Degree Project in Machine Design, SECOND CYCLE, 30 ECTS
STOCKHOLM, SWEDEN 2019
Mechanical Face Seals:
Test Rig Development and Analysis of the Frictional Behaviour
Àlex de la Cruz Gargallo Víctor Pérez Palomino
KTH SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT
Mechanical Face Seals:
Test Rig Development and Analysis of the Frictional Behaviour
Àlex de la Cruz Gargallo Víctor Pérez Palomino
Master of Science Thesis TRITA-ITM-EX 2019:248 KTH School of Industrial Engineering and Management
Department of Machine Design Stockholm, Sweden 2019
Abstract
Mechanical face seals, MFS, are important machine components in the heavy- duty-vehicle industry which generate a high frictional power loss. In order to lower these losses an investigation has to be performed on the frictional behavior of the MFS. A physical test rig allows an evaluation of existing seals and new concepts to compare their performances.
Prior to this project, a test rig for MFS had been partially designed and manufactured by a group of students within a design project at KTH. This report presents the continuation of this work.
A solid background study was performed in which the best testing equipment and theoretical models in the literature to characterize MFS were reviewed. The design and manufacturing of the test rig prototype has been completed and most importantly, it has been validated, achieving an acceptable assessment of repeatability. Tests have been planned and executed to assess the repeatability of the test rig and to study the effect of different parameters: speed, preload, eccentricity and angular misalignment. A theoretical loss model was performed which gave discrepant results for studies with mixed lubrication, making a comparison with the experiments impracticable. Further work has to be done to evaluate the applicability of the proposed model. In addition, a valuable Matlab tool has been developed and packaged to facilitate processing data of the test rig for future users.
Keywords
Mechanical Face Seal, Friction, Test Rig, Lubrication Regimes
Sammanfattning
Mekaniska glidringstätningar är viktiga maskinkomponenter inom tung fordonsindustri, men de orsakar höga friktionsförluster. För att sänka dessa förluster behöver först en undersökning utföras för att få förståelse för friktionsbeteendet hos de mekaniska glidringstätningarna. En fysisk testrigg gör det möjligt att utvärdera befintliga tätningar och jämföra deras prestanda med nya koncept. Denna rapport är en uppföljning på ett koncept av en testrigg för mekaniska för glidringstätningar som en grupp studenter på KTH tidigare tagit fram. Innan detta projekt hade en testrigg för konceptet delvis utformats och tillverkats.
Fortsättningen av detta arbete innebar att en heltäckande bakgrundsundersökning utfördes samt även en granskning av teoretiska modeller för att karakterisera tätningarna. Konstruktion och tillverkning av den tidigare testriggen slutfördes. Dessutom validerades testriggen vilket visade att dess repeterbarhet var accepterbar. Tester planerades och utfördes för att studera effekten av följande parametrar: hastighet, kast, excentricitet och snedställning.
En teoretisk förlustmodell utvecklades men visade sig ge avvikande resultat för blandsmörjning vilket gjorde jämförelser med experimenten ogenomförbara.
Ytterligare arbete bör göras för att utvärdera tillämpligheten av den framtagna modellen. Utöver detta har även ett användargränssnitt i Matlab utvecklats och paketerats för att underlätta för framtida användare att bearbeta data från testriggen.
Nyckelord
Mekanisk glidringstätning, Friktion, Provningsbänk, Smörjningsregimer
Acknowledgements
This master thesis project has been carried out at the Machine Design Department in KTH in collaboration with Volvo CE. The cooperation between both KTH and Volvo CE has been excellent, so we would like to thank both parts which made this project possible. Especially, we would like to thank our supervisors Henrik Strand and Johan Abrahamsson at Volvo CE and Stefan Björklund at KTH. It has been a privilege to learn from you.
We would also like to thank all who helped us at KTH. A special thanks to Tomas Östberg, whose help with manufacturing and assembling was extremely important; Staffan Qvarnström, who helped us with the soldering of some sensor connections; Jonny Gustafsson from the Production Department in KTH, who helped us with the surface measuring instrument; Ulf Olofsson, who gave us advice on our loss model and test plan; Andreas Almqvist, professor at Luleå University of Technology, whose great expertise in tribology guided us in the right direction to perform the theoretical model; Anders Pettersson also from Luleå University of Technology, who gave us another point of view; Ulf Sellgren, examiner of the project, whose feedback from the Project Planning seminar was very beneficial; friends who have been supporting us during the whole process.
Finally to our parents, without their help and support this would have never been possible.
Authors
Àlex de la Cruz Gargallo <alexdlcg@kth.se>
Víctor Pérez Palomino <victorpp@kth.se>
Engineering Design - Machine Design KTH Royal Institute of Technology
Place for Project
Stockholm, Sweden Brinellvägen 85
Examiner
Ulf Sellgren Brinellvägen 83
KTH Royal Institute of Technology
Academic Supervisor
Stefan Björklund Brinellvägen 83
KTH Royal Institute of Technology
Industrial Supervisor
Henrik Strand and Johan Abrahamsson Eskilstuna, Sweden
Volvo Construction Equipment AB
Contents
1 Introduction 1
1.1 Background . . . . 1
1.2 Purpose . . . . 2
1.3 Delimitations . . . . 2
1.4 Methodology . . . . 3
2 Frame of reference 6 2.1 Mechanical face seals (MFS) . . . . 6
2.1.1 Definition . . . . 6
2.1.2 Types . . . . 8
2.1.3 Applications . . . . 11
2.2 Physical phenomena in MFS . . . 12
2.2.1 Lubrication . . . 12
2.2.2 Friction and wear . . . 14
2.2.3 Leakage . . . 16
2.3 Materials in MFS . . . 16
2.3.1 Physical and Mechanical Properties . . . . 17
2.3.2 Tribological Properties . . . . 17
2.3.3 Chemical Properties . . . 18
2.3.4 Common materials . . . 19
2.4 Test rigs . . . 21
2.4.1 Input variables . . . 21
2.4.2 Output variables . . . 22
2.4.3 Validation . . . 24
2.5 Loss models . . . 26
3 Test rig 29 3.1 Design . . . 29
3.1.1 Static system . . . 30
3.1.2 Dynamic system . . . 30
3.2 Mechanical parts . . . . 31
3.2.1 Dynamic bearing unit . . . . 31
3.2.2 Seal housings . . . 32
3.2.3 Dynamic shaft . . . 32
3.2.4 Static shaft adapter . . . 32
3.2.5 Base . . . 33
3.3 Measurement system . . . 34
3.3.1 Temperature sensor . . . 34
3.3.2 Force sensor . . . 36
3.3.3 Torque sensor . . . 37
3.3.4 Pressure sensor . . . 39
3.4 Safety . . . 40
3.4.1 Safety switch . . . 40
3.4.2 Emergency button . . . 42
3.5 Assembly . . . 42
3.5.1 Order of assembly . . . 42
3.5.2 Shaft alignment . . . 43
3.5.3 Grounding issues . . . 45
3.6 Test Plan . . . 45
3.6.1 Validation Test / Speed & Preload Test . . . 46
3.6.2 Eccentricity Test . . . 47
3.6.3 Angular Misalignment Test . . . 47
4 Theoretical loss model 48 4.1 Operating lubrication regime . . . 48
4.2 Film thickness . . . 49
4.3 Friction forces . . . . 51
4.4 Losses . . . . 51
5 Visualization and process data tools 53 5.1 EVIDAS . . . 53
5.2 GUI Matlab tool . . . 53
5.2.1 Visualize data . . . 53
5.2.2 Create fG curve for Loss Model . . . 54
6 Results 56
6.1 Test Rig . . . 56
6.1.1 Validation Test . . . 56
6.1.2 Speed and preload study . . . 60
6.1.3 Eccentricity Test . . . 63
6.1.4 Angular Misalignment Test . . . 64
6.1.5 Distance - Preload Test . . . 66
6.1.6 Surface measurements . . . 67
6.2 Loss Model . . . 69
7 Conclusions 70 7.1 Discussion . . . . 71
7.2 Future Work . . . . 71
References 73
1 Introduction
In this section a general introduction of the project is provided. The purpose and delimitations of the study and the methodology to follow in order to achieve the main goals are briefly discussed below.
1.1 Background
Mechanical face seals are often used in rotating applications in extremely arduous conditions. It is an important machine element in the axles of heavy-duty vehicles.
Their main function is to prevent leakage and to keep contaminants from entering the axles. All of them have in common that one half of the seal is rotating and the other is stationary. The separation between both halves has to be small to minimize leakage of the lubricant. Therefore, an increase of friction and wear occurs, which results in power loss and reduced life time.
A lot of products from Volvo CE include mechanical face seals. Some data is provided by the suppliers of the seals, but the engineers at Volvo CE working with this important machine component believe that further information about the frictional losses in the seals will be very valuable. If the frictional power loss can be lowered the speed can be increased and bigger seals used.
A test rig would allow Volvo CE to evaluate face seals being currently used, compare new concepts with existing ones and compare face seals from different suppliers. With the test rig, Volvo CE can obtain valuable information for selecting seals for their products and generating new concepts. The company’s cost can be reduced and efficiency increased.
This project was started in January 2018 - December 2018 by a group of students at KTH within the Machine Design Advanced Course. A test rig had been partially designed and manufactured to investigate the frictional behavior of mechanical face seals. This thesis work undertakes the continuation of the project.
1.2 Purpose
The main purpose of the thesis work was to investigate the frictional behavior of mechanical face seals. For that, a functional test rig was to be built and validated.
First of all, the design and manufacturing of the test rig needed to be completed and the performance of the machine evaluated. In addition, a loss model for mechanical face seals was to be developed and validated with frictional loss tests in the test rig.
To summarize, the main deliverables of the thesis work are listed below:
• Background study on mechanical face seals.
• Functional test rig for mechanical face seals.
• Theoretical loss model.
• Concept design and ideas suggestion to improve mechanical face seals performance.
1.3 Delimitations
The time frame of the project has been limited January 2019 until June 2019. A Gantt chart of the timeline of the project is shown in Appendix A.
From the very beginning, concept design for improvement of MFS performance was placed in the backlog, and it would only be accomplished in the case there was time left.
The influence of environment contaminants is not tested. The seal chamber is not pressurized, it is filled with the oil without any pump or regulating system.
The prototype had the following limitations:
• Seal diameter ≤ 500 mm
• Motor velocity ≤ 4000 rpm
• Actuator force ≤ 10000 N
1.4 Methodology
To achieve the objectives of this project, a method was defined. Firstly, a background research was conducted, where papers, patents, projects and books were studied and analyzed in order to get a deep understanding of mechanical face seals, tests performed on them and other related topics. After that, the test rig was fully designed, assembled and calibrated. At the same time, a loss frictional model of MFS was developed, which would be validated with the experiments carried out in the test rig. At the end, ideas to improve MFS and its performance would be generated if there was time available.
The structure of the methodology is shown in Figure 1.1.
Figure 1.1: Methodology flowchart.
On the other hand, a Work Breakdown Structure (WBS) was performed (Figure 1.2), where the project was broken down into all the deliverables (first row of the diagram) and their corresponding work packages needed to achieve the completion of each deliverable.
Figure 1.2: Work Breakdown Structure.
In addition, to carry out this project successfully and in an organized way, two other tools were used.
A Gantt chart has been developed to define the different working blocks and their subtasks, as well as their duration and deadlines. It was constantly updated, and allowed us to permanently have a general overview of the project time plan, so that the long-term goals were not overlooked. It is displayed in Appendix A.
Kanban method was used on our day-to-day work. It is a visual method of organization, where there are four columns “Backlog, To Do, Doing and Done”, and all the tasks to be done in the project are written in post-it and placed in Backlog. Every agreed period of time (usually weekly), some post-it are moved to the To Do column. During the period those post-it will go from To Do column to Done, through Doing where there should be as few post-it as possible. This allowed to organize the short-term goals and be more efficient, thereby, dispersion with many open topics was avoided.
Figure 1.3: Kanban board.
Moreover, weekly meetings of approximately 1-hour duration were scheduled with industrial supervisors. In these meetings, the progress done every week was presented, open/new issues were discussed and the plan was reviewed. In addition, discussions were also held occasionally with the academic supervisor to receive feedback and advice.
2 Frame of reference
The frame of reference of the project includes the definition of all the available types of mechanical face seals, the materials used and their applications. The physical phenomena that characterize them when operating and the test rigs and technology available to test them is also discussed in this section. The existing theoretical knowledge of models to predict and simulate the performance of the mechanical face seals is also of interest.
2.1 Mechanical face seals (MFS)
In this section mechanical face seals are defined. Different existing types and their applications are presented.
2.1.1 Definition
Mechanical seals are designed for machinery applications in arduous environments in which conventional seals lack the required durability.
MFS generally have two seal rings and two elastomers mounted in independent housings (Figure 2.1). One of the two halves is stationary, while the other one rotates. The two seal rings mate at some annular surface. The separation between both halves has to be small to minimize leakage. Therefore, an increase of friction and wear occurs, which results in power loss and reduced life time.
Figure 2.1: Mechanical face seal parts [35].
The seal rings are preloaded by the elastomers, closing the gap between the seal faces. The elastomers also transmit the friction torque to the housing components and provide static sealing between each seal ring and its housing bore.
Both the rotating and the stationary seal rings are flexibly mounted to allow axial movement, misalignment and run-out to some degree. Both seal rings, when placed together, form a cone-shaped gap that opens out from the seal faces towards the centre axis. This feature enables easy admission of the lubricant to the seal faces. Also, when wear occurs, the seal face will continuously shift towards the centre axis, providing wear reserves that would last until the inside diameter is reached. [31]
Main advantages of mechanical face seals are:
• Simple and reliable design.
• High sealing ability in arduous environments i.e. dirt, dust, water and abrasive media from the outside and against oil and grease from the inside.
• Long service life.
• Self-centering to compensate for shaft run-out or misalignment.
• Maintenance-free.
• Easy assembly.
2.1.2 Types
Mechanical face seals have several applications. Depending upon the purpose and environment where the seal will be working, it will need different geometry and configuration.
In terms of geometry, mechanical face seals can be classified in the following categories:
• DO - O ring:
This type of seals consists of two tapered seal rings and two o-rings, which are elastomeric rings with round section. The o-rings are compressed between the tapered contact surfaces of the seal ring and the housing.
Figure 2.2: O-Ring mechanical face seal geometry.
• DF - Trapezoid:
This kind of seal is made of two metallic angular seal rings and two trapezoid elastomers. The elastomer parts are positioned against the particular housing bore.
Its housing is easier to manufacture and the seal more simple to install than the O-ring.
The spring characteristic of this combination is usually stiffer, hence allows less axial movement and tolerance errors than other geometries.
Figure 2.3: Trapezoid mechanical face seal geometry [31].
This kind of seals may have some problems. A modification of the design is implemented with the addition of retaining lips (Figure 2.5). It helps to prevent axial displacement caused by dirt-encrusted elastomer parts, and prevents the metal+elastomer spinning at higher break-away and friction torque. Moreover, the seal rings have slots to allow the elastomer parts to grip into the retaining lip.
Figure 2.4: Normal trapezoid seal [31].
Figure 2.5: Trapezoid retaining lip [31].
These kind of geometries can also be used in different ways. The components can be arranged in the seal system following different configurations.
The sealing rings can be symmetrically tapered (Figure 2.6), but can also be
inverted (Figure 2.7) with one of the elastomers facing the shaft and the other one facing the outside.
Figure 2.6: Symmetrical seal [31]. Figure 2.7: Asymmetrical seal [31].
Furthermore, for applications with strong abrasive environment, adapter rings are placed between the elastomer and the housing to avoid significant retaining lip wear on the housing contour (Figure 2.8).
Figure 2.8: Seal with adapter rings [31].
Although the geometries and configurations above are the standards and most used, different and special mechanical face seals can be designed in order to meet particular requirements for unusual or specific applications.
2.1.3 Applications
Mechanical face seals are used in many different areas. They are found wherever the task to carry out gets really tough. Usually operating in rough environment with abrasive media such as dust, sand, mud or under water.
One of the most common fields of application of mechanical face seals is mining, where the conditions are severe. Dust, mud, pressure and other elements require the use of mechanical face seals to keep the components safe. In this area, they can be found in the following components and systems:
• Tracked vehicles (including bulldozers and excavators)
• Heavy trucks
• Conveyor systems
• Axles
• Heavy duty machinery in agriculture and mining
• Oil-cooled disc brakes
• Armoured face conveyors
• Tunnelling machines
• Tension sprockets
• Suspension arm
Mechanical face seals are also an essential component when talking about construction, which has similar requirements as mining. They are used in these applications:
• Hub seals
• Oil Cooled Disc Brakes
• Track Rollers
• Final Drive Systems
• Hydrostatic transmissions
Besides, mechanical face seals can be also found in military applications such as the following:
• Final Drive Gearboxes
• Suspension Arms
• Track Rollers
• Road wheels
• Tension Sprockets
2.2 Physical phenomena in MFS
The performance of mechanical face seals is governed by the rules of some physical phenomena. In this sections the main factors of their behavior when operating are described.
2.2.1 Lubrication
Lubrication is indispensable in mechanical face seals. It has two important functions: reducing friction between the seal faces and cooling the seal and its environment i.e. shaft and housing.
Generally oil lubrication is used since it gives better results than grease regarding wear and service life. Grease lubrication is suitable in cases of low speeds only. Transmission oils, grade SAE 80 and SAE 90 are preferred for typical face seals.
When the correct amount of oil lubricant is supplied, the oil will generally suffice for the lifetime of the seal and the face seal requires no additional maintenance.
[35], [31].
Lubrication regimes
Friction and wear behavior of MFS is governed by the lubricating conditions.
Three lubrication regimes can be distinguished:
• Full film or Hydrodynamic Lubrication regime:
Is the case when the surfaces are fully separated by a fluid film due to pressure build-up, which is caused by rotation of the faces. The pressure in the lubrication film is sufficient to carry the load and there is no mechanical contact. This means that wear can be completely avoided. Friction is limited to viscous shearing in the film and friction forces become very low (μ=0.001- 0.1). Therefore, machines have much less energy losses and are subjected to a smaller thermal load. [32]
• Boundary Lubrication regime:
Is the case when most part of the load is transmitted by mechanical contact.
Usually occurs at low velocities, when the hydrodynamic pressure build-up between the surfaces is negligible. [29]
When boundary lubrication occurs in high speed seals, frictional heating and wear increase significantly and seals cannot survive for long duration. [11]
• Mixed Lubrication regime:
This is the most common case for many seals. A great fraction of the total load is transmitted by fluid pressure, but also by mechanical contact. The flow field in the film depends not only on the face profile, but also on the surface roughness. [15]
Research in mixed lubrication is not very advanced since experimental work is extremely complicated [8]. Thus, numerical studies are more usual.
The study of mixed lubrication is not only relevant for the case of face seals.
There is a general trend in technology to apply increasingly thinner films and in this respect, mixed-lubrication regime will gain in importance [29]. The so-called texturised surfaces, where specific surface patterns are applied, might have a very positive impact in this regard [12] [18].
In consideration of the foregoing, boundary lubrication would be the most suitable regime to minimize leakage, but friction and wear are high in this case. Face seals often have to operate at high pressures and speeds, so a sufficient fluid pressure in the sealing interface is essential to avoid excessive wear, friction and temperature
rise and consequently, guarantee a long seal life [33]. The optimum situation for mechanical face seals is the transition region from hydrodynamic to mixed lubrication. In this situation, friction and wear are low and the film thickness can be kept relatively small to avoid leakage. In this respect, H. Lubinge found in [33] that the transition from hydrodynamic to mixed lubrication regimes mainly depends on the coning angle, the load and roughness of the surfaces.
Stribeck curve
In the Stribeck curve, the relation between friction and the relative velocities of the surfaces is observed and the different lubrication regimes are distinguished.
Figure 2.9: Stribeck curve of a journal bearing [29].
The velocity is usually plotted on a logarithmic scale since the boundary regime is very small. At low velocities, Boundary Lubrication (BL) regime is found.
Then, friction generally decreases with pressure build-up in the lubricant. In hydrodynamic lubrication viscous friction increases with increasing speeds, although it keeps very small values. Regarding the film thickness, it decreases when velocity is reduced, until the surfaces come into contact.
2.2.2 Friction and wear
Friction and wear should be kept low to ensure a long duration of the seal and a low power loss of the machine. However, as mentioned before, friction is compromised by the necessity of avoiding leakage. Hence, it is important to
find a situation in which the best combination between leakage and friction is acquired.
H. Lubinge in [33] performed friction experiments to determine the transition from hydrodynamic to mixed lubrication, where the optimum situation regarding leakage, friction and wear exists for mechanical face seals. He also verified with these experiments a theoretical friction model based on the combination of a contact model and a film thickness equation.
Friction under full film lubricated conditions is mainly caused by shear of the lubricant in the contact as a consequence of the relative movement of the mating faces. Since the shear stresses in the lubricant are relatively low, the lubricant behavior can be assumed as Newtonian. Thus, the shear stress can be expressed as:
τH = η ˙γ = ηUseal
h , (1)
where, η is the dynamic viscosity and ˙γ is the shear rate. In [33] the friction force caused by shear of the lubricant is defined as:
Ff =
∫∫
AH
τHdAH =
∫∫
AH
ηωr
h dAH, (2)
with AH the contact area of the hydrodynamic component, Useal = ωr and h the film thickness.
In the boundary lubrication regime, friction is determined by the shear strength of the protective boundary layers at the surface and in [33] it is defined as:
Ff =
∑N i=1
∫∫
Aci
τcidAci, (3)
where N is the number of asperities in the contact, Aci is the area of contact of the single asperity i and τci is the shear stress at the asperity contact i.
Under mixed lubrication conditions friction is influenced by all the previous factors, and can be calculated as the sum of the friction force at the interacting asperities and the shear force of the hydrodynamic component:
Ff =
∑N i=1
∫∫
Aci
τcidAci +
∫∫
AH
τHdAH (4)
Surface texturing has been proved as an efficient technique to reduce friction and improve the performance of mechanical face seals [3], [7].
2.2.3 Leakage
The leakage rate is an important aspect in the performance of a seal. The leakage in face seals is mainly determined by the average gap between the seal faces.
As a matter of fact, if the film thickness over the entire seal is known, leakage may be estimated. [32] If leakage is low, the seal is most likely operating in the mixed lubrication regime. In this case, the value of the combined mean-to-peak roughness heights is in the same order of magnitude as the fluid film thickness and leakage may be estimated using this value. In the case of full fluid lubrication, leakage is larger and can be estimated in the same way. If leakage is high, it is probable that in some regions the seal gap is several times the average peak roughness height. In addition to this, seal leakage is also affected by the roughness orientation of the mating faces [15].
Pustan et. al [19] carried out experiments where the variation in leakage with time in mechanical seals was observed. He found that in the beginning of the running-in period, high values of leakage are obtained and then leakage achieves a stabilized value. He also found that in addition to the thickness of the sealing gap, leakage is also influenced by the pressure of the sealed fluid and by the rotational speed of the rotor in a mechanical seal.
A very rough estimation of an acceptable leakage in a mechanical face seal is below milimeters3/h.
2.3 Materials in MFS
Mechanical face seals materials are selected based on different criteria. A lot of properties are taken into consideration when materials are chosen, but the
most important ones are described in [32], where they classify them in three categories:
2.3.1 Physical and Mechanical Properties
• Tensile Strength: It is important to both thermal and mechanical shock resistance.
• Compressive Strength: Usually much higher than tensile strength, but in some carbon materials it might be a limiting factor.
• Modulus of Elasticity: It determines how seal faces behave together under load. Materials with low modulus of elasticity easily lead to deformations in undesired and unexpected ways, but increase the thermal shock resistance.
Moreover, tight fits and thermal loads can over-stress materials with high modulus.
• Poisson’s Ratio: It is needed to calculate many of the stresses and deflections for seal rings.
• Hardness: Tribological performance of the MFS highly depends on the hardness of the materials.
• Coefficient of Thermal Expansion: It is significant in relation to the fit of parts in the presence of temperature changes. It is an essential parameter for thermal shock resistance.
2.3.2 Tribological Properties
In addition to the physical and mechanical properties, there are some other material properties affecting the tribological behaviour:
• Friction coefficient: In [32] they claim that the difference in contact friction coefficient is significant between different materials. This fact supports the idea of considering the contact friction coefficient as a material property.
• Wear rate: There is not much information and correlation between materials, since the data available may not repeat from seal design to
seal design because the condition of lubrication may be totally different.
However, it is noted that beyond some critical PV-limit, the wear rate will dramatically increase.
• PV (pressure times velocity) limit: The highest combination of load and speed under which the material can operate properly will be definitely critical for the tribological behaviour, since beyond that limit both wear rate and friction will increase.
In [14] they list some values of PV to give a general indication of the variation between materials.
Face Counterface (PV)max
SiC Carbon 20
SiC SiC 20
WC Resin-carbon 8
WC WC 4
“Stellite” Metal-carbon 3
“Ni-Resist” Carbon 3
Alumina Carbon 3
Lead-bronze Metal-carbon 2 Stainless steel Metal-carbon 1
Table 2.1: Approximate PV limit values (MPa m/s) for water at 40°C [14].
More information about friction and wear can be found in 2.2.2.
2.3.3 Chemical Properties
Chemical resistance is also a decisive and essential parameter to be taken into account when choosing material for mechanical face seals. The purpose of this components is to work in harsh and arduous environments, where they will be in contact with many different chemical elements that will threat their integrity and performance. Therefore, it is critical to determine the material of each part of the seal according to the external factors which it will be exposed to.
2.3.4 Common materials
Over time, materials used for mechanical face seals have been changing, but current suppliers use similar materials between them to design both seal ring and O-ring.
Seal ring materials
Seal ring material has two main requirements, among others: wear and corrosion resistance. Suppliers have developed different solutions for this component:
• In [31] they offer cast iron as a base of this component, but with different variations. They use other alloying components to increase the corrosion and abrasive wear resistance. Inclusions of fine graphite are made to achieve excellent tribological characteristics.
• In [35] hardness of the face material is shown as most important parameter.
They offer different alloys with their specific hardness (Table 2.2).
Material Chemical Composition
Bearing Steel Material: 100Cr6
Ni Hard Alloy Material:
Chromium Base Alloy Material:
C 0.95 - 1.10 2.3 - 3.6 3.4 - 3.8
Cr 1.4 - 1.6 1.2 - 2.0 15.00 - 19.00
Ni - 2.0 - 5.0 -
Fe Balance Balance Balance
Hardness 63 +/- 3 HRC 60 +/- 4 HRC 66 +/- 3 HRC Table 2.2: Hardness for alloys from [35].
• In [30] also some alloys are categorized and evaluated according to wear, corrosion and scoring resistance:
C6 Stellite NiHard Formed Forged
Material Nickel-Alloy Iron-Alloy Iron-Alloy SAE 1074 SAE 52100
Process Cast Cast Cast Stamped Forged
Wear life High High Low/Medium Low Low
Corrosion
Resistance High Medium/High Low/Medium Low Scoring
Resistance High Low Medium/High Low Low
Table 2.3: Alloys and properties in [30].
O-ring materials
When it comes to the O-ring material, high thermal resistance and low compression set are the essential criteria for the elastomer to be fulfilled.
• In [31] they use mainly four compositions, with degrees of hardness:
– Nitrile-butadiene-rubber (NBR): meets the standard requirements.
– Hydrogenated nitrile-butadiene-rubber (HNBR): For applications with high thermal stresses.
– Silicone rubber (VMQ): For applications with high thermal stresses.
– Fluorinated rubber (FPM): For applications with high thermal stresses.
• In [35] similar compounds are offered, which are summarized in Table 2.4.
Type Temperature Range °C
Hardness Range available SH
Mechanical Resistance
Oil
Suitability
NBR -25 to 100 50 to 70 Good Good
NBR LT -50 to 100 60 to 70 Good Good
HNBR -30 to 150 60 to 65 Very Good Good
HNBR LT -40 to 150 65 Very Good Good
FKM -15 to 200 65 Good Good
VMQY -50 to 175 65 Quite Good No EP
Table 2.4: O-ring materials from Trelleborg[35].
• [30] offers O-rings made of Nitrile and Silicone as the most common, but they also have fluoroelastomer (FKM) and Hydrogenated Nitrile (HNBR) for more specialized applications.
It can be seen that all the suppliers use similar materials to achieve the required conditions in the O-ring component. Nitrile compounds are a common denominator between them.
2.4 Test rigs
Over the years many test rigs have been developed to characterize and try to know better how the mechanical face seals work and which are their limitations.
Predicting how they fail and why, has always been a challenge to the research community.
Regardless of what parameter or phenomena they are looking into, mechanical face seals test rigs usually have similar inputs in common.
2.4.1 Input variables
It is needed to have an accurate control of the velocity of the rotating ring. A motor and its driver are in charge of that task.
To ensure that the conditions during the test runs are as similar as possible to the actual application environment, the pressure inside the seal housing needs to be regulated. Most of the test rigs use water vapor with a control system to do so [1] [4]. Using steam to pressurize the chamber may induce to a two-phase phenomena, where vapor and liquid share the contact surface between the rings, which increases the frictional torque as it is shown in [13]. However, for the applications relevant to this thesis, pressurized oil would be more suitable.
Misalignment in the seal configuration is another variable to control when testing this components. A wrong alignment of the faces may have different effects, and lead to several kinds of dysfunctionalities. There are some ways to create this disturbance in the system, such as using shims or adapter plates with screws.
Figure 2.10: Misalignment used in [4].
S. Goilkar and H. Hirani claim that the misalignment is the root cause of seal failure, ahead of other factors like heat or vibration [1].
Another element to control when testing mechanical face seals is the preload. The force applied between the contact faces of the seals must be defined and monitored in order to properly characterize their frictional behaviour. T. Wang et al. used a spring-loaded device applied to the bottom of the stationary ring to supply the contact force to the system [13], but other mechanisms can be used for the same purpose, for instance a linear actuator, which would be more accurate.
2.4.2 Output variables
Depending on the purpose of the study, the test rig will be equipped with different sensors or devices to get information from the mechanical face seals and its performance.
Frictional torque is a common denominator in all test rigs, since the frictional behaviour and the performance of the seal clearly relies on it. It is usually recorded by a force sensor placed at a certain distance of the shaft, attached to the static-side of the seal.
The main function of the mechanical face seals is to prevent leakage from inside, and outer dust to come in. Thus, this is one of the most important outputs of this type of test rig. However, it is difficult to detect this phenomena automatically, and all the test rigs available use visual inspection to determine if there is leakage or not.
When studying failure in mechanical face seals, temperature is a significant aspect
to observe. Most of the test rigs track the temperature in the stationary ring, usually using thermocouples or thermal resistors attached to the seal or close to it. But temperature in the dynamic ring is barely considered, since it is more difficult to access to it. However, some researchers used wireless solutions to make it possible. In [2] M.C. Valigi et al. combined a thermocouple with a radio device to transfer the data, and fastened it to the rotating ring so they would turn together. They found that the temperature in the rotating ring is always lower than the temperature in the static ring, due to forced convection.
Figure 2.11: Thermocouple and radio device used in [2].
On the other hand, Lokesh A. Gupta et al. used permanent magnets sensed remotely by a Hall effect sensor. They validated the method and they were able to get the temperature of the rotating ring through the magnetic field variation [9].
In [13] T. Wang et al. also study the phase change and distribution of the fluid in the contact surface of the seal. Using a camera they are able to picture the face of the rotating ring through the static ring, which is made of quartz. They proved that the friction and its effects increase in areas where vapor prevails over liquid.
Figure 2.12: Two phases in contact surface from [13].
2.4.3 Validation
To ensure that the results and measurements of the test rig are reliable, a validation is performed. Studies of repeatability and reproducibility are standards to validate measuring systems.
Reproducibility, which is the variation of the data due to the measuring system or the variation coming from different operators measuring the same part using the same device. In case there was only one operator running the experiments, this factor would not be studied.
A study of repeatability analyzes the variation of the data due to the measuring device, i.e. the test rig, or the variation coming from when the same operator uses the same measuring device measuring the same part repeatedly.
Gage R&R (Repeatability and reproducibility) analysis is one of the most accepted methods to assess the quality of the measuring system.
The study calculates variances and standard deviations for:
• The effect of the measuring device itself (repeatability).
• The effect of the operator and process (reproducibility, 0 in this case).
• The effect of both repeatability and reproducibility (gage R&R).
• The effect caused by the natural difference between parts.
Another calculation is made for all the error sources: the study variation. It is defined as the assumed percentage of error, and is calculated as follows:
StudyV ariation = 5.15(StandardDeviation) (5)
Assuming a normal distribution, approximately 99.73% of the data fall within 6 standard deviations (± 3 standard deviations of the mean), but most commonly used, approximately 99% of the data fall within 5.15 standard deviations (± 2.575 of the mean) [27].
The standard criteria to evaluate and assess the results of the study is summarized in [34]. Table 2.5 shows how to interpret the data obtained from the analysis.
% Variance % Study Variation Conclusion
X <1% X <10% The system is capable 1% <X <9% 10% <X <30% The system is acceptable.
Might need revision depending on the application.
9% <X 30% <X The system is not capable. Should be improved.
Table 2.5: Gage R&R analysis criteria.
One last indicator for the evaluation of the system is extracted from the study:
number of distinct categories. It represents the number of groups that the system can differentiate from the data itself.
The Automotive Industry Action Group [28] has standard guidelines on how to assess and analyze this parameter, summarized in Table 2.6.
Number of distinct categories Conclusion
X <2 The system is not capable. Should be improved.
2 <X <5 The system is acceptable.
Might need revision depending on the application.
5 <X The system is capable Table 2.6: Gage R&R analysis criteria.
2.5 Loss models
There is an extensive research on models for mechanical face seals to predict the various phenomena occurring around this mechanical component.
Many investigators have proposed models for fully hydrodynamic or fully hydrostatic seals. However, mechanical face seals are often operating in mixed lubrication regime, where the surface roughness must be taken into account, since contact between the seal faces may occur. Nevertheless, the published research into mixed lubrication is very little. Besides, experimental work is very complicated to carry out, so numerical studies are usually preferred.
When mixed lubrication is considered, some fraction of the load is carried by the asperity contact between the seal faces. In addition some part of the load is supported by hydrostatic and hydrodynamic pressures.
The roughness induced load support is due to the combined effect of height variations in the circumferential direction and local micro cavitation eliminating negative pressure values [25]. Most of the authors do not add the effects of the contact load support into their models for simplicity. However, there are a few who take this into account, as discussed below.
In [17] two possible approaches are suggested to study the contact pressure, which will determine the film thickness when contact occurs. The first is a Gaussian plastic model where the roughnesses of the two faces in contact are combined to just one surface and it is assumed that the stress develops on the asperities whose height from the mean line exceeds the film thickness. The second approach is an elastic contact model, the classic Greenwood-Williamson model [26], based on a distribution of peak heights. The two approaches were compared in [17] and it was found that there was a significant difference between them. The elastic model predicts a larger film thickness and leakage under heavily loaded operating conditions. The elastic model is usually preferred, but much information is needed to apply it, and real surface measurements should be done. Noël Brunetier in [23] develops an analytical model in which the contact force is calculated using a simplified Greenwood and Williamson model. In this model, he suggests to approximate the various characteristics of the summits of the asperities by only
using two measurable surface roughness parameters: the standard deviation of the roughness height and the correlation length of the surface.
In addition, there are two more opening forces acting between the seal faces. The hydrostatic contribution due to the sealed fluid pressure and the hydrodynamic, due to relative sliding motion. This contribution might be neglected when using water or low viscosity fluids. However, it is important to consider for oils.
Empirical equations are proposed to estimate the hydrodynamic load support in [23].
There are other numerical flow models of mixed lubrication in MFS as [8], in which cavitation effects and Hertzian asperity contact are considered. In [3] M.
Adjemount et al. present a complex model which also includes a heat transfer and deformation model for the seal rings, introducing the distortions caused by the thermal and mechanical loads. Numerical simulations are used to predict the film thickness.
Temperature in the seals is one of the most researched factor. It produces stresses in the seal, deformations and changes the fluid viscosity. Therefore, it is of major importance for the control and prediction of the seals behavior.
Many researchers have focused on how to measure the temperature in seals. Since the setup of the seals is usually closed and hardly accessible, a way to reach the contact face of the seals, and mainly the rotating ring, has been object of investigation. M.C.Valigi et al. present a in [2] a new telemetry system consisting in a combination of a thermocouple and a wireless device, both attached to the rotating side of the seal, which is able to send the temperature of the dynamic ring through WiFi. More complex methods are used by Lokesh A. Gupta et al. in [9], where temperature-induced magnetic field change in permanent magnets is remotely sensed by a Hall effect sensor. Besides, M. Adjemout et al. introduce infrared thermography to measure the temperature of the rotating ring [3].
On the other hand, there is a lot of theoretical research on heat transfer and temperature distribution. N. Brunetière et al. have developed models for numerical analysis of heat transfer in mechanical face seals using CFD [6], as well as other researchers [21] [22]. Analytical methods are also used as M. Rahimi
Takami et al. do in [5] to investigate the influence of different factors on the temperature distribution.
The effect that the rise of temperature produces in mechanical face seals is also studied by investigators such as Francis E. Kennedy, JR. who carried out an analytic study of thermocracking in [10].
3 Test rig
To obtain experimental information about the performance of mechanical face seals, a test rig has been built to investigate the friction between the two seal faces of MFS of different sizes. In addition, temperature in the vicinity of the inner ring and leakage were tested. The normal load applied, the sliding speed, the containing pressure, the angular misalignment and the eccentricity were controlled parameters.
A preliminary design and selection of parts was done by the students of the previous project, but the components and devices in the following sections have been designed, selected and manufactured in this thesis.
3.1 Design
The test rig built had to meet the requirements in Appendix B. Two significant parts can be distinguished: the static and dynamic sides.
The test rig has two actuators to control and change the input factors and allow the user to perform tests with different operational conditions. An inductive motor supplies torque to the dynamic seal housing, being able to control its rotational speed. On the other side, an electric linear actuator applies an axial force on the static seal housing, creating a closing force between the contacting faces.
All components are resting on a stiff base, which is composed of two parts in order to make it easier to manufacture and transport.
Figure 3.1: Test Rig Design.
3.1.1 Static system
The static part of the test rig (left part in Figure 3.1) holds the face of the seal that does not rotate. The seal housing is attached to two plates which allow to set eccentricity and angular misalignment to the contact. These plates are at the same time attached to the torque hub, which has a lever arm that connects the shaft to the torque sensor. This shaft rests on two bushings that allow axial movement, and is aligned with a linear actuator that pushes it against the other face of the seal to apply the required preload.
The electric linear actuator used is provided by Volvo CE (Thomson DD24-30B5- 16-485) and is rated for a static load of 20 kN, more than sufficient regarding load requirements. It is controlled by a control box, which allows the user to set the direction of the movement and the speed, as well as type of movement: constant or by steps pushing a button.
3.1.2 Dynamic system
The dynamic side (right part in Figure 3.1) holds the face of the seal that rotates.
The seal housing is attached to the dynamic shaft, which rests on the dynamic
housing. To transmit the torque, this shaft is then connected to the shaft of the motor by a safety coupling.
In order to meet the requirements in Appendix B, a 5.3 kW AC inductive motor from ABB has been chosen(ABB M3EB 100B 6, 3GEB103724-HDA). It is able to provide the maximum torque and speed required. The motor is controlled using a drive from ABB as well (ABB ACS 880-01-017A-3+B056+E200+R705).
3.2 Mechanical parts
The following mechanical parts had to be designed, redesigned or adapted in order to fulfill the requirements of the test rig.
3.2.1 Dynamic bearing unit
The dynamic bearing unit that was designed in the previous project has been redesigned. The previous design was impossible to assemble, so the housing, the spacer and the right seal have been redesigned and selected again to ensure an easy and proper assembly and functionality.
Figure 3.2: Dynamic bearing set.
Number Component 1 Dynamic shaft
2 Labyrinth seal 70x90x10 3 Dynamic housing
4 Bearing set
5 Spacer
6 Lock washer
7 Lock nut
Table 3.1: Dynamic bearing set components.
3.2.2 Seal housings
The seal housings have been adapted in order to fit a different type of seal. The previous model was designed to hold asymmetrical seals of smaller diameter.
The seals to analyze in this thesis are symmetrical type of seals of 270 mm outer diameter. Drawings of the housings can be found in Appendix C.
3.2.3 Dynamic shaft
The dynamic shaft has been redesigned and manufactured again. The shaft made by the students from the previous project did not have the required tolerances, therefore the fit between it and the bearing set had clearance instead of interference. The drawing of the new dynamic shaft can be found in Appendix C.
3.2.4 Static shaft adapter
The previous design of the test rig was not able to set the seals in the position where they should be. It was impossible to approximate the static part to the other side of the seal, because the force sensor was colliding with the bushing, as it can be seen in Figure 3.3.
Figure 3.3: Force sensor collision.
In order to solve this problem, an adapter has been designed and manufactured to make the static shaft longer (Figure 3.4).
Figure 3.4: Adapter.
3.2.5 Base
The base was already built in the previous project. However, some modifications had to be done to make the assembly possible and to guarantee a firm mount of the motor. In addition, the base was painted to avoid corrosion and provide an aesthetic look.
The base is composed of two parts made of steel beams, which basically hold the dynamic and static components respectively. Moreover, steel plates are placed on these beams, where all the components are fixed, allowing minimum misalignment. The attachment of these plates to the beams was already a challenge, since the holes of the beams and the plates were not always aligned due to inaccurate manufacturing. Therefore, bigger holes had to be drilled to assembly all the components.
In addition, two steel beams were added on the base to support the motor. The design of the previous students only included two points to support the motor on the base. The beams indicated in Figure 3.5 were welded on the existing base to ensure a secure support of the motor.
Figure 3.5: Added beams to support motor.
3.3 Measurement system
There is a total of four sensors in the test rig. They are used to measure the temperature in the seal, the preload applied by the electrical actuator, the friction torque generated by the seals and the pressure inside the seals.
Each sensor is connected to a combined amplifier and data acquisition (DAQ) unit: Quantum X, from HBM. Thus, all the data is acquired in the same software tool: EVIDAS, from HBM, which will be further discussed in Section 5.1.
The characteristics of each sensor as well as the transducer settings in the data acquisition tool and the calibration procedure will be explained below.
3.3.1 Temperature sensor
• Description and specifications
A type K thermocouple cable from Pentronic was placed in the inner ring of the static seal, which can be accessed from the outside, in between the seal
