Friction Characterization of Turbocharger Bearings

(1)

Friction Characterization of Turbocharger Bearings

ERIK SJÖBERG

Master of Science Thesis Stockholm, Sweden 2013

(2)

(3)

Friction Characterization of Turbocharger Bearings

Erik Sjöberg

Master of Science Thesis MMK 2013:06 MFM 149 KTH Industrial Engineering and Management

Machine Design SE-100 44 STOCKHOLM

(4)

(5)

I

Master of Science Thesis MMK 2013:06 MFM 149

Friction Characterization of Turbocharger Bearings

Erik Sjöberg

Approved

2013-03-07

Examiner

Sergei Glavatskih

Supervisor

Sergei Glavatskih Hans-Erik Ångström Habib Aghaali Matthew Cha

Commissioner

Internal Combustion Engines Division, KTH

Contact person

Hans-Erik Ångström

Abstract

In this study the main objective was to determine the mechanical frictional losses in a standard automotive turbocharger. For future work, the result should be the basis for a formula describing the frictional losses in the turbocharger, dependent on its various operation conditions.

The methodology used to estimate the frictional losses included numerical simulation, analytical calculations and measurement techniques. Isoviscous simulation models for journal bearings and thrust bearing were developed and used to simulate bearing friction. In addition, analytical calculations based on Petroff’s equation were used to determine frictional losses. A third method based on a calorimetric measurement technique was used to determine the frictional losses and compare the results from analytical calculations and simulations. The method involved measurement of the oil flow through the turbocharger and the temperature difference of the turbocharger oil between the inlet and outlet. From the measurement and for a given value of the oil heat capacity the total bearing power loss was determined. The cooling water flow to the turbocharger was adjusted to obtain a bearing housing temperature close to the oil temperature in the bearing housing and so minimize heat transfer between the turbocharger bearing housing and the oil.

The axial force acting on the turbocharger thrust bearing has a great influence on the oil film thickness in the bearing and so also the frictional losses. The axial force was measured and calculated to be used as input for friction simulations and to correlate it to other significant parameters. Three different methods were used to determine the force: axial force calculation with force balance analysis, axial force calculation based on the axial displacement measurement of the turbocharger shaft and axial force measurement with strain gauges.

The thrust bearing simulation model gave an approximation of the bearing frictional losses in the turbocharger. However, the results tend to be overestimated due to the isoviscous assumption

(6)

and the approximation of the oil viscosity based on the assumption that the bearing oil temperature could be estimated as an average of oil temperature at the inlet and outlet.

The measurement of frictional losses was slightly overestimated, which might be a result of invalid heat transfer assumptions and inaccurate measured bearing housing and oil temperature.

In addition to the approximation of the oil viscosity, this explains the deviation between simulated and measured frictional losses.

Finally, it was found that the axial force was strongly correlated to the axial displacement of the turbocharger shaft, the pressure before the turbine and the turbocharger rotational speed. The pressure on the compressor wheel back face is mainly affected by the turbocharger rotational speed and is also a large contributor to the total axial force on the turbocharger shaft.

(7)

III

Examensarbete MMK 2013:06 MFM 149 Redogörelse för lagerfriktion i turboladdare

Erik Sjöberg

Godkänt

2013-03-07

Examinator

Sergei Glavatskih

Handledare

Sergei Glavatskih Hans-Erik Ångström Habib Aghaali Matthew Cha

Uppdragsgivare

Avdelningen

Förbränningsmotorteknik, KTH

Kontaktperson

Hans-Erik Ångström

Sammanfattning

I denna studie var huvudsyftet att bestämma de mekaniska friktionsförlusterna i en standard turboladdare för personbilar. För framtida arbete skulle resultatet ligga till grund för en formel som beskriver friktionsförlusterna i turboaggregatet beroende på dess olika driftpunkter.

I den metod som används för att uppskatta friktionsförlusterna ingick numerisk simulering, analytiska beräkningar och mätteknik. Isoviskösa simuleringsmodeller för radiallager och axiallager utvecklades och användes för att simulera lagerfriktion. Dessutom användes analytiska beräkningar baserade på Petroff’s ekvation för att bestämma friktionsförlusterna. En tredje metod baserad på kalorimetriska mättekniker användes för att bestämma friktionsförlusterna och jämföra resultaten från analytiska beräkningar och simuleringar. Metoden involverade mätning av oljeflödet genom turboladdaren samt temperaturskillnaden av oljan mellan inloppet och utloppet i turboladdaren. Från mätningen och för ett givet värde av oljans värmekapacitet kunde den totala effektförlusten bestämmas. Kylvattenflödet till turboladdaren justerades för att erhålla en lagerhustemperatur nära oljetemperaturen i lagerhuset och på så sätt minimera värmeöverföringen mellan turboladdarens lagerhus och oljan.

Den axiella kraften som verkar på turboladdarens axiallager har en stor inverkan på oljefilmstjockleken i lagret och så även friktionsförlusterna. Den axiella kraften mättes och beräknades för att användas som indata till simuleringarna samt för att korrelera den till andra viktiga parametrar. Tre olika metoder användes för att bestämma kraften: beräkning av axialkraft med kraftbalans, beräkning av axialkraft baserad på mätningen av den axiella förskjutningen av turboladdarens axel samt mätning av axialkraft med trådtöjningsgivare.

Simuleringsmodellen för axiallagret gav en uppskattning av lagerfriktionsförlusterna i turboladdaren. Dock tenderar resultaten till att vara något överskattade på grund av det

(8)

isoviskösa antagandet och approximationen av oljans viskositet baserat på antagandet att oljetemperaturen i lagret kan uppskattas som ett medelvärde av oljetemperaturen vid inloppet och utloppet.

De uppmätta friktionsförlusterna var något överskattade, vilket kan vara ett resultat av ogiltiga antaganden gällande värmeöverföring samt felaktigt uppmätt lagerhus- och oljetemperatur.

Förutom approximationen av oljans viskositet, förklarar detta avvikelsen mellan simulerade och uppmätta friktionsförluster.

Slutligen kunde slutsatsen dras att den axiella kraften var starkt korrelerad till den axiella förskjutningen av turboladdarens axel, trycket före turbinen och turboladdarens rotationshastighet. Trycket på kompressorhjulets baksida påverkas främst av turboladdaren rotationshastighet och är också en stor bidragsgivare till den totala axiella kraften på turboladdarens axel.

(9)

V

ACKNOWLEDGEMENT

This project was initiated by the CCGX, Competence Center for Gas Exchange, which is a part of the Internal Combustion engine division at the Royal Institute of Technology. As author of this master thesis I would like to thank and give my greatest gratitude to all people who has supported and helped me throughout this project.

Many thanks to my supervisor and examiner, Professor Sergei Glavatskih for contribution of professional excellence in the field of tribology.

Professor Hans-Erik Ångström has shown tremendous enthusiasm for this project from the initial phase and supported with technical support likewise ingenious technical solutions.

Special thanks to co-supervisor PhD student Habib Aghaali for support, understanding and patience.

Thanks to co-supervisor PhD student Matthew Cha for help with simulations.

I would also like to thank technician Bengt Aronsson for manufacturing and serving the test equipment.

Finally, thanks to Per-Inge Larsson at Scania CV AB for help with bearing geometry measurement and to Mayte Pach at Svenska Statoil AB from providing us with lubricating oils.

Stockholm March 2013 Erik Sjöberg

(10)

NOMENCLATURE

Symbols, subscripts and abbreviations are defined below.

Symbol Definition Unit

A Area m²

! Diameter m

! Force N

! Length m

! Power W

Ra Surface roughness m^-6

Re Reynolds number -

! Temperature K

U Velocity ms^-1

!_! Heat capacity Jkg^-1K^-1

! Radial clearance in journal bearing m

! Acceleration of gravity ms^-2

ℎ Film thickness m

! Mass kg

! Mass flow kgs^-1

! Pressure Pa

!!_!"#$%!& Pressure on compressor wheel exducer Pa

!!_!"#$%&' Pressure on turbine wheel inducer Pa

!_{!_!"#$} Pressure on compressor wheel back face Pa

!_{!_!"#$} Pressure on turbine wheel back face Pa

! Radius m

!!_!"_!"# Compressor wheel tip radius at inlet m

!!_!"#_!"# Compressor wheel tip radius at outlet m

!!_!"_!"# Turbine wheel tip radius at inlet m

!!_!"#_!"# Turbine wheel tip radius at outlet m

!!_!"#$$%& Turbine wheel scallop radius m

!_!" Compressor backplate seal radius m

!_!" Turbine backplate seal radius m

! Time s

!, !, ! Cartesian coordinates -

! Angular extent of thrust pad rad

! Axial clearance in thrust bearing m

! Angle rad

! Efficiency -

∆!_!"# Oil temperature difference between °C

inlet and outlet

! Film thickness parameter

! Dynamic viscosity Pas

! Kinematic viscosity cSt

! Density kgm^-3

! Partial derivative operator -

! Shear stress Nm^-2

! Angular velocity rads^-1

(11)

VII

Subscript Definition

1 Compressor inlet

2 Compressor outlet

3 Turbine inlet

4 Turbine outlet

! Compressor

!_! Compressor wheel inducer

!_! Compressor wheel exducer

!_! Compressor wheel back face

! Friction

! Inner radius

! Journal bearing

! Outer radius

!"#1 Bearing housing oil inlet

!"#2 Bearing housing oil outlet

! Turbine

!_! Turbine wheel inducer

!_! Turbine wheel exducer

!_! Turbine wheel back face

!ℎ Thrust bearing

!" Turbocharger

!"! Total

! !"# Average oil temperature

Abbreviations

krpm kilo-revolutions per minute

rpm Revolutions per minute

SI Spark ignited

TC Turbocharger

(12)

1 INTRODUCTION

1.1 Background

Turbocharging is a common feature in today’s internal combustion engines and widely used on diesel engines while the usage on gasoline engines is increasing. The mechanical design of the turbocharger is basically a rotor system consisting of turbine and compressor, and a connecting shaft. On small high-speed turbochargers the shaft is supported by a bearing system, usually oil- lubricated floating journal bearings in combination with a thrust bearing [1]. There are simple well-known formulas for calculating the frictional losses in these types of bearings, such as Petroff’s equation [1]. However, these types of formulas cannot be directly introduced for turbocharger bearing systems if reasonable accuracy is desired.

The Internal Combustion Engines division of the Machine Design Department conducts research within turbocharging. Simulation in research projects is often linked to GT-Power [2], a commercial software for engine simulation. When specifying the turbocharger mechanical efficiency in the software, it is normally pre set as a percentage of the turbine power, which is a rough assumption. If replacing the inbuilt friction loss assumption by a friction formula describing the mechanical friction losses dependent on the turbocharger operating conditions, this will give more accurate estimation of the turbocharger efficiency in the simulations.

1.2 Objective

This master thesis project is part of the research at the Internal Combustion Engines division and aims to estimate the frictional losses in the oil-lubricated journal bearings and thrust bearing used in a Garrett GT 1752 turbocharger. The estimation should give a good approximation of the frictional losses with higher accuracy than Petroff’s equation. The result should be the basis for a formula, which describes the mechanical frictional losses in the turbocharger. By using the formula in 1D engine simulation the estimation of the mechanical frictional losses in the turbocharger can be improved and give a more realistic value of the turbocharger efficiency. The estimation of the mechanical frictional losses should be based on bearing friction simulation and bearing friction measurement techniques, where the simulated results l should be compared with the empirical calculations. Following parameters mentioned below are aimed at calculating:

• Pressure distribution on the front side of the compressor and turbine wheels.

• Pressure distribution on compressor and turbine wheel back faces.

• Axial- and radial forces on the turbocharger shaft.

• Mechanical frictional loss in the turbocharger more accurate than Petroff’s equation.

• The effect of changing oil temperature and so viscosity.

• The effect of changing oil pressure.

• Finally discuss whether turbocharger friction characteristics are desirable, in order to separate them from aerodynamic losses.

1.3 Delimitations

To specify the frame of this thesis several delimitations were taken into consideration:

• Frictional losses originated from wet oil cavity and form sealings isolating the bearing housing were not taken into account.

• The engine providing the turbocharger with exhaust gas pulses was only used to control the turbocharger operating conditions. Therefore, input engine operating parameters affecting the turbocharger characteristics were not given any further analysis.

• Only steady state load points were studied.

(15)

2

1.4 Methodology

Initially, an extensive literature survey was presented. The survey included previous similar investigations as well as studies of different turbocharger bearing systems, theory of hydrodynamic lubrication, calculations of axial force balance in turbochargers, measurement methods of bearing friction etc.

Two different approaches were used to estimate the mechanical frictional losses in the turbocharger; calorimetric measurement technique and bearing friction simulation. The bearing friction measurements formed the basis of input parameters for the bearing friction simulation considering parameters such as axial force, turbocharger rotational speed and oil viscosity.

Furthermore, measured results were compared with results obtained from the the simulations.

In conjunction with the design of experiment it was decided to design and manufacture separate oil lubrication system for the turbocharger, referred to as oil rig. The rig enabled separate oil lubrication for the turbocharger independent of the engine and the engine oil lubrication system.

In addition, inlet oil temperature and pressure could be varied and controlled.

The measurements of turbocharger bearing friction were performed on an engine in the KTH engine test facility. From this measurement, bearing friction could be calculated using the calorimetric measurement.

Journal bearing and thrust bearing isoviscous friction models were created in COMSOL [3], a multi physics simulation software. In the models, measured bearing geometries were introduced and input parameters such as axial force, turbocharger rotational speed and oil viscosity were used. Measured bearing friction and simulated bearing friction results were compared and evaluated.

(16)

2 FRAME OF REFERENCE

In this section fundamentals of turbocharging and turbocharger bearings are described.

Furthermore, friction simulation models and friction measurement methods for turbocharger bearings are studied.

2.1 Turbocharging

A turbocharger is basically a turbine-driven compressor. Compressors are used to increase the cylinder content by means of air, prior to combustion in internal combustion engines. The increased air mass led into the engine allows more fuel to be injected and burnt, causing a large increase in output power. This can be seen as the original benefit of turbocharger usage. The closest related compressor device to the turbocharger is the supercharger. The basic idea of increasing air charge led into the engine is equal to the turbocharger but the major difference is the driving method. The engine craft shaft usually drives superchargers, while turbochargers are driven by the engine exhaust flow running the turbine and so also the compressor. Since the energy in the exhaust gas for a turbocharged engine is reused and not wasted as in a natural aspirated engine, fuel consumption can be reduced and total engine efficiency improved.

Small turbochargers for automotive usage can reach speeds up to 250 000 rpm, while turbochargers for other applications, for instance large marine turbochargers have a speed range much lower than this. Small high-speed passenger car turbochargers usually use oil lubricated journal bearings and large turbochargers use multilobe or rolling element bearings. Due to axial force imbalance between compressor and turbine side radial bearings are used in combination with thrust bearings. The power dissipation in large low speed turbochargers can be as low as 2- 3% of the total shaft power while in small turbochargers it can be 10% or even more. It must be pointed out that these figures should be regarded as rough estimates [1].

2.1.1 Turbocharger within the chain of engine components

The airflow path through the engine components is illustrated in Figure 2.1. Once the exhaust valve opens, high temperature exhaust gas flows through the exhaust manifold and to the turbocharger, running the turbine wheel. A connecting shaft links the turbine wheel with the compressor wheel and ambient air is sucked into the intake system by the compressor.

Turbochargers are usually used in combination with some type of air cooling system. In Figure 2.1 the system uses a charge air cooler, which serve as a type of heat exchanger and cools the compressed air before it enters the combustion chamber.

Figure 2.1 Intake air and exhaust gas flow paths on a turbocharger engine. Figure from [4].

(17)

4

2.1.2 General turbocharger mechanical design

The turbocharger rotating assembly consisting of shaft, turbine and compressor wheel is supported by journal bearings attached to the bearing housing. A cut out section of a typical small automotive turbocharger can be seen in Figure 2.2.

Figure 2.2 The main components of a small automotive turbocharger. Intake air, exhaust gas and lubricating oil flow paths are illustrated. Figure adapted from Mahle Aftermarket [5].

The journal and the thrust bearings are provided with lubricating oil via the oil inlet. The turbocharger oil lubrication system is typically shared with the engine in order to reduce cost and complexity, which increases with a separate oil system. Consequently, the turbocharger bearings have to be designed for engine oil operation in terms of pressure, temperature and oil composition.

The main tasks for the turbocharger lubrication system are to lubricate and cool the bearings and the bearing housing. Cooling occurs when lubricating oil is heated by friction and heat is conducted away through the oil. Lubrication also provides damping, reduce vibrations and noise, maintain a long service life, improve operational reliability and keep maintenance costs low [6].

(18)

The turbocharger shaft is sealed on both compressor and turbine side. Oil leakage from bearing housing into the compressor may result in oil transferred to the combustion chamber where the oil vaporizes and burns, leading to soot formation and emission production. Hot exhaust gases leaking into the bearing housing from the turbine side may result in an oil temperature rise. The temperature increase in turn may cause a viscosity drop to unacceptable levels leading to bearing failure. For the same reason it is of great importance to reduce heat transfer from the hot turbine to the bearing housing. A heat shield is usually installed between the turbine and the bearing housing in order to reduce heat transfer. In turbochargers with very high exhaust gas temperature it may also be necessary to use water-cooling to cool the bearing housing [1].

2.2 Turbocharger bearing systems

The turbocharger bearing system can be arranged in several ways. The most common configuration, used in current small turbochargers and also in many larger turbochargers is the bearing between wheel assembly presented in Figure 2.3, configuration A. The bearings oil supply is located between the compressor and turbine wheel, which makes it rather simple to isolate the lubricating oil from the gas path. Bearings in this configuration must be designed for long life with minimum maintenance due to poor accessibility. The layout in (B) allows a smaller shaft with high bending critical speed and it also minimize the need of water-cooling.

However, the benefits do not compensate for disadvantages of using the outboard and inboard design. The configuration in (C) is for instance used in small gas turbines and enables separation of the lubricant from the hot turbine. The close-coupled turbine and compressor may cause heat transfer from the turbine to the compressor side, which reduce the charge density. Large turbochargers have used the configuration in (D), where the bearings are outboard mounted. The major benefit of this system is the accessibility to the bearings [1].

Figure 2.3 Bearing system configurations. Red corresponds to turbine and blue to compressor.

2.2.1 Roller bearings

Roller contact bearings are very advantages regarding frictional losses. They generate less frictional loss compared to a conventional floating ring journal bearing. These types of bearings are especially beneficial running at low speed, which results in increased idle speed and faster transient response. Due to durability issues at high turbocharger speeds and high manufacturing costs, roller bearings are not yet used in small mass-produced turbochargers. However, as the bearing technology progresses and manufacturer costs are decreasing roller contacts bearing will be used more and more in small turbochargers [1].

(19)

6 2.2.2 Air bearings

Turbocharger manufacturers have introduced extensive development for air bearings in turbocharger applications, although these types of bearings have not yet reached success. This may be due to a high selling price and problems with start and stop friction, but also that the air bearings have a lower bearing-load capacity than conventional oil lubricated plain bearings.

Air bearings must have a larger diameter and length compared to oil journal bearings due to the lower load capacity of the air film. Additionally, a large thrust disc is required, resulting in increased inertia, which in turn reduces transient response, and increase the turbocharger weight.

The air provided to the bearings may also be isolated from the hot turbine in order to minimize heat transfer. Bearing parts are manufactured with high precision and are coated with high temperature resistant material, which leads to high manufacturing costs and so also a high selling price [1].

2.2.3 Semi- and fully floating ring journal bearings

The fixed sleeve bearing, predecessor to current journal bearings was initially used in small turbochargers and pressed into the bearing housing. It suffered problems with balancing and had a short lifetime. However, large low speed turbochargers may still use these types of bearings.

Later on the fixed sleeve bearing was replaced with various types of floating bearings, which soon became industry standard.

In addition to the inner oil film between the shaft and the bearing inner surface, an outer film between the bearing outer surface and the bearing housing supports the floating bearing in order improve damping and resilience. Floating bearings are either fully or semi floating. Unlike semi floating bearings, the fully floating bearing is rotating with the rotating shaft, while the semi floating bearing is prevented from rotation. This is usually accomplished using an anti-rotation pin locking the bearing. Semi-floating bearings are typically one piece with two bearing surfaces and can be seen as a plain journal bearing in combination with squeeze film damper, which becomes effective when the inner film is stiff enough to transmit shaft motion to the bearing [1].

The one-piece semi floating journal bearings used in the Garrett GT 1752 turbocharger is presented in Figure 2.4.

Figure 2.4 One-piece semi floating journal bearings.

2.2.4 Thrust bearings

Journal bearings are used in combination with some type of thrust bearing to compensate for axial force imbalance. Pressure forces on turbine and compressor blades give rise to imbalance

(20)

in the rotating assembly, which in turn results in a net axial force acting on the shaft, especially when running transient.

In mass produced turbochargers the thrust bearing is usually a plain disc, unlike the ramped or taper land thrust bearing. The ramped bearing can resist high load. High oil film pressure is generated in such a bearing due to forced oil flow and for that reason a smaller contact area is obliged for a specified axial load [1]. The thrust bearing for the turbocharger used in this thesis is presented in Figure 2.5.

Figure 2.5 Thrust bearing used in the Garrett GT 1752 turbocharger

The center hole is where the shaft is located, surrounded by the six thrust pads. Since the bearing is double acting there are sex equal thrust pads on the opposite side.

Thrust bearings are used in combination with thrust collars, which belong to the rotor and act as its sliding surfaces towards the thrust bearing pads. The rotating assembly is floating while the thrust bearing is fixed to the bearing housing. Figure 2.6 illustrates a cut out section of the rotating assembly in the Garrett GT 1752 turbocharger and shows the location of the thrust collars in the bearing system.

Figure 2.6 Thrust collars mounted on the turbocharger shaft. The green arrows symbolize axial motion and the red arrow corresponds to rotational motion.

(21)

8

The thrust collar to the left abuts a thicker section on the shaft whereas the second thrust collar abuts the first thrust collar. The compressor wheel is mounted onto the axis from the right and clamped together with the two thrust collars. The thrust bearing is fitted between the two thrust collars with a small clearance typically a few hundreds of a millimeter. During operation, when the thrust collars together with the shaft rotate, the axial clearance is filled with lubricating oil giving rise to hydrodynamic lubrication.

2.2.5 Rotor dynamics

Shaft rotation unbalance, inertial forces due to engine vibration, and gyroscopic forces from vehicle motions are the three contributors to dynamic loads in the turbocharger. The largest contributor is rotation unbalance. There are also static loads originated from aerodynamic side loading and gravity. In turbo compound machineries there are additional loads from gears.

Balancing is performed individually to turbine and compressor wheels. By doing this the rotating assembly can be reassembled and assembled again and still maintaining an eccentricity ratio within the limits, which facilitates maintenance especially for inboard mounted bearings [1].

2.2.6 Thrust load

The sum of axial forces acting on compressor and turbine originates from pressure and momentum forces corresponding to the axial thrust.

Radial turbines typically have scalloped back faces, which decreases moment of inertia and bore stress. Unlike the radial turbine, compressors impellers are usually designed with full back faces in order to improve aerodynamic efficiency. Due to larger back face area of the compressor wheel compared to the scalloped back face of the turbine wheel, the back face of the compressor wheel carries higher axial load than the back face of the turbine wheel.

It may be difficult to predict axial thrust. In addition, the resultant axial force is calculated from turbine and compressor forces, thus estimations and errors calculating these opposing forces will affect on the net axial force.

To get a true picture of the turbocharger thrust load engine tests have to be performed, where real load and running conditions occur. Stationary tests on turbochargers may not reveal the whole truth about the thrust load characteristics. When running transient the imbalance between compressor and turbine side may be larger compared to high load point of stationary running [1].

2.3 Friction theory for journal and thrust bearings

According to Lamquin and Gjika [7] the main sources for frictional power loss in a rotor bearing system for a turbocharger are radial bearings, thrust bearings, wet-oil-cavity and seals. Friction due to wet-oil-cavity and seals were not taken into consideration in this thesis because these two parameters only give a small contribution to the total frictional losses compared to the contribution from journal and thrust bearing.

The friction characteristics in thrust bearings and journal bearings can be calculated in various ways. The following section describes the different methods and which method that is used in this thesis.

2.3.1 Petroff’s equation for journal and thrust bearings

The power dissipation in the journal bearing is primarily dependent on bearing geometry, oil viscosity, rotational speed and oil film thickness. Generally, the frictional power loss in journal bearings can be estimated with Petroff’s equation

(22)

!_! =! 4

!!^!!!^!

ℎ (2.1)

where ! is dynamic viscosity of the oil, ! is rotational speed, ! is bearing length, ! is bearing diameter and ℎ is oil film thickness. The equation assumes no lubricant flow in the axial direction, no eccentricity between the bearing and the shaft and the oil film is assumed to be unable to support load. According to Petroff’s equation the bearing diameter is the largest influencing factor on power dissipation followed by the rotational speed. Decreasing the shaft diameter is not advisable because of the rotor dynamic issues.

The frictional power loss in double sided thrust bearings can be described by a modified Petroff’s equation

!_!! = !"!^! !_!^!− !_!^!

!_!! (2.2)

where !_! and !_! denote outer and inner radius, and !_!! is axial clearance. Petroff’s equation reveals that the power dissipation is highly dependent on geometry, as for the journal bearing.

Larger bearing contact surface results in higher power loss.

Petroff’s equation only yields a rough estimation of the bearing friction losses. To increase the accuracy and to get a true picture of the losses, more complex methods have to be used.

2.3.2 Hydrodynamic lubrication - Thrust bearing

Hydrodynamic lubrication is characterized by full film lubrication, which is attained when the oil film pressure reach such a level where it can carry the load with no contact between the bearing surfaces. Full film lubrication friction is basically represented by viscous shearing of the oil, where internal friction forces are low [6].

The tapered land pad (A), the michell bearing (tilted bearing) (B) and the rayleigh step bearing (C) are all examples of slider bearings and are illustrated in Figure 2.7.

Figure 2.7 Pressure gradients in the different slider bearings A, B and C. Figure adapted from [6].

A flat surface slides with a velocity U next to the bearing contact surface and forces the lubricant into the converging wedge, which results in a pressure gradient. The shape of the pressure gradient is partly governed by the sliding surface geometry.

(23)

10

The tapered land pad is the type of thrust bearing pad used in the Garret GT 1752 turbocharger.

It consists of two sections; taper (inclined surface) and land (flat surface) section, see Figure 2.8.

Pressure calculation in the taper section is similar to the pressure calculation for the tilted bearing, but with changed boundary conditions. In the land section the film is parallel and the film pressure decreases linearly, as for the stepped bearing.

Figure 2.8 Pressure gradient in a tapered land pad. The z-coordinate describes the film thickness h between the two separated surfaces, ! = 0 at the sliding surface boundary. Figure

adapted from [6].

Reynolds equation is a differential equation, which describes the pressure distribution in the lubricating film between two separated surfaces in motion. The equation is derived in [8].

The modified Reynolds equation for a tapered land pad considering the two sections can be written as

−ℎ^! 12!

!"

!"+!ℎ

2 = −ℎ_!^! 12!

−!_!!

! − !_!+!ℎ_!

2 (2.3)

The pressure gradient can be written as

!"

!" = ℎ_! ℎ

! !_!!

!_!− !+ 6!"ℎ − ℎ_!

ℎ^! (2.4)

where !_!! = ! ! = !_! is unknown. If integrating the pressure the equation can be written as

! ! = ℎ_!

ℎ

! !_!!

!_!− !+ 6!"ℎ − ℎ_! ℎ^!

!

!" (2.5)

If !_!! = ! ! = !_! , the parameter !_!! can be derived from

(24)

!_!! = ℎ_! ℎ

! !_!"

!_!− !+ 6!"ℎ − ℎ_! ℎ^!

!!

!

!" (2.6)

and if substituting ℎ = ℎ !

ℎ ! = ℎ_!−ℎ_!− ℎ_!

! ! (2.7)

this results in

!_!! = 6!"#

ℎ_!^!

!^! !^! − 1 ! − 1

−2!^!+ 2!^!!^!− !!^! − !^! (2.8) where !^! is defined as

!^! =!_!

! (2.9)

and ! is defined as

! =ℎ_!

ℎ_! (2.10)

The final equation for describing the pressure distribution ! ! is a result of substituting the expression (2.8) for !_!" into equation (2.5).

The load capacity in the bearing can be obtained by integrating the pressure ! ! :

!

! = ^!^!! !

!

!" +!_!!

2 ! − !_! (2.11)

Solving equation 2.10 gives !^! as function of !^! and n. Table 2.1 shows the maximum load capacity !^!. The maximum value !^! is derived from values of !^!and by calculating !"^!/!" = 0.

Table 2.1. Maximum load capacity for a tapered land pad. Table adapted from [6].

!^! 1/5 2/5 3/5 4/5 1

n 2.01 2.08 2.149 2.239 2.189

!^! 0.07 0.13 0.173 0.192 0.16

The shear stress, !_!" in the film can be calculated by

!_!" = !!"

!" (2.12)

The velocity ! can be obtained by rearranging equation 2.3 with ! in the left term. If ! = ℎ, the modified equation 2.3 with ! in the left term can be derived with respect to the variable !.

(25)

12

By integrating the shear stress and multiply with the velocity U, the total power loss can be determined with following equation

!_!,! = !_!"∙ ! ∙ !"

!

! (2.13)

If considering the frictional power loss over a surface described with polar coordinates the corresponding equation is

!_!,!" = ^!!"!^!!"!#

!

!!

!_!

(2.14)

where !_! and !_! are inner and outer radius of the thrust bearing pad. The variable ! is the angular extent of the thrust pad. The equation is derived in [9].

The above model with the isothermal assumption gives a reasonable approximation of the frictional power losses in a thrust bearing. Since the model assumes constant oil viscosity in the bearing, it may not give realistic values of the power loss. The model can be improved by taking thermal effects into account and add the energy equation [11] to the model. This enables the usage of variable viscosity, which may give more realistic results. Due to constraints in the thesis time plan the thermal effects were partially taken into account by using an isoviscous model. The oil viscosity in the bearings were calculated from an average of the oil temperature between inlet and outlet in the bearing housing.

2.3.3 Hydrodynamic lubrication - Journal bearings

Hydrodynamic lubrication in journal bearings is based on the same idea as for the thrust bearing.

Isoviscous models were decided to be used for the journal bearings friction simulations as well.

The equations in chapter 2.3.2 were modified according to [11].

2.3.4 Reynolds number

By calculating Reynolds number for the oil flow in the bearings it can be determined whether the flow is laminar or turbulent. Reynolds number for fluid flow in a thrust bearing [12] can be expressed as

!!_!! =!_!!"ℎ

! (2.15)

To determine to characteristics of the oil flow an extreme case was considered. Assuming a maximum turbocharger rotational speed of 100 000 rpm, a film thickness as thick as the total bearing clearance (0.095 mm), a minimum oil viscosity at 100 ℃ of 15 cSt and oil density of 888 kg/m³. Oil density is obtained from chapter 3.5.4. With assumed values Reynolds number was calculated to Re = 384. The transformation between laminar and turbulent flow is in the region of Re = 1000 [12]. Consequently, the oil flow in the thrust bearing can be considered as laminar.

Reynolds number for journal bearings [13] is written as

!!_! =!"#

! (2.16)

where ! denotes the radial clearance and is expressed

(26)

! = !_!,! − !_!!!"# (2.17) The radius, !_!!!"#, correspond to the raduis of the turbocharger shaft were the journal bearing is located and !_!,! is the inner radius of the journal bearing. With the same assumptions as for the thrust bearing calculation and a radial clearance of 0.01 mm, Reynolds number was calculated to 28, which indicates that the oil flow is laminar.

This means that no turbulence model must be introduced in the bearing friction simulation.

2.4 Experimental methods to measure frictional losses in turbocharger bearings

According to [14] there are three different methods to measure mechanical frictional losses in the turbocharger: Calorimetric measurement techniques, measuring gas enthalpy variation and measuring the friction torque with a torque meter. Each method is described below.

2.4.1 Calorimetric measurement techniques

If measuring the frictional heat up of the oil in the turbocharger, two assumptions have to be made: there is no heat exchange inside the turbocharger and the mechanical friction is totally dissipated in the oil flow. With these assumptions the mechanical frictional losses can be described as

!_!"# = !_! = !_!"#∙ !_{! !"#} ∙ ∆!_!"# (2.18)

where

∆!_!"# = !_!"#!− !_!"#! (2.19)

where !_!"#! and !_!"#! correspond to outlet and inlet oil temperature in the turbocharger. The parameter !_!!" is the cooling effect on the oil, !_!"# represents oil mass flow through the turbocharger and !_{! !"#} is oil heat capacity.

As stated by [14], the quality of the result depends on the effectiveness of the assumption of no heat exchange and on the accuracy of temperature and mass flow measurements. In order to reduce the heat transfer the turbocharger should be insulated and it should be fed with oil and air at the same temperature. Another challenge using this method is the temperature measurement of the outlet oil. According to [7] it is claimed that the outlet oil is spread in the form of droplets in an area where oil and air are mixing, making the measurement of the oil outlet temperature quite difficult.

2.4.2 Measuring gas enthalpy variation

According to [14] this method assumes adiabatic conditions of the gas evolution in the compressor and turbine. Consequently, the power extracted by the turbine and the power received by the compressor can be written as

!_!= !_!∙ !! !"!!"#$ !_!− !_! (2.20)

(27)

14

where !_! is turbine power, !_! is exhaust flow through the turbine and !! !"!!"#$ is exhaust gas heat capacity. The temperatures !_! and !_! are outlet and inlet temperatures in the turbine.

The compressor power, !_! is written as

!_! = !_!∙ !_{! !"#} !_!− !_! (2.21)

The temperature difference corresponds to air temperatures after and before compressor. The mechanical friction power dissipation can be expressed as

!_! = !_!− !_! (2.22)

Furthermore, it is stated by the author that the accuracy of this model is dependent on the effectiveness of the adiabatic evolution. As for the method using a calorimetric measurement technique the turbocharger has to be insulated and the temperatures chosen to reduce heat exchange as much as possible.

2.4.3 Measuring the friction torque with a torque meter

This is the method used in [14]. As stated in the paper, the rotating torque meter is preferably installed between a drive (the turbine or an electric motor) and the central housing containing the bearings. The torque meter measures the friction torque directly and the accuracy of the measurement is equal to the precision of the torque meter accuracy, which in this case is

±0.00019 Nm.

2.4.4 Experimental method used in this thesis

The experimental approach used in this thesis was based on the calorimetric measurement technique described above. Still the two above mentioned assumptions had to be fulfilled i.e. that there is no heat exchange between the oil and the bearing housing and the mechanical friction is totally dissipated in the oil flow.

A major difficulty measuring frictional losses in the turbocharger is the presence of heat transfer.

Due to large temperature differences in the turbocharger, there is heat transfer from the turbine to the compressor side. The heating of the lubricating oil originates from two sources: friction and heat transfer. Furthermore, this disables to determine the frictional power loss from heat capacity and oil mass flow and oil temperature measurements without taking the heat transfer into consideration [14]. Adjusting the bearing housing temperature to be in the same range as the oil temperature can minimize the heat transfer from the turbocharger to the oil. Since the temperature difference is minimized, also the heat transfer between the oil and the bearing housing is assumed to be minimized. Therefore, the heat up of the oil can be assumed only coming from the friction. This experimental method was used in this thesis. The power loss in the turbocharger can be expressed with equation (2.18).

(28)

3 IMPLEMENTATION

In the implementation chapter it is described how the thesis work was performed. The different subsections are bearing geometry measurement, axial force analysis, bearing friction experiment and bearing friction simulation.

3.1 Bearing geometry measurement

When the simulation models for the thrust and journal bearings were created geometrical input parameters had to be identified and determined. The bearing geometry, particularly bearing contact surfaces have a great influence on the bearing performance, hence it has to be measured with high precision. The bearing geometry measurements are further described in Appendix A.

3.2 Axial force analysis on turbocharger shaft

The axial force acting on the thrust bearing has a significant influence on the film thickness in the bearing and so also on the bearing frictional power loss. In this thesis the axial force was determined in three different ways: axial force calculated with force balance analysis, axial force calculated as a function of measured axial displacement and axial force measured with strain gauges. The two latter methods are further presented in Appendix B. The axial force based on force balance described in 3.2.1 was used as an input parameter for the simulations.

3.2.1 Axial force and pressure analyses

As stated by Baines [1], the sum of axial forces acting on compressor and turbine originate from pressure and momentum forces, corresponds to the axial thrust. The model used to calculate the resulting axial force is based on force equilibrium calculations from [15]. Geometry used in the calculations is defined in Figure 3.1 below.

Figure 3.1 Geometry of turbocharger rotor. Figure adapted from [15].

In Figure 3.2 the resultant axial force acting on the compressor wheel, !_!, is comprised of three components, !_!!, axial force at compressor wheel inducer, !_!!, axial force at compressor wheel exducer and !_!!, axial force on compressor wheel back face. The left side of the rotor corresponds to the compressor side and the right side corresponds to turbine side.

(29)

16

Figure 3.2 Axial forces balance on turbocharger shaft. Figure adapted from [15].

If considering the force equilibrium on the compressor side the force !_! can be expressed as

!_! = !_!!− !_!!− !_!! (3.1)

The axial force at the compressor inlet, !_!!, can be written as

!_!!= ! ∙ !!_!"_!"#! ∙ !_!+ !_! ∙ !_! (3.2) where !!_!"_!"# denotes inlet tip radius of the compressor wheel and !_! is the compressor inlet pressure. The axial airflow velocity at compressor inlet, !_!, was calculated as

!_! = !_!

!_!∙ !_! (3.3)

where !_! is the compressor inlet pipe cross sectional area and !_! is the air density at the compressor inlet. The force !_!! is defined as

!_!! = ^!!_!"#_!"#!!_!"#$%!& !

!!_!"_!"#

∙ 2!"#" (3.4)

where !!_!"#$%!& ! is the compressor wheel exducer pressure distribution along the wheel radius and can be written as

!!_!"#$%!& ! = !_! ∙ !

!!_!"#_!"#

! (3.5)

The radius !!_!"!_!"# defines the outlet tip radius of the compressor wheel. The parameter !_! denotes pressure after compressor and were measured during engine testing together with !_!. The axial force acting on the compressor wheel back face, !_!!, can be derived from the exact solution of Navier Stokes equations [16]. This assumes that the compressor wheel back face can be approximated as a flat disk rotating with uniform rotational speed. In addition, the air flow around the disk is assumed to be incompressible. The pressure on the compressor wheel back face can be written as

(30)

!_{!_!"#$} = 1

2!_!!^!!^! (3.6)

and by integrating the pressure over the compressor wheel back face the force can be expressed as

!_!! = ^!!_!"#_!"#!_{!_!"#$}∙ 2!"#"

!_!" (3.7)

The variable !, indicates the rotational speed of the turbocharger shaft and !_!! is the air density after the compressor. The densities !_!! and !_!! are calculated with the ideal gas law based on current temperatures and pressures for the different locations. Due to geometry similarities between the turbine and the compressor in this turbocharger configuration same types of equations are used for the flow analysis for the turbine side. The equations for the turbine side are stated below.

!_! = !_!!+ !_!!−!_!! (3.8)

!_!! = ! ∙ !_!!!_!"#_!"#∙ !_!+ !_!∙ !_! (3.9)

!_!! = !!_!"#$%&' !

!!_!"_!"#

!!_!"#_!"#

∙ 2!"#" (3.10)

!_!! = ^!!_!"#$$%&!_{!_!"#!} ! ∙ 2!"#"

!_!"

(3.11)

Variable definitions for equations above can be find in the nomenclature. Finally, the total resulting axial force, !_!", acting on the turbocharger shaft can be expressed as

!_!" = !_! + !_! (3.12)

The resulting axial force is either directed form the compressor side to turbine side or vise versa.

3.3 Radial forces on turbocharger shaft

As described in 2.2.5 forces in the turbocharger are both dynamic (mainly due to shaft rotation unbalance both also because of inertial forces due to engine vibration, and gyroscopic forces from vehicle motions) and static loads (originated from aerodynamic side loading and gravity).

As input data for the journal bearing simulations the load was estimated by assuming that the gravitational force corresponds to the total load. This should be regarded as a rough estimation since the assumption does not reflect the real dynamic running conditions. The rotor assembly was measured to !!"#"! = 140 !. The center of mass is assumed to be in between the two journal bearings resulting in equal radial forces on each bearing. The total radial load can be calculated with equation (3.13):

!_! = !!"#"!∙ ! (3.13)

(31)

18

For journal bearing simulation the same method estimating the radial load is used in [10].

However, in reality the center of mass is located closer to the turbine wheel resulting in another force distribution.

3.4 Experimental setup

3.4.1 Experimental setup – engine and turbocharger

The engine used in the experiment is an Opel LNF, 2 liter, four-cylinder, direct injected SI- engine equipped with a water-cooled GT1752 turbocharger from Garrett. Figure 3.3 shows the on-engine turbocharger together with the separate oil lubrication system.

Figure 3.3. The on-engine turbocharger together with the oil rig.

As described in chapter 2.4.4 the bearing housing temperature was adjusted to be equal to the average of inlet and outlet oil temperature in order to minimize heat transfer. By installing a valve for adjusting the cooling water flow to the turbocharger it was possible to maintain a certain bearing housing temperature. The water used in the turbocharger cooling system was taken from a separate cooling system and had a temperature close to room temperature. An illustration of the bearing housing oil and water flow paths is presented in Figure 3.4.

(32)

Figure 3.4 Illustration of the thermocouples locations as well as cooling water and oil flow paths through the bearing housing. The cooling jacket close to the oil inlet is located behind the oil

channel.

Initially, the bearing housing temperature was calculated as an average of the temperatures measured with thermocouple 1 and 2. Unfortunately, due to the location of the thermocouples radiation from the hot turbine caused major impact on the temperature. Also, the cooling jacket, which is located in between the bearings and the thermocouples contributed to an unrealistic estimation of the bearing housing temperature. Since the aim was to measure the bearing housing temperature close to the oil and the bearings it was decided to install a third thermocouple closer to the oil channel. Thermocouple 3 is located 8 mm in the goods of the bearing housing as shown in Figure 3.4. This location is beneficial since the effect of radiation is much smaller compared to the surface mounted thermocouples, but also because the impact on the cooling water is reduced.

3.4.2 Axial force measurement

The turbocharger thrust bearing was provided with strain gauges in order to measure the axial force applied on the thrust bearing. To make the strain gauges more sensitive material was removed around the bearing thrust surfaces. The axial force calibration is further described in Appendix B. The modified bearing with strain gauges connected can be seen in Figure 3.5. Axial force measurement with strain gauges in turbochargers is a well-tried method and were also used by [7,17,14].

Figure 3.5 Strain gauges attached to the thrust bearing.

(33)

20 3.4.3 Displacement measurement

The turbocharger is provided with a displacement sensor that measures the shaft displacement, see Figure 3.6. Since a large amount of material is removed from the thrust bearing it was assumed that the thrust bearing was flexible and consequently had the characteristics of a spring.

Therefore the displacement of the shaft could be calibrated as a function of force. The calibration is described in Appendix B.

Figure 3.6 Location of displacement sensor.

3.4.4 Compressor wheel back face pressure measurement

In order to verify the analytical compressor wheel back face pressure calculation the pressure on the compressor back plate was measured. The intake hole for the air leading to the pressure transducer is located on the compressor back plate 17.2 mm from the compressor wheel center.

3.4.5 Experimental setup – separate oil lubrication system

A separate oil lubrication system for the turbocharger was developed in order to control the condition of the oil independent on engine condition and operation point. A flowchart of the oil rig is enclosed in Appendix C. Initially, the oil rig was designed to be able to switch between two different lubricating oils during operation. The purpose was to investigate the effect of changing oils with different viscosities. Despite oil viscosity, parameters affecting the turbocharger operation conditions such as turbocharger speed, oil temperature, oil pressure etc. were aimed to be constant for each operation point. When changing oils during operation, likely there would be a shift in turbocharger rotational speed. Either a reduction in speed if the changed oil had a viscosity higher than previous oil or a speed increase if the oil viscosity was lower than the previous oil. Theoretically, from the speed shift the friction torque difference between the two cases of oil viscosities could be derived. The two cases were aimed to be run in the 1D engine simulation software GT-Power, where there is an existing model of the particular engine used in this thesis. By applying a counter acting torque on the turbocharger shaft whose magnitude corresponds to a value, which adjusts the turbocharger rotational speed equal to the speed of the opposite case with different oil viscosity, the applied torque can be assumed to be the difference in friction torque between the two cases. Thus the results of friction torque from the simulations could be verified by this method.

Due to a tight time schedule the oil rig was not finished to such extent that the oil switch function would work. Instead, the oil rig was redesigned to only run with one oil type. Still oil pressure and oil temperature could be controlled independent on engine operation conditions.

(34)

Originally the oil rig was built with two sets of individual oil systems in order to be able to switch between the two different oils. However, only one system was used in thesis. The oil rig consists of heat exchanger, oil pump, oil tank, water heater, pressure sensor, thermocouples, flow meter and bypass valve, see Figure 3.6.

Figure 3.6 Oil rig set up.

The bypass valve was used to adjust the oil pressure and so also the oil flow into the turbocharger.

The oil temperature was controlled by a heat exchanger. A water circulator pump provides the heat exchanger with water, warmed up by the water heater. A return hose guides the oil back from the turbocharger to the oil tank.

3.4.6 Pressure and temperature measurement equipment

• All pressures used in the calculations were measured with pressure transducers, which were calibrated before the experiment.

• Turbocharger oil inlet and outlet temperature as well as turbocharger cooling water inlet and outlet temperatures are measured with PT100 temperature sensors. Remaining temperatures were measured with thermocouples.

Friction Characterization of Turbocharger Bearings

Friction Characterization of Turbocharger Bearings

Friction Characterization of Turbocharger Bearings

Erik Sjöberg

Abstract

Sammanfattning

ACKNOWLEDGEMENT

NOMENCLATURE

Symbol Definition Unit

Subscript Definition

Abbreviations

TABLE OF CONTENTS

1 INTRODUCTION

1.1 Background

1.2 Objective

1.3 Delimitations

1.4 Methodology

2 FRAME OF REFERENCE

2.1 Turbocharging

2.2 Turbocharger bearing systems

2.3 Friction theory for journal and thrust bearings

2.4 Experimental methods to measure frictional losses in turbocharger bearings

3 IMPLEMENTATION

3.1 Bearing geometry measurement

3.2 Axial force analysis on turbocharger shaft

3.3 Radial forces on turbocharger shaft

3.4 Experimental setup