system • Operating pressure range •
Environmental standards • Nature of gas being compressed • Rod
size, stroke length and speed • Bolt load requirements • Project
cost limits and timing restraints
The design should meet the requirements described in the API 618
standards for reciprocating compressors.
2.1.1 Packing cups
The packing cups which retain the packings rings are assembled back
to back closing the rings in a confined space. Henceforth they are
also referred as retainers [6]. The cups are machined to provide
certain amount of radial clearance around the piston rod. This is
to corroborate that the piston rod will have no contact with the
packing retainers; if any lateral movement of the rod occurs when
running. The radial clearance value is influenced by the rod
diameter, operating pressure and the type of compressor. The
packing ring set is held between the cups and some side clearance
in the cup allocates for radial movement inside the cup. Since, the
sealing rings can float around within the cup, this type of
mechanical packing is termed as ‘floating’.
Packing cup faces are ground and lapped depending on the pressure
and gas being used. Manufacturing of the component needs to be
precise as per the drawings and quality check insures that faces
are both flat and parallel such that it is assembled properly, and
surfaces of the cup are perpendicular to the rod. They are
precision machined with pertinent grade of material so that it
meets the performance requirements [7]. Commonly used materials are
carbon steel, alloy steel, stainless steel, cast iron, or
bronze.
2.1.2 Packing rings
Pressure packing rings serves as crucial component in reciprocating
compressor, which offers a dynamic, mechanical seal around piston
rod and against the sealing surface of the packing cups to prevent
leakage from the cylinder. The rings which are widely used till now
are the design patented in late nineteenth century by A.W.France
[8].
Different types of cut rings are assembled and held together within
a cup which constitute a ring set. The cut rings are held together
with a garter spring and is referred as segmental packing ring.
Similarly, different ring sets are used in a rod packing technology
based on the application. In section 2.2, a detailed discussion
about different ring designs and existing ring pairs is
presented.
9
Figure 2.2 Pressure packing rings Source: Image taken from
Compressor Products International – “CPI Packing Rings How to
Install”
The common features of pressure packing rings are as follows
[7]:
• Pressure loads the rings around the rod and against the sealing
surface of the packing cup
• The compressed gas fills the gap between the cup and the rings
and leaves as the rod reciprocates. In this gap (radial clearance)
rings can float.
• Gas leakage is blocked by ring overlap. • End gaps allow the
rings to self-adjust for wear which extends its life-time.
2.1.3 Flange
The flange is the end part of the mechanical rod packing. The
packing cups, ring sets and flange are usually connected via
threaded rods commonly called as tie rods as shown in figure 2.1.
The packing rings assembly is bolted with the compressor cylinder
through the flange. A wide range of flange sizes and number of
bolts are available, based on the application, and operating
conditions. Lube and vent connections are machined on the flange.
Lube connection is furnished based on lubricated / non-lubricated
type of packing rings.
2.2 Packing ring designs This section introduces common types of
rod packing ring and ring sets. Different compressor duties require
unique packing ring designs and suitable material of the rings.
Other factors like gas properties, operating pressure, and
compressor speed play a vital role in determining the appropriate
combination of ring style [7].
Single acting rings, double acting rings and oil wiper rings are
variety of segmented ring types and ring pairs which will be
elucidated in the following sub-sections.
2.2.1 Single acting rings
As the name implies, single acting ring seals gas on one-side only.
These rings prevent pressurized gas from being enclosed between the
ring sets, during the suction stroke of the reciprocating
compressor. Most common types in the market are enumerated as
follows and refer figure 2.3 shown below.
Pressure breakers
Pressure breakers are usually installed in the first cup of the
packing assembly. On one side of the pressure breaker is etched
with a letter and that side faces high pressure. It throttles the
gas pressure pulsations and do not seal them off. Hence it should
be placed in the high-pressure side. It is typically required for
applications where the pressure difference between the suction and
discharge is over 20 bar [7].
Radial Tangent pair
Radial Tangent pair is a pair of 2 differently cut ring types. One
is radially cut and is called radial ring and the other one is
tangentially cut and hence tangential ring. These two are paired
together and it provides sealing along the piston rod and against
the sealing face of the next packing cup. It is a fundamental
sealing element in mechanical rod packing. These two rings are
doweled together, where the radial cut ring faces pressure which is
followed by the tangential ring. A dowel pin in the tangential ring
prevents rotation of one ring with respect to the other. It is
manufactured for a wide variety of materials based on
application.
Balanced-cap rings
Balanced cap rings are a four segmented radial cut ring held
together around the rod with a spring. It comprises two caps (upper
& lower) which bridges 2 side segments. This design produces a
balanced pressure breakdown that does not diminish sealing
capacity. The compact design provides advantages such as less
frictional heat, easy to install and low leakage. It can be used
for applications with high temperature, high load and can also be
used for a wide range of piston speeds.
Backup rings
The main task of backup rings is to prevent the extrusion of
non-metallic sealing rings into the clearance between the piston
rod and packing cup. The bore diameter of the backup ring is
slightly bigger than the sealing rings. Hence it does not seal on
the rod. It is combined usually with a radial tangent pair. Then
the backup ring will seal against the packing cup face. They are
quintessentially manufactured using cast iron or bronze
material.
11
Figure 2.3 Single acting rod packing rings Source: Image taken from
Hoerbiger, “Ring and packing”, Sealing systems for reciprocating
compressors [7]
Side loaded rings
Side loaded rings comprises of three ring types, which are placed
one behind the other, as mentioned below following the same order
(from the pressure side)
1. Radial cut ring with chamfered recess 2. Radial cut ring with
chamfered boss 3. Tangent cut ring
It provides effective sealing when pressure is below 4 bars. Hence,
it is usually located in the low-pressure end of packing or behind
the vent line.
2.2.2 Double acting rings
Double acting rings are commonly used at low pressures and as vent
seals. They seal gas in both directions. Two types of double acting
rings as shown in figure 2.4 are described below:
• Tangent pair • Double side loaded pressure rings
Figure 2.4 Double acting rod pressure packing rings Source: Image
taken from Hoerbiger, “Ring and packing”,
Sealing systems for reciprocating compressors [7]
Tangent pair
Tangent pair is a common double acting ring configuration. The ring
set incorporates 2 tangent rings pinned together so that the gaps
do not align. Both the rings seal against the metallic cup faces
and with the piston rod.
Double side loaded pressure rings
Double side loaded pressure ring sets uses two pairs of side loaded
pressure rings with the tangent cut rings faces the low pressure.
It is usually used in compressors with purge system. Purge gas
usually used is nitrogen due to its inert nature.
2.2.3 Oil wiper rings
As the name suggest, oil wiper rings scrape off the oil from the
piston rod, so that oil does not travel along the piston rod
further away from the mechanical packing case. These wiper ring(s)
as shown in figure 2.5 are held within oil scraper cup, which is
similar to other packing cups. Oil scrapper cup with rings are
usually placed after the flange. The oil wiper cup is either
separated from the mechanical packing case or attached with the
flange at the end. It is used when a lubricated type mechanical
packing seals are installed in the compressor.
Tangent cut wiper
Tangent cut wipers have two scraping edges on the inner diameter of
the ring, drilled internally with drain holes in the radially
outward direction. Tangent cut wiper ring comprises of three
segments, where the cut on each segment is tangential to the piston
rod and these segments are held together with a spring. The front
face of the ring is machined with drainage slots. The scrapped off
oil along the piston rod is channeled radially outwards through
these
13
drainage slots. For double-acting rings, drainage slots are
machined on both faces, but holes only on the front face
Figure 2.5 Oil Wiper rings Source: Image taken from Hoerbiger,
“Ring and packing”, Sealing systems for
reciprocating compressors [7]
Radial High-volume wiper (HVOL)
Radial High-volume wiper constitutes two radial cut rings with a
drainage hole milled on one face of the ring. The two rings are
joined with each other through a dowel pin. It helps to suppress
the rotation of rings and eradicate the issue of gaps from
aligning. Rings are manufactured with variety of metallic and
non-metallic materials, the most common is bronze.
Radial wiper rings set
The set of radial wiper rings provides effective oil wiping action.
Usually it includes 3 radial wiper rings that are attached behind
on another with dowel pin. The construction of the rings is almost
similar to that of tangential wiper rings, except the fact that the
cut is radial to the rod. In these rings, channelization of the
scrapped oil is through the drainage slot. However, they do not
contain oil drainage holes on the face of the ring. It is usually
made of cast iron or bronze.
2.3 Viscous fluid flows It is indispensable to have background
knowledge about fluid mechanics and dynamics before solving the
problem description as stated in section 1.2, of chapter 1. Viscous
flow of fluid [9] is an important topic. First, it is pre-requisite
to understand fluid flow regimes, flow through channels and losses
indulged, to design piping line connections for the test-rig to
administer the experiments. The thesis work comprises the use of
both compressible and incompressible fluids, but the methodology
acquires the comprehensive understanding of incompressible viscous
fluid flow to solve the dynamic system using numerical and
experimental methods.
2.3.1 Flow regimes
The flow of fluid which can be an internal or external flow through
a circular or non-circular channel, it is important to note that
flow can be either laminar or turbulent in nature. The intermediate
regime between these two flows is called transition. Now let’s
review about various flow regimes in brief.
When fluid flows in a medium and the layers of fluid film are
parallel to each other; then it is known as streamline / laminar
flow. In other words, there is no disruption or lateral mixing
(mixing at right angles to the flow direction) between the layers
during the flow. In laminar flow, when examined microscopically,
the fluid particles move in orderly path in straight lines parallel
to the pipe walls. Lateral mixing is attributable to the action of
diffusion of fluid layers. In general for laminar flow, diffusive
mixing is slow. However, if diameter/ perimeter of the channel is
small, then the diffusive mixing can be significant.
The turbulent flow regime is signalized by swift property changes.
It means there will be rapid variation of velocity and pressure in
space and time for turbulent flow. In contradiction to laminar
flow, the flow is highly susceptible to diffusive mixing. It is
hard to measure the mean velocity or pressure, since it requires
highly sensitive instruments which complicates the process.
Hot-wire anemometer or a piezoelectric pressure transducer can be
used to measure the turbulence.
When the smooth and steady flow terminates, and becomes fluctuating
and agitated, then the changeover phase is called transition.
Transition depends on many effects [9], for example, wall
roughness, fluctuations in the inlet stream or Reynolds number.The
Reynolds number is the primary factor influencing the flow regimes.
Hence, it is important to know the Reynolds number.
Reynolds number
In 1883, Irish scientist Osborne Reynolds discovered the number
that predicts fluid flow based on static and dynamic properties
like flow velocity, dynamic viscosity and density. It can be
defined as the ratio of inertial force and viscous force. As the
name indicates, it is a dimensionless number, which basically
indicates whether the flow past an object or in a duct is steady or
turbulent. If the flow is laminar, then the viscous force dominates
over inertial force and vice-versa for turbulent condition. Thus in
certain application where turbulence pertains, the fluid is
inviscid. For internal fluid flow, the Reynolds number is computed
by the following formula
Re = ρ∗v∗d µ
(2.1)
Where is the density (fluid property), v is the fluid velocity
(flow property), is the dynamic viscosity (fluid property) and d is
the diameter of pipe (geometrical property). For external flow L
replaces d which is also a geometrical property. The critical
Reynolds number is a value which specifies the transition of fluid
flow from laminar to turbulent, diversifies
15
regarding type of flow and geometry. Thus the applicability of the
Reynolds number varies depending on the enumeration of fluid flow
like alteration of density, variation of viscosity, flow being
internal or external and so on. As an example, for internal flow of
fluids in a pipe, transition occurs as indicated by critical
Reynolds number of 2300.
2.3.2 Flow through channels
A hydraulic fluid (SAE 15W-40) is used as pressure balancing fluid
for the dynamic component in the novel rod packing seal technology.
The fluid flows between the pump and the component through steel
tubes. Hence it is vital to perceive internal flow of viscous fluid
through pipes.
Head loss or pressure drop prevails in a piping system and there
are several reasons for this [10]. Forcing the fluid through a
channel or pipe fittings consumes energy which results in a
pressure drop along the channel and fittings. It is caused by
elevation of the pipeline, shaft work, friction along the walls and
turbulence provoked by abrupt changes in direction or
cross-sectional area. For a fully developed pipe flow, the
following relation for head loss is obtained from steady flow
energy equation.
= + ∗
(2.2)
The pressure drop across a horizontal pipe can be computed by the
equation given below:
∇ = ∗ ∗ (2.3)
Where is the head loss, is the height, is the pressure drop, is the
density and is the acceleration due to gravity. The amount of
kinetic energy contained in a stream of fluid is the velocity head
[11]. Velocity head can be alternatively stated as the amount of
potential energy necessary to drive a fluid to its flowing
velocity. Thus the potential energy is converted to kinetic energy.
The velocity of the stream is used to compute the velocity head, ,
as given below.
= 2
2∗ (2.4)
For pipes with fittings like pipe couplings, reducers, the excess
loss in a fitting is indicated by a dimensionless ‘K-factor’.
= ∗ 2
2∗ (2.5)
Head loss due to friction in pipes and fittings should also be
considered when computing the pressure drop across the flow. Thus
to calculate the total head loss accounting in the piping system
due to turbulence and friction, add to each sum of K factors the
friction loss and multiply the sum by the velocity head.
= ( Σ + ∗
)( 2
2∗) (2.6)
By combining equations (2.6) with (2.3), the pressure drop in the
piping system is given as follows
= Σ + ∗ ∗ ∗ 2
2 (2.7)
Friction factor
Friction factor f is a dimensionless parameter, which represents
the effect of friction between the fluid and walls in a pipe flow.
It can be used to compute the pressure drop and head loss in a
flow. It primarily depends on the velocity v, diameter D, density ,
viscosity , and wall roughness ε. Since friction factor is
dimensionless, the parameters it depends on should be in
dimensionless form. Thus friction factor is a function of Reynolds
number and relative roughness as given below.
= (,
) (2.8)
Friction factor as a function of Reynolds number and relative
roughness is formulated by various scientists and experts, some are
implicit and explicit relations. The most commonly used for flow
through commercial pipes is based on Darcy’s friction factor or
moody chart. It is based on the Colebrook-White equation. There are
many variants of Colebrook-White equation which can be solved
explicitly. For laminar flow through smooth pipes, the wall
roughness tends to be zero and hence the friction factor depends on
Reynolds number only and is inversely proportional to it as given
in equation 2.9. Friction factor for turbulent flow of fluid in
both smooth and rough pipes can be obtained either from the Moody
diagram or by computing Colebrook equation.
= 64
2.4 Governing Equations Pivotal topic like Navier-Stokes equation,
Hagen-Poiseuille flow will be briefed in the following
sub-sections. The topics discussed in this chapter will also
support background knowledge for numerical simulations carried out
with ANSYS Fluent, a computational fluid dynamic tool.
2.4.1 Navier-Stokes equation
The motion of a fluid is basically described by the Navier-Stokes
system of equations [9]. Application of three laws of conservation
namely, (1) conservation of mass, (2) conservation of linear
momentum and (3) energy conservation are applied to derive the
system of
17
equations. The law of conservation of mass and linear momentum are
expounded in this section, as it is of major significance for this
project to evaluate the pressure and velocity of flow.
Conservation of mass is often known as continuity relation, denotes
that the fluid mass cannot change. It is basically a differential
equation derived by deeming either an elemental control volume or
an elemental system. It results in a partial differential equation
involving the derivatives of density and velocity and is known as
continuity equation. General forms of the equation for
incompressible flow, in both Cartesian and cylindrical coordinates
are given below.
+
+
The differential linear momentum for an infinitesimal element,
constitutes the following terms
1. Gravity force per unit volume 2. Pressure force per unit volume
3. Viscous force per unit volume
The sum of these terms corresponds to density multiplied by
acceleration. The component form of the law of conservation of
momentum is represented as
− ∇ + ∇. =
(2.13)
For a Newtonian fluid, the viscous stresses () are proportional to
element strain rates and the coefficient of viscosity. The
differential momentum equations of an incompressible flow for one
direction only in both rectangular and cylindrical coordinates are
given below.
−
(2.15)
It can also be written for other dimensions for both the
co-ordinates. This system of equations is essentially known as
Navier-Stokes equations.
Navier-Stokes equations have only a limited number of analytical
solutions, but computer numerical modeling is highly built upon
these fundamental equations. It is possible to attain approximate
and realistic results for various complex two- and
three-dimensional viscous flows using any commercially available
CFD tools.
2.4.2 Hagen-Poiseuille flow
G. Hagen in 1839 and J.L. Poiseuille in 1840 observed
experimentally the incompressible flow in a straight circular pipe.
For fully developed laminar pipe flow, the above mentioned
Navier-Stokes equation is solved by stating initial assumptions to
simplify the complex mathematical equation and then implementing
the right set of boundary conditions according to the application.
This leads to an empirical relation to evaluate the velocity of
flow as given in equation 2.16 below.
= (−
= 2
8 (2.17)
This is known as Hagen-Poiseuille flow equation. From equation
2.16, we can interpret that velocity of flow in a pipe varies only
in radial direction and its profile is parabolic. The corresponding
relationship for velocity of flow in z-direction is valid only for
incompressible laminar flow.
2.5 Analytical Background This chapter covers the analytical
background to develop the analytical model for numerical analysis.
The model provides a correlation between pressure difference and
axial displacement that varies the orifice gap. The mathematical
formula is derived from hydrostatic bearing design, which gives
relationship between flow rate and pressure drop for a given gap
height. In other words, the formula is obtained from pressure
distribution in stepped film and radial flow through long thin film
of hydrostatic bearings. Hence pressure distribution from these
models will be discussed in brief in this section.
Stepped film –
The pressure distribution in the stepped film is derived based on
Poiseuille flow with a film thickness of h [12]. In the part with
film thickness of h, pressure drops from to . is the pressure in
the supply side and is the pressure in the return side.
(2.18)
The above equation gives the flow rate of the fluid per unit length
for a known pressure drop across the cross-section. It has a linear
pressure drop across the fluid film of thickness h which has a
length w.
Radial Flow through long thin gap –
For radial flow along long thin parallel film, the pressure
distribution and flow resistance are derived from Reynolds
Equation. Similar to the equation as obtained for stepped film,
flow
19
rate in relation with pressure difference and film thickness can be
derived from RE in polar coordinates. For linear pressure
distribution, the following equation is used.
= 3
12 1+(1 ) 1+(1 )
(2.19)
Chapter 3
Methodology This chapter expounds the implementation of methods and
tools to accomplish desired outcomes; as well as helps to
comprehend the behaviour of the dynamic component. In the
beginning, in order to plan the testing and to solve numerically /
analytically, identifying the influential parameters that produces
an effect on the system is crucial. This is followed by exploring
the experimental test-setup and design modification. This
sub-section also furnishes a clear picture on control drives
configuration / parameterization for pump and motor. It then
extends to numerical/computational analysis of the dynamic system.
This encompass various steps, namely modelling fluid domain,
setting-up physics for the domain in the ANSYS fluent solver and
solving differentials using appropriate solver in MATLAB.
3.1 Parameter identification In order to ascertain the effects
(positive / negative) on the behaviour of the dynamic component,
the essential part is to identify the influential parameters. The
novel rod packing technology with the dynamic component is shown in
figure 3.1 below. The result on the effects of the parameters will
be covered in later chapter. It is significant to identify and
analyse the parameters, so that it can facilitate product
development / design improvements.
Figure 3.1 Dynamic component of the novel rod packing
technology
• Flow rate of hydraulic fluid • Mass of the dynamic component •
Friction acting on the dynamic component • Tube length of hydraulic
fluid flow • Geometry of the dynamic component
21
Flow rate of hydraulic fluid –
The fluid pressure (hydraulic oil), corresponds to flow rate and
the axial distance between the orifice and the dynamic component.
The displacement also tends to vary with the flow rate. This term
is represented either as mass or volumetric flow rate. In CFD, for
applying boundary condition, it’s usually given as mass flow
rate.
Mass of the dynamic component –
The geometry and material selection for the dynamic component
constitutes the mass. Mass of the dynamic component have an effect
on the motion such as displacement and velocity of the
component.
Friction on the dynamic component –
There will be two types of contacts established on the assembly of
the dynamic component.
• Polymer-metal contact • Metal-rubber contact
The polymer-metal contact is between the piston rod and sealing
rings. These rings are made of different materials. For example, a
set of rings consists of cast-iron ring and polymer rings. Polymer
rings are usually PTFE-filled, but the constituent of this ring is
proprietary. The O- ring in association with high pressure oil cup
(encompasses the dynamic component) results in metal-rubber
contact. Material of the O-ring used is FKM-75. Friction is of
paramount importance as it affects the system dynamics to greater
extent. The problem is convoluted and so it is hard to model it
analytically or to analyse numerically. In this thesis work,
arbitrary friction force value is assigned to solve the equation of
motion to simplify the complexity in formulation of friction force.
In section 3.3.6, detail explanation on equation of motion is
provided.
Tube length of the hydraulic fluid flow –
Viscous flow of fluid through pipes as reviewed in section 2.3.2
has an effect on the fluid dynamics. Thus the geometry (diameter
and length) of the tube and wall roughness need to be taken into
consideration.
Geometry of the dynamic component –
The design of the dynamic component (packing ring cups which holds
the sealing rings) needs to be considered for analysis. The
geometry of the design might have an effect on the dynamics of the
component and on to the functioning (pressure balance). Different
geometry configurations can be studied to assimilate its effects
accordingly.
3.2 Experimental Testing
Experimentation is considered as an effective method for scientific
study of a concept or to discern the principle behaviour of a
system. Precise result is one of the main advantages of exerting
experimental methods. The prototype of the novel rod packing
technology for the Ariel reciprocating compressor JGH/4, which is
manufactured for the original size, is used for testing.
Manufacturing of the prototype requires special care, so that exact
tolerances are maintained as specified in the drawings. Other
components which forms up the test rig is also designed and
manufactured. The experiments are conducted at KTH Royal Institute
of Technology, in the Machine Design department, where the Test-rig
is set up. The following sub-section will briefly describe the
test-rig setup, instrumentation of the rig, data acquisition, and
control drives.
3.2.1 Test-rig
The test-rig comprises of several parts / components which are
assembled and connected mechanically with each other. The test
frame provides support to the various parts and components as shown
in figure 3.2 below. Steel beams of square and rectangular cross-
section are welded to each other, which constitutes the frame. It
is constructed in such a way that it is rigid, and the legs of the
frame is provided with rubber dampers to reduce the vibration.
Vibration is induced by the running compressor. A double acting
piston compressor is used which simulates the principle function of
the reciprocating compressor used in the methane gas transmission
application.
Figure 3.2 Test-rig setup
23
The experimental setup incorporates various parts which lead to
successful working of the compressor and the rod packing
technology. As shown in the schematic representation of the
arrangement in figure 3.3, it contains an electric motor, a single
cylinder modified marine engine, a Bosch Rexroth pump, a novel rod
packing, a lube pump, and a double acting piston compressor. In
this arrangement, the motor transmits torque to the single cylinder
Stirling engine. Through a coupling, the piston rod of the
compressor is connected to the modified piston-cylinder of the
engine. The axial coupling selected should withstand a force of
4900 - 8000 N. This force (pull and push) is from compressing the
gas across the piston cross- sectional area. Thus the rotational
motion of the motor is converted to reciprocating motion which
imparts movement to the double acting piston cylinder compressor.
The Bosch Rexroth gear pump supplies hydraulic fluid to the dynamic
component of the rod-packing seal, which recirculates back to the
pump. Steel tubes of various dimensions are used as channels for
fluid flow.
Figure 3.3 Schematic arrangement
Nitrogen is the gas, which is used for experimental testing, to
prove the concept of pressure balancing of the dynamic component.
Since, nitrogen is inert in nature; it neither reacts with oil nor
ignites at high temperature and pressure. It is stored at 300 bar
(g) pressure in a plastic cylinder which can contain a total volume
of 6000 normal liters. At a regulated pressure, the gas is supplied
through a steel reinforced flexible hose to the inlet of the double
acting piston cylinder.
The electrical motor is controlled with a frequency controller
(Danfoss drive) and speed is varied from 0 to 1500 rpm. The testing
is conducted at 1000 rpm with a piston velocity of approximately 2
m/s. When the compressor runs at around 400 and 700 rpm, the whole
set-up vibrates which is undesirable.
3.2.2 Design and construction
The following parts constitute the compressor, which is a
reciprocating type, double acting cylinder, piston compressor. The
cross-section of the design with the novel rod packing seal
assembly is shown in 3.4 below.
• Piston and piston rod • Main cylinder
Figure 3.4 Cross-section of double acting piston compressor with
novel rod packing assembly
The cylinder is designed based on the compression ratio required,
stroke length of the engine and the rod packing assembly so that
both can be fitted together conveniently. The cylinder bore is a
known value based on the piston dimension which is fixed (39.5 X 40
mm). The required compression ratio on the bottom side of the
piston (for a double acting cylinder) is set as 3, and the length
of the cylinder bore is calculated based on the simple formula
given below.
C.R = ∗2−2∗ ∗(2−2)∗
=
(3.1)
R is the cylinder bore radius, r is piston rod radius which is also
fixed value. The main cylinder is designed such that it can be
easily attached and detached with the existing cylinder head and
rod packing assembly. They are assembled together with bolts. The
cylinder needs
25
to equip a pressure transducer, a quick coupling for gas inlet, and
a pressure relief / drain valve. The exact end connection needs to
be machined. A volume adjuster is designed to improvise the control
volume and adjust the compression ratio. Refer appendix A for
detail drawings.
A double acting piston seal and piston rings are selected based on
cylinder bore value and type of material required. Accordingly,
piston is designed to fit the selected seal and guide rings. Proper
tolerance is specified in the drawing as required between piston
seal and cylinder bore. Tolerance on the piston seal is given in
the manufacturer’s catalogue. Material selection is also a part in
the design phase. The material used are steel, aluminum and
stainless steel for different parts accordingly.
The rod packing sealing rings used are the existing market products
as discussed in section 2.2 (packing ring designs) which are
purchased from Ariel corporation. These rings are almost similar to
the set of ring pairs used in a traditional packing design.
Different ring pairs (double-acting BTR rings, single-acting BD
ring pair and scrapper rings set) are placed within the rod packing
assembly and the dynamic component. Design of rod packing assembly
and dynamic component is shown in figure 3.5 below. Based on this
design a fluid model is developed for CFD analysis which is
discussed in detail in the following section 3.3. Manufacturing of
the dynamic component and the pressure cups (low- and high-pressure
cup) needs to be precise with regards to the specified
tolerances
Figure 3.5 3D CAD model of a) Rod packing assembly b) Dynamic
component
3.2.3 Test objective
The main goal to be achieved from experimental analysis is to study
the pressure curves of hydraulic fluid and compressed gas and how
it behaves regarding changes with parameters like flow rate, tube
connection, and geometry. Another objective is to estimate the
leakage rate of the product and compare with the benchmark result.
The measured pressure data is analysed with LabVIEW software.
3.2.4 Instrumentation
The gas is compressed inside the main cylinder and cylinder head;
hence it needs to be rigged with the required instruments. When the
piston reciprocates, the bottom side of the piston is supplied with
nitrogen at a constant pressure and flow rate from the storage
cylinder. The reciprocating motion compresses the gas to high
pressure up-to 150 bar. In the rod packing seal, the dynamic
component is immersed in hydraulic oil. When the gas compresses and
expands, the dynamic component also reciprocates which pressurizes
the fluid. The pulsating pressure needs to be measured. The
frequency of the pulsation depends on the speed of the motor. Hence
a high frequency pressure measurement device is selected. The
pressure transducer is installed at 3 locations, in the main
cylinder, high pressure and low-pressure oil cups of the rod
packing seal. The pressure transducer adopted can measure both
liquid and gas pressure with a range of 0-250 bar. Pressure
transducers need to be excited with an external DC supply. The
input voltage range for the transducers is 0-30VDC. An AC to DC
power supply unit is connected to these transducers to supply
24VDC. The output of the transducer is connected with National
Instrument’s pressure module (data acquisition module) to analyze
the measured signals.
Figure 3.6 Instrumentation of the test-rig parts
At the cylinder head the temperature of the compressed gas is
measured over time. A thermocouple is used to record the
temperature. J-type, in-built cold junction thermocouple is
connected to the cylinder head. The other end of the thermocouple
has two terminals, positive and negative terminal. The two
terminals are connected to the temperature module to visualize the
data using NI MAX tool. The modules will be discussed in brief in
the following sub-section.
27
Since the compressor will be working at high pressure, it is
required to provide safety. Henceforth, a pressure relief valve is
connected to the main cylinder as shown in figure 3.6 above. The
pressure relief valve is set to relieve when the pressure reaches
200 bars.
The temperature on the oil is measured from the Bosch pump on the
return line. Both the oil contamination sensor and temperature
sensor are integrated with the pump. Thus when temperature reaches
the limit it indicates a warning in the IndraWorks software and
shut down the pump.
3.2.5 Data Acquisition
The pressure module and the temperature module as mentioned in the
above section, are connected to the NI compact DAQ which is also
known as chassis for the modules. This chassis is connected to the
system to observe and analyze the measured signals. This compact
DAQ can accommodate 4 modules. Four of the pressure sensors and
thermocouple can be connected to the pressure module (NI-9239) and
temperature module (NI-9211) respectively.
National Instruments- Measurement and automation explorer (NI-MAX)
software interfaces with the hardware to analyze the signal. The
working platform of the NI-MAX is shown in figure 3.7 below. Using
c-DAQ assist feature, the device linked to the corresponding module
can be connected to record the measured signals. The temperature
data of the compressed gas is reviewed in NI-MAX.
Figure 3.7 NI-MAX platform
The gas and liquid pressures are both analyzed using LabVIEW
software. DAQ-assist function from the express VI function palette
is used to interface with the pressure module. LabVIEW has two
windows, namely front panel and block diagram. A graphical program
is developed with the help of function palettes in the block
diagram window and the waveform charts / graphs to plot the
pressure data is created and viewed in the front panel. To
record,
analyze and store the pressure data of both fluids (gas and oil),
the graphical program is developed as shown in figure 3.8. The data
measured is stored as excel file using write to measurement file
express VI. Write to measurement file express VI icon is seen in
LabVIEW front panel of figure 3.8. It is used to edit and plot the
required range of datas.
Figure 3.8 LabVIEW front panel and block diagram window
3.2.6 Drives and Control
The remote operation and control of the electrical motor and the
Bosch Rexroth CytroPac pump, necessitates the use of control
drives. The electrical motor is connected with Danfoss frequency
converter. Using an Ethernet cable a connection is established with
the computer. VLT motion control tool MCT 10 is the supporting
software to interface with the frequency converter. The related
parameters that allow access to remotely control the motor is set
in the MCT 10 parameterization dialog box as shown in figure 3.9
below. The parameter file edited in the project is used to access
the drive remotely. The motor speed is set and controlled using
this tool, by varying the frequency value in terms of percentage.
Desired working speed of the motor is 1000 to 1200 rpm. Update
output frequency button of the remote controller is used to update
the ongoing request to change the motor speed. Similarly, start and
stop control option is used to switch on and off the motor
intermittently.
29
Figure 3.9 VLT Motion Control Tool for remote control of the
motor
Figure 3.10 Parameter editor dialog box for IndraWorks DS
14V20
Sytronix FcP 5020, the frequency converter is inbuilt with the
Bosch CytroPac pump, controls the motor. USB cable connects the
pump system with the computer. IndraWorks DS 14V20 is the software
tool to communicate with the frequency converter. It is also used
to view and analyze the working of the system and diagnose warnings
and errors. There are many parameters related to the FcP 5020
converter, but the essential parameters are edited according to the
application. Parameter values can be viewed in parameter viewer
file as shown in figure 3.10 and is edited in the parameter editor
dialog box. The inlet flow rate for the rod packing seal is set to
0.8 l/min using the flow command parameter (F1.12). The pump motor
automatically shuts when the pressure in the return line reaches
the set pressure limit of about 150 bar.
3.3 Computational analysis of the dynamic system
The sealing cups with the packing rings in the novel rod packing
technology, together constitutes as dynamic system as discussed in
section 3.2.2. It is expensive to analyze the dynamic behaviour and
functionality of the system experimentally for different
parameters. Hence it is essential to make computational analysis to
vary certain parameters and analyze the outcome. For FE modelling
and analysis, the model is developed with SolidEdge surface
modelling and imported to ANSYS. Fluent pre- and post-processor is
used for setting-up physics and analyzing the results respectively.
A mathematical model that represents the physics of the system is
solved using MATLAB numerical solver. The analysis of both the
fluid model and the mathematical model are discussed further in the
following sub-section.
3.3.1 Scope and method development for CFD
Computational fluid dynamic analysis is required to understand the
pressure development of the hydraulic fluid, which acts on the face
of the dynamic component as shown in figure 3.11, and for various
displacement of the dynamic component. Different methods are
identified to build the FE model.
1. A 2D- axisymmetric model that constitutes a solid domain, a
liquid domain, and air domain.
2. A 2D-axisymmetric model of the fluid domain 3. A 3D model of the
fluid domain
Figure 3.11 Hydraulic fluid acting on the face of the dynamic
component
The cross-sectional view of the novel rod packing technology as
shown in figure 3.4 above is analyzed to construct a 2D-
axisymmetrical model as described in the first method. In this
method, the contour of the dynamic component which acts as a solid
domain is modelled. The two fluid domains contour are modelled. All
the domains need to be connected using the mesh interface. The
simulation from this method will represent the function/behavior of
the dynamic component as analyzed experimentally. One fluid domain
is defined for the hydraulic fluid (viscous oil) properties and
other domain represents the air. Aluminum is defined for the solid
domain. This model is complex as it requires complex dynamic mesh
motion and meshing methods for the successful simulation of the
model. Also, the objective to obtain the correlation between the
fluid pressure and the displacement can be obtained with the second
method as stated above. In this method the fluid model is developed
by analyzing the cross-sectional contour of the fluid region inside
the pressure cups (high pressure and low- pressure cup). The
contour comprises boundaries of different elements of the novel rod
packing seal. Geometry of this model is explained further in the
following sub-section. This model is simple, takes less
computational time to achieve the above stated objective. Hence, it
is the chosen method for CFD. The third method is just obtained by
revolving the fluid model developed for second method into a 3D
model. The result for fluid pressure obtained will be like that of
the previous method, and it also has the disadvantage of consuming
higher
31
computational time and meshing is limited. So, this method is also
discarded. The model developed for the three methods is shown in
figure 3.12 below.
Figure 3.12 Fluid model for CFD analysis a) 2D-axisymmetric model
for fluid domain b) 3D model of the fluid domain c) 2D-axisymmetric
model with both solid and fluid domains
3.3.2 Fluid model
For the second method as discussed in section 3.3.1 the fluid model
is developed. The hydraulic fluid from pump as discussed in section
3.2.1 enters and leaves the novel rod packing seal in the radial
direction i.e., radially outwards from the piston rod. Around the
dynamic component the flow of this viscous fluid is both
circumferential and tangential. As the flow in the direction of the
cylindrical coordinates of the system is not of importance, a
2D-axisymmetric model (x-y plane of rectangular coordinates) is
developed. SolidEdge surfacing feature is used to create the 2D
model which is imported to ANSYS for further analysis. As shown in
figure 3.12 a) above, the contour of the fluid model comprises of
boundaries of various elements which faces the fluid. In simple
terms, the face of the components that is in contact with the fluid
forms the boundary of the defined model. To obtain the correlation
of the high-pressure development with the displacement of the
dynamic component, the boundary lines of the contour of dynamic
component is modified for different position (termed as x). So, it
is a repetitive method, where the pressure of the hydraulic fluid
is analyzed for different displacements.
3.3.3 Meshing
The essential stage in a computational fluid dynamic analysis is to
mesh the model. ANSYS mesh modeler is used to generate the 2D
quadrilateral linear order mesh elements for the fluid domain. In
the fluid model, the gap in the orifice is very small as shown in
figure 3.13 below. The gap in the orifice means, the distance
between two boundary lines (one is the slanting line of the dynamic
component face and other is the slanting line of the low-pressure
cup). Meshing in this gap needs to be refined since, it is where
the fluid pressure drops, and the velocity profile is
parabolic.
Figure 3.13 Linear quadrilateral mesh elements
Proximity and curvature mesh sizing option and fine meshing is
chosen for the models. Mesh convergence study is done for a liquid
model with 0.3mm axial gap in the orifice. It is observed that the
meshing did not have major effect on the pressure drop along the
orifice. So, for different models (with varying axial distance in
the orifice) an average of about 100,000 mesh elements is
generated. The computational time is also comparatively low
compared to meshing the 3D fluid model for about 50,000 elements.
Skewness in the mesh elements is maintained to a low value around
0.005. Further the meshing quality is also checked under fluent
processor.
3.3.4 Setting up physics
A steady-state, pressure based, axisymmetric fluent solver is
selected. Since the flow from the pump through the channel to the
rod packing seal inlet is smooth and constant, the laminar flow
model is considered to solve the 2D axisymmetric model. For the
fluid model, the hydraulic oil used in the experiments is defined
to the model by inputting the properties of the fluid as given in
table 3.1 below.
Table 3.1 Hydraulic fluid properties
33
Property Value Unit
Density 879 kg/m3
Viscosity 0.132 Kg/m-s
The boundary condition is a major step in setting-up physics to
solve flow problems. In the experiments conducted, a known flow
rate from the pump is supplied to the inlet line. At the inlet
boundary (marked blue) of the fluid model as shown in figure 3.14,
a mass flow inlet is defined as the boundary type. Mass-flow rate
is given as 0.0103 kg/s. At the outlet the boundary type is set as
pressure outlet, and a gauge pressure value of 300000 Pa is
set.
Figure 3.14 Boundary conditions defined to the fluid model in
setting-up physics.
3.3.5 Solving and Post processing
There are different numerical solvers for Fluent to solve the
problem it encounters. The details about the solvers can be found
in [13] and here the type of solution methods used will only be
discussed. PISO type pressure velocity coupling scheme and for
pressure spatial discretization PRESTO are selected for solution
methods. For other values and methods, it is set to default
options. This solution methods solves the flow equation. The
numerical solver that the fluent implements is an iterative type to
obtain converged solution. The problem needs to be initialized and
a hybrid setting is provided for initialization.
The main objective for computational analysis is to obtain
correlation of the fluid pressure build up for different
displacements of the dynamic component. From the fluent post
processor, the following results is obtained for different orifice
gap.
• Pressure drop across the orifice and the maximum pressure
developed surrounding the dynamic component.
• Average flow velocity and streamline function
In fluent post-processing, a contour plot is used to obtain the
graphics for absolute pressure and total velocity component of the
fluid model. As stated in the objective, it is essential to note
down the maximum absolute pressure value for every model with
different orifice gap. The values of pressure against the
displacement is plotted in a line chart.
3.3.6 Analytical model and Equation of motion
Let’s consider the dynamic component (with sealing rings and
packing cup) as a single body with mass m. Air pressure acts on one
side of the dynamic component which is sinusoidal in nature due to
compression and expansion of air over time . Due to this pressure
the body moves towards the other side and the pressure builds up on
the hydraulic fluid side, since an incompressible viscous oil is
used. When the pressure is greater than air pressure, the body
moves back. Henceforth, back and forth motion is obtained. The
free-body diagram that describes the physics of the system is shown
in figure 3.15
Figure 3.15 Free body diagram of the dynamic component.
From the above free body diagram, an equation of motion can be
formulated that entails all essential parameters.
x = − + 0 − x − x () (3.2)
The above ordinary differential equation is solved using a
numerical solver like ode 45 using MATLAB. Refer Appendix B for the
code. The air pressure is generated using the equation 3.3 given
below, which is sinusoidal in nature and is given as an input
parameter. Similarly the damping coefficient, c of the hydraulic
fluid is defined as constant. A frictional force value is guessed
and is given as arbitrary input to the friction component of the
above given equation of motion.
35
= + ( ∗ sin (2 ∗ ∗ )) (3.3)
With M as mean pressure, A is the amplitude of sine wave, and T is
the periodicity. Thus, air pressure is generated which is a
function of time. Oil pressure term is calculated within the
numerical solver loop, since it is a function of x displacement. So
everytime when x is solved it will be given as input to compute the
oil pressure. So a correlation between oil pressure and
displacement x is obtained and the formula is presented
below.
= ∗ ∗∗∗
+ (3.4)
The formula obtained as given in section 2.5 is the analytical
model that represents the CFD model developed to obtain the
pressure accretion in the hydraulic fluid domain based on the
displacement of the dynamic component. So the CFD model is
validated by this analytical model.
Hydraulic fluid pressure force projected on to the cross-sectional
area of the component is not the same area as the air pressure
force projected area as shown in figure 3.16 below. Hence for the
given geometrical design of the component, the pressure developed
in the hydraulic fluid side is slightly greater than the generated
air pressure.
Figure 3.16 Fluid pressure acting on the dynamic component
The 0 component of the equation of motion, where the projected area
on the oil side () is divided into three sections. 1 is the
projected area where maximum oil pressure acts; 2 is the projected
area where minimum oil pressure acts. The force acting on the
orifice gap is computed by integration of varying oil pressure
force (pressure multiplied with the circumference). This
constitutes the third section. Thus the equation 3.2 is rewritten
as
× = − + 10 + 20 + ∫ () × 2 × − × − × ()
The integration of pressure force is performed over the limits of
and , which represent the inner and outer radius of the orifice
respectively.
3.3.7 Geometry design configurations
Different design configurations are developed by varying the
geometry outline of the dynamic component. This is used to study
the geometry’s effect on the dynamics and as well for functional
analysis of the system. For basic understanding of different
configurations, the outline of the fluid domain is developed using
surfacing feature with SolidEdge as shown in figure 3.17 below.
Instead of representing the geometrical change for different
configurations in the 3D model of the dynamic component it is
represented through the 2D outline of the hydraulic fluid domain.
The 2D fluid model is a result of cross-section of the rod-packing
assembly as shown in figure 3.4.
Configuration 1 represents the chamfer edge of the orifice with an
angle of 45°; configuration 2 represents the chamfer edge of the
orifice with an angle of 30°; configuration 3 represents the
chamfer edge of the orifice with an angle of 60°. More geometry
configurations can be studied, but for this thesis work analysis of
only three configurations are conducted to check if the developed
analytical model can be used to study their effects.
Figure 3.17 2D-fluid model for different configurations a)
configuration 1 b) configuration 2 c) configuration 3
37
Results and Discussions
This chapter provides the results from numerical simulation carried
out by CFD and solving the analytical model. The first section
describes the oil pressure development in the hydraulic fluid side
of the dynamic component using ANSYS Fluent. The CFD model
developed for the fluid domain is validated with an analytical
model. Then the results from transient simulation for the dynamic
system, by solving the equation of motion with MATLAB, are
presented. After validating the model with the previous work of
experimental results, analysis of pressure difference and
displacement for different configurations are also presented in
this chapter.
4.1 CFD simulation of Liquid Domain The result from ANSYS Fluent
gives the pressure distribution of oil in the developed liquid
domain over a range of orifice gap. Varying the orifice gap
represents the displacement of the dynamic component. Thus from
this static CFD analysis, a corresponding relationship between the
maximum pressure and displacement is obtained. The pressure
distribution in the liquid (hydraulic oil) domain for different
gaps is shown in figure 4.1 below focus the orifice gap.
39
Figure 4.1: Pressure distribution for different orifice gap a) x=
0.07mm b) x= 0.12mm c) x=0.16 d) x= 0.22mm
The pressure in the red region as shown in figure 4.1 above
represents the maximum pressure formed that acts on 3/4th of the
projected area of the dynamic component. The maximum pressure
acquired for different displacement is plotted and an exponential
curve is obtained as shown in figure 4.2 below.
Figure 4.2 Displacement versus oil pressure plot
4.1.1 Validation with Analytical Model
As discussed in section 3.3.6, analytical models that represent the
pressure drop over the orifice, as obtained with the CFD model, are
solved analytically with MATLAB. The results of various analytical
models are compared with the CFD result as shown in figure 4.3. The
three analytical models closely match the CFD result. The models
are solved using the same boundary conditions. Hence the CFD model
is validated.
3,00E+05
8,00E+05
1,30E+06
1,80E+06
2,30E+06
2,80E+06
3,30E+06
Figure 4.3 Comparison of analytical model with CFD model
4.2 Dynamic Response Analysis In this chapter the dynamics of the
novel sealing solution is analyzed through numerical simulation of
the equation of motion of the system which is solved using MATLAB.
At first the pressure curves for both generated air pressure and
computed oil pressure is plotted together for analyzing the
pressure difference. Then the displacement of the dynamic component
is analyzed. The performance of the component dynamics for varying
friction force and geometry (chamfer angle) is also discussed and
analyzed.
4.2.1 Time-domain analysis
The major analysis presented in this thesis work involves the
pressure balancing of the fluid for the rod packing seal and the
displacement of the dynamic component. Pressure balancing plot as
shown in figure 4.4 below is obtained for initial design
configuration of the dynamic component as discussed in section 3.4
with inlet flow rate of 0.8l/min of hydraulic fluid, supplied air
pressure ranges from 5.67 bar to 6.72 bar. The force acting on the
component from the hydraulic side is greater than the force acting
on the other side of the component. This is reflected in the
pressure plot, as the oil pressure is slightly greater than the
compressed air pressure. The plot shown below is generated for an
arbitrary friction force value of 200N acting on the component.
This plot is comparable with the experimental results from previous
work and hence the model is validated. The average pressure
difference is 9.5e4 Pascals as given in figure 4.4 below.
41
Figure 4.4 comparison of air pressure with hydraulic fluid
pressure
The corresponding displacement of the dynamic component which is
influenced by the forces acting on the component is analyzed in the
figure 4.5 given below. For the base configuration of the design
with a friction force of 200N the travel of the component is found
to be 9 µm.
Figure 4.5 Displacement plot of the dynamic component
4.2.2 Performance Analysis
The dynamic behavior and the functionality of the sealing component
is analyzed for various frictional force and for three different
configurations. As shown in figure 4.5, with increase in friction
force the travel of the component drops which is indicated by
flattening out of the oil pressure when friction force reaches
about 500N. Thus, the amplitude of the oil pressure decreases over
increasing friction. The plots shown in figure 4.6 are obtained for
the base configuration of the dynamic component.
Figure 4.6 Oil pressure and air pressure plots for different
friction force starting from top left corner with friction force of
a) 50N b) 100N c) 300N d) 500N
The corresponding displacement plot for 50, 100, 300 and 500 Newton
of friction force is shown in figure 4.7 below. Travel of the
dynamic component for these friction forces are 0.013mm, 0.011mm,
0.005mm and 0.001mmrespectively.
43
Figure 4.7 Displacement of the dynamic component for different
friction forces starting from top left corner with friction force
of a) 50N b)100N c) 300N d) 500N
By comparing the results with previous experimental results for the
base configuration, based on the amplitude of oil pressure and the
pressure difference (Δp), the friction force on the dynamic
component is estimated to be around 200 N. For this frictional
force, the displacement of the dynamic component and pressure
balance for three different configurations is plotted and analyzed.
As shown in figure 4.8 below, the computed oil pressure for three
configurations and generated air pressure are plotted over time. In
the plot, configuration 1 represents a chamfer angle of 45°;
configuration 2 represents a chamfer angle of 30°; configuration 3
represents a chamfer angle of 60°.
Figure 4.8 Comparison of oil pressure for three different
configurations
The corresponding travel of the dynamic component for the three
configurations are plotted in the figure 4.9 shown below. The data
acquired by computing the equation of motion, is solved
with a friction force of 200N acting on the component. The axial
orifice gap between the chamfer edge and the component varies for
different configurations which can be noticed from the plot given
below. The values for displacement range and pressure difference
for the three configurations as obtained from the plots as shown in
figure 4.8 & 4.9, are tabulated below.
Figure 4.9 Comparison of travel of the component for different
configurations
Table 4.1 Values of average pressure difference and displacement
range for different configurations
Types Average Pressure difference (Pa)
Average displacement range (mm)
Configuration 1 (45° chamfer) 9.4527E+04 0.009
Configuration 2 (30° chamfer) 6.5229E+04 0.011
Configuration 3 (60° chamfer) 5.9466E+04 0.007
To analyze the effect of friction force on the pressure balancing,
an assumption is made to the model. The plot as shown in figure
4.6, cannot be used to find the effect of friction force onto the
pressure difference. Reason is computing average pressure
difference for different friction forces will give similar values
which doesn’t not make sense. So force for maximum oil pressure 0
is assumed to act on the entire cross-sectional area of the
component. But in reality, oil pressure force varies over the face
of the component as given in section 3.3.6.
45
Normalized pressure plot for the base configuration with different
friction force is obtained as shown in figure 4.10 below.
Figure 4.10 Normalizing pressure plot for different friction forces
a) 50N b) 150N c) 250N d) 500N
From the plots as given in figure 4.10 above, average pressure
difference is computed which is plotted against friction forces as
shown in figure 4.11 below. From the influence of friction force on
pressure difference plot, a linear relation between friction force
and pressure difference is obtained. Also from this plot it is
noted that the pressure difference value will not change once a
limit is reached. In this case, 450 N is the limiting friction
force.
4,00E+03
9,00E+03
1,40E+04
1,90E+04
2,40E+04
2,90E+04
3,40E+04
0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0
Δ P_
A V
E R
A G
E I
N P
A SC
A L
ΔP for 45degree(Pa)
Figure 4.11 Influence of friction force on pressure difference for
base configuration
4.3 Experimental result Test-rig modification was the beginning
part of the work which involved designing and fabricating parts for
double acting piston cylinder with additional attachments as per
the requirements. Refer to appendix A, for detailed drawings of the
components. Benchmarking test for the competitor’s product was
conducted, in which long time test run was unsuccessful due to heat
buildup and the temperature did not stabilize as expected. Tried
couple of water- cooling methods to bring down the temperature,
which did not work as well.
Before starting the tests for the novel rod-packing seal
technology, lot of time and work was spent on troubleshooting the
test-rig. Moreover, at the beginning of testing the technology,
problems were faced with oil leakage along the rod out from the
packing flange and scrapper cup. A reason could be a manufacturing
defect of the packing flange, where proper tolerance needs to be
maintained at the BD ring seat.
These problems resulted in that the experimental part of thesis
work could not be completed as planned
47
Chapter 5
Conclusions This chapter delineates the conclusions which are
inferred from the computational analysis of the analytical model
developed for the design of the novel rod packing seal technology.
In summary from the previous chapters, an analytical model was
framed based on the CFD analysis and was used to solve the physics
of the system for transient analysis. The behavior of the dynamic
component is analyzed with an analytical model through numerical
simulations but could not obtain results through physical testing.
Also parameters like friction force and geometry are studied with
the validated analytical model. From the influence of friction
force on pressure difference study, a linear relation between
friction and pressure difference is observed. Also, by changing the
geometry (chamfer length and angle of the dynamic component) of the
dynamic component, it can be observed that design configuration
with 60° chamfer angle gives smaller pressure difference value
compared to the original design.
• Results from CFD analysis closely match with the analytical model
which provides correlation between hydraulic fluid pressure and
axial displacement of the dynamic component.
• The model developed and analysed in this thesis is suitable for
functional analysis of the component and influence of certain
parameters like geometry, friction force, mass and flow-rate can be
studied.
• The friction force value is estimated to be around 100-200 N,
from the model which represents the dynamics of the component and
on comparison with the previous testing results.
• From the performance analysis for different friction forces and
its influence on pressure difference, it can be noted that, a
linear relation is obtained between friction force and pressure
difference. Also friction affects the dynamics and for the case
analysed in this thesis it can be seen that when friction force is
above 450 N oil pressures become constant. This means that the
component does not perform as it should.
• The developed model provides a good estimate of the pressure
balance and trav