Integrated engine waste heat recovery by combination ofevaporative engine cooling and Rankine bottoming cycle.

(1)

March 27, 2014

Integrated engine waste heat recovery by combination of evaporative engine cooling and Rankine bottoming cycle.

Application to heavy duty engines.

Vincent CHOQUET choquet@kth.se

Master of Science Thesis MMK 2014:09 MFM 155 MF205X Internal Combustion Engine KTH Industrial Engineering and Management

Machine Design

SE-100 44 STOCKHOLM

(2)

Examensarbete MMK 2014:09 MFM 155 Återvinning av motorvärme i lastbilsmotorer genom att kombinera ett evaporativt motorkyl- ningssystem med en Rankine-cykel

Godkant Examinator Handledare

March 27, 2014 Crohnjort Andreas Latz Gunnar Kontaktperson Thomas Reiche

Sammanfattning

Motorns kylförluster utgör ca 20 % av energin som injiceras i dieselmotorn på en modern lastbil. Målet med detta examensarbete är att undersöka om flödes kokande kylning är en termiskt effektiv metod för att återvinna spillvärmes och en effektiv lösning för precisionskyla.

Först genomfördes en motorvärmeöverföringsmodell på GT-suite för att beräkna värmeflöden i motorcylindern. Eftersom cylinderfoder är mindre termiskt begrän- sade än topplockundersöktes flödes kokande kylning i cylinderfoder. En mer an- passad värmeöverföringsmodell med hänsyn till både avgas- och kylmedelsidan på cylinderfoder genomfördes således med Simulink. Till skillnad från kommersiella programvaror, gör denna enkla modell det möjligt att utföra 2-fas värmeöver- föringskorrelationer och studera flödes kokande beteendet i detalj. De viktigaste parametrarna (vattenmantelns hydrauliska diameter, vätsketrycket och ytan av värmeöverföringarna) studerades för olika massflöden för att analysera hur de påverkar väggtemperatur och värmeöverföring.

Undersökningen visade goda arbetsförhållanden för mycket låga massflöden (ca 1 % av det typiska massflödet för konvektiv vätskekylning). På grund av prob- lem med flödesregleringen, behövde andra vätskor beaktas som köldmedier men hade god potential för att kyla systemet effektivt. För att studera potentiella förbättringar av energieffektiviteten infördes den flödes kokande modellen slutli- gen i en komplett modell av en Rankine-krets där vatten användes som kylmedel.

En Rankine-krets med vatten skulle förbättra värmeåtervinningen på den avsedda

(3)

Msc Thesis Report March 27, 2014

motorn med 4,8 % motorns bromskraft, genom att återvinna värme från cylinder- foder och avgaserna vid 1800 RPM, full belastning. Ytterligare simulationer har också hållits med R245fa, som visar en återvinning av spillvärmen med 5,5 % av motorns bromskraft vid 1800 RPM, full belastning.

KTH MF205X ii Vincent CHOQUET

(4)

Master of Science Thesis MMK 2014:09 MFM 155

Integrated engine waste heat recovery by combi- nation of evaporative engine cooling and Rankine bottoming cycle.

Application to heavy duty engines.

Approved Examiner Supervisor

March 27, 2014 Crohnjort Andreas Latz Gunnar Contact person Thomas Reiche

Abstract

Engine Cooling losses constitutes about 20% of the injected fuel energy in a mod- ern heavy duty truck diesel engine. The objective of this Master Thesis Project is to investigate flow boiling cooling as a thermally efficient method for waste heat recovery as well as a good solution for precision cooling.

First, an engine heat transfer model was implemented on GT-suite software in or- der to estimate heat fluxes within the engine cylinder. Liners being less thermally constrained than the cylinder head, flow boiling cooling was then investigated in the liner’s water jackets. A more adapted heat transfer model taking into account both gas side and cooling side of the liner was thus implemented on Simulink.

Unlike commercials software, this simple model allowed to implement the relevant two-phase heat transfer correlations and to study in details the boiling flow behav- iors. The hydraulic diameter of the water jackets, the fluid saturated pressure and the surface area of heat transfer are the major parameters and they were studied for various mass flow rate in order to analyze how they influence wall temperature and heat transfer.

This study showed good operating conditions for very low mass flow rate (about

1% of the typical mass flow rate for liquid convective cooling). Due to flow control

issues, it implied the consideration of other fluids such as refrigerants but showed

good prospect for cooling system simplification. This flow boiling model was finally

(5)

Msc Thesis Report March 27, 2014

inserted in a complete Rankine loop model using water as a working fluid to study potential efficiency improvements. A Rankine loop using water as a working fluid would thus improve the heat recovery of the considered engine of about 4.8% of the net engine brake power, recovering heat from the liners and the exhaust gases at 1800RPM, full load. Further simulations have also been led with R245fa, which shows a WHR of about 5.5% of the net engine brake power at 1800RPM, full load.

KTH MF205X iv Vincent CHOQUET

(6)

Nomenclature

Symbols

A Surface Area m

²

B Cylinder bore m

C

_p

Thermal capacity of the material J/K.kg

d or D Diameter m

e Thickness m

F Force N

h Enthalpy J/kg

k Material conductivity W/m

²

.K

˙

m Mass flow rate kg/s

N Engine Speed RP M

p Pressure P a

P Power W

Q ˙ Heat transfer W

T Temperature K

v Instantaneous speed m/s

V Volume m

³

V

_p

Mean Piston speed m/s

W T Thrust N

W ˙ Mechanical Work W

α Heat transfer coefficient K/m

²

.K

µ Dynamic viscosity kg/s.m

ρ density kg/m

³

Dimensionless Numbers

Bo Boiling Number

E Two phase convection coefficient

f Friction Factor

N u Nusselt Number

P r Prandtl Number

Re Reynolds number

x or X Vapor quality

η Efficiency

(7)

Msc Thesis Report March 27, 2014

Sub and super-scripts

cold cold side of the wall

E exchange

exh exhaust

exp expander

g gas

hot hot side of the wall

i element number (tube discretization)

in input

L liquid

lam laminar

n time step number

o oil

out output

pump pump

sat saturated state

s skirt

tb turbulent

tp two phase

turb turbine

KTH MF205X vi Vincent CHOQUET

(8)

Abbreviations

1-D One dimension (numerical modeling) ATR Advanced Technology and Research

CAC Charge Air Cooler

CAD Crank Angle Degree

CFD Computational Fluid Dynamic

EG Ethylene Glycol

EGR Exhaust Gas Recirculation

EIWHR Engine Integrated Waste Heat Recovery

ER Engineering Report

Eu European Pollutant Emissions Standards (3, 4, 5 or 6 in this re- port)

FE Finite Element

GHG Greenhouse gas

GT Gamma Technologies

GWP Global Warming Potential

HDE Heavy Duty Engine

ICE Internal Combustion Engine

PCU Piston Cylinder Unit

SCR Selective Catalytic Reduction

TC313 Relative to engine heat balance tests

WHR Waste Heat Recovery

(9)

List of Tables

2.1 Conditions in the case of the highest cooling power, 1800RPM, full load . . . . 9 3.1 GT-Suite technical Applications . . . 18 3.2 Engine Characteristics of the 11L diesel engine considered . . . 21 3.3 Convective ports available with EngCylStructCondelement (Lim-

ited to the use of 2D water jacket model) . . . 22 3.4 Operating points considered for the calibration . . . 33 3.5 Description of the fluid states along the Rankine loop model . . . . 53 3.6 Definition of basic Rankine cycle quantities . . . 54 4.1 Exhaust gas conditions . . . 74 B.1 Display of the Simulink and test wall temperature results for liquid

convective cooling with EG-water(40/60) for the considered operat- ing points (full load). . . 97 B.2 Display of the modeled and tests heat balances results for liquid con-

vective cooling with EG-water(40/60) for the considered operating points (full load). . . 98 B.3 Display of the modeled heat flux results for liquid convective cooling

with EG-water(40/60) for the considered operating points (full load). 99 B.4 Display of the Simulink and GT Wall temperature results for liquid

convective cooling with water for the considered operating points (full load). . . 100 C.1 Interpolation of the EG-W mixture properties from EG and water

ones (international system units). . . 102 E.1 Sorting of few fluids available with REFPROP tool. . . 107

x

(12)

1.1 Total GHG emissions by sector in the EU-27, 2011. [2]. Transport

sector includes marine transportation and internal flights. . . . 2

1.2 Energy balance of a heavy duty diesel truck engine, Volvo 11L at full load. Speed normalized with the highest possible speed. . . . . 3

2.1 Turbine and reciprocating Rankine cycles [12]. . . . 8

2.2 Rankine cycle in a Temperature-Entropy diagram. . . . 8

2.3 Boiling flow heat transfer regimes in a vertical tube. Source : Collier & Thome (1994) . . . 11

3.1 Overall structure of the investigations. . . 17

3.2 Engine and cooling circuits coupled model overview. . . 20

3.3 Coolant Side of the model, zoom on one cylinder. . . 23

3.4 Evolution of the heat transfer coefficient as function of crank an- gle for Hohenberg Correlation and Woschni correlation (adapted by GT) at 1800 RPM and full load. . . 26

3.5 Evolution of the heat transfer coefficient as function of the distance from cylinder top. Hohenberg Correlation. . . 27

3.6 Evolution of the oil film thickness as function of the CAD. . . 29

3.7 Evolution of the Heat transfer coefficient as function of the CAD, 1800RPM Full load. . . 29

3.8 Evolution of the average heat transfer coefficient between the piston and the liner. . . 30

3.9 Evolution of the friction power losses as function of the CAD [deg]. 31 3.10 Evolution of the friction losses power as function distance from the cylinder top. . . 32

3.11 Evolution of the averaged friction losses power as function of the

engine speed [RPM] at full load. . . 32

(13)

Msc Thesis Report March 27, 2014

3.12 Nodes for GT FE analysis and the thermocouples postition for tests. 34 3.13 Simulated temperatures and measured temperatures as function of

the operating point (full load). . . 34

3.14 Difference between simulated temperatures and measured temper- atures for various operating points (full load). E=Exhaust, I=Intake. 35 3.15 Difference between simulated temperatures and measured temper- atures for various operating points (full load). . . 37

3.16 Distribution of temperatures as a result of GT FE analysis for 1800 RPM, full load. . . 38

3.17 Piston Geometry, nodes taken into account for GT simulation and position of the thermocouples. . . 39

3.18 Evolution of the second land and skirt temperatures as funciton of the operating point. . . 39

3.19 Difference between simulated temperatures and measured temper- atures of the second land for various operating points (full load). . . 40

3.20 Comparison of the model with TC313 experimental investigations. . 41

3.21 Architecture of the liner heat exchange model. . . 43

3.22 Architecture of the gas side heat exchange submodel. . . 44

3.23 Architecture of the liner wall submodel. . . 44

3.24 Architecture of the coolant side heat exchange submodel. . . 44

3.25 Percentage of errror between the simulated wall temperature and the results from tests (Ethylene glycol 40-60). . . 48

3.26 Percentage of errror between the simulated coolant power and the results from tests in relation to test results(Ethylene glycol 40-60). . 49

3.27 Percentage of error between the simulated wall temperature on GT and on Simulink. . . 50

3.28 Sketch of the complete Rankine cycle loop. . . 53

4.1 Volume flow rate parameter study . . . 56

4.2 Volume flow rate parameter study : Influence on Coolant Temper- atures (Inlet, liner outlet, cylinder outlet) . . . 56

4.3 Volume flow rate parameter study : Influence on Wall Temperatures (Head, Liner) . . . 57

4.4 Torque parameter study . . . 58

4.5 Torque parameter study : Influence on Coolant Temperatures (In- let, liner outlet, cylinder outlet) . . . 59

KTH MF205X xii Vincent CHOQUET

(14)

4.6 Torque parameter study : Influence on Wall Temperatures (Head, Liner) . . . 60 4.7 Inlet temperature study: Influence on Coolant Temperatures (Inlet,

liner outlet, cylinder outlet) . . . 61 4.8 Inlet temperature parameter study: Influence on Wall Temperatures

(Head, Liner) . . . 62 4.9 Saturated pressure parameter study : Influence on output quality . 64 4.10 Saturated pressure parameter study : Influence on maximum wall

temperature . . . 65 4.11 Saturated pressure parameter study : Maximum wall temperature

contours . . . 66 4.12 Diameter parameter study : Influence on output quality . . . 67 4.13 Diameter parameter study : Influence on maximum wall temperature 67 4.14 Diameter parameter study : Maximum wall temperature contours . 68 4.15 Area of heat exchange parameter study : Influence on output quality 69 4.16 Area of heat exchange parameter study : Influence on maximum

wall temperature . . . 69 4.17 Area of heat exchange parameter study: Maximum wall tempera-

ture contours . . . 70 4.18 Comparison of the wall temperature with the typical liquid cooling

case (1800RPM). Wall temperature with flow boiling is higher, but still below usual design standards. . . 72 4.19 Heat flux distribution (1800RPM) . . . 73 4.20 Distribution of the cooling losses and comparison with the typical

values (1800RPM) . . . 73 4.21 Preheating power as function of the total mass flow rate for various

saturated pressures p

sat

(1800RPM). . . 75 4.22 Evolution of the net recovered power at the output of the expander. 76 4.23 Evolution of the Rankine efficiency [%] as function of the coolant

mass flow rate for various case of saturated pressure. . . 76 4.24 Bypassing of the fluid between the evaporator and the superheater.

1= The whole flow goes through the expander. . . 77 4.25 Architecture for an integrated engine WHR system using exhausts

as a preheating and superheating source . . . 77 4.26 Distribution of recuperated heat for ˙m

total

= 0.013kg/s per cyl. and

, at 1800RPM full load. . . 78

(15)

Msc Thesis Report March 27, 2014

4.27 Evolution of the net recovered power and the recuperated power as function of the vane bypass. . . 79 4.28 Evolution of the net recovered power as function of the engine

coolant mass flow rate. . . 81 4.29 Proposed architecture for WHR system working with R245fa. . . . 81 4.30 Rankine efficiency as function of the mass flow rate and the satu-

rated pressure for R245fa. . . 82 4.31 Distribution of the recovered heat at ˙m

total

= 0.2kg/s and p

sat

=

1000kP a . . . 82 4.32 Distribution of the power (recuperated and net) at ˙m

total

= 0.2kg/s

and p

sat

= 1000kP a . . . 83 A.1 Engine Side of the model . . . 92 A.2 Coolant Side of the model. . . 93 A.3 Liner temperatures simulated with GT for various operating points

(full load). . . 94 B.1 Display of the Simulink model for a discretization of the tube in 20

elements. . . 96 C.1 Evolution of the heat transfer coefficient as function of the mass

flow rate in liquid convective cooling for water and EG-W(40-60) mixture. . . 101 D.1 Sketch of the counter flow model. . . 104 D.2 Evolution of the cold side fluid enthalpy in the case of ˙Q

cold

< ˙ Q

_hot

. 105 D.3 Evolution of the cold side fluid enthalpy in the case of ˙Q

cold

> ˙ Q

_hot

. 106

KTH MF205X xiv Vincent CHOQUET

(16)

Introduction

The following part aims at presenting the origins of the project and detailing its objectives. After a short description of the background, the problematic is laid and the objectives of the project are defined.

1.1 Background

Up to now, the Euro norms for passenger and commercial vehicles have been regu- lating pollutant emissions such as nitroxid, particles and carbon monoxid emissions . The Euro 6 norms are the latest to date and have been applied in 2013. Today, the European Commission is taking steps toward a reduction of GHG emissions and a decarbonisation of the transport.

In the context of a necessary reduction of carbon emissions and the fight against climate change, the EU commission would need to reduce its emissions of GHG by 80% below its 1990 level by 2050 [1]. As transport accounts for 20.3 % of the total GHG emissions in EU in 2011 and road transportation represents 94% of it (see Figure 1.1, [2]), the European commission target for 2050 is thus to reduce by 60% the level of GHG emissions according to the White Paper on Transport 2011 [1].

In addition to the coming norms on GHGs emissions for commercial road vehicles, oil price increase pushes transportation industries to further improve their vehicles fuel consumption.

The importance of fuel efficiency in the truck market encourages companies to

investigate cutting edge technologies to recover waste heat from their heavy duty

diesel engines. Rankine bottoming cycles, 6-strokes engines and thermo-electricity

are mainly investigated. However, Rankine bottoming cycle technologies are likely

(17)

Msc Thesis Report March 27, 2014

to be released first on the market, due to their simplicity, their low cost and their significant efficiency.

Rankine cycles have been commonly used for power generation from heat, such as in power plants or steam engines for instance. Nowadays, the implementation of such waste heat recovery (WHR) system is also investigated on heavy duty en- gines.

As a matter of fact, waste heat sources on an engine are manifold. In addition to the common heat sources such as exhaust gases and engine cooling, the exhaust gas recirculation (EGR) cooler and the charge air (CAC) cooler constitute signifi- cant losses.

The exhaust gas heat source is the most considered for WHR thanks to its high temperature level. However, engine cooling stands for significant losses (the ratio of cooling losses to injected fuel energy ranges from 22 to 25% on a Volvo 11L engine, see Figure 1.2). It constitutes therefore an non-negligible source of heat.

Figure 1.1: Total GHG emissions by sector in the EU-27, 2011. [2]. Transport sector includes marine transportation and internal flights.

KTH MF205X 2 Vincent CHOQUET

(18)

Figure 1.2: Energy balance of a heavy duty diesel truck engine, Volvo 11L at full load. Speed normalized with the highest possible speed.

1.2 Problem statement

Direct heat recovery from the engine without affecting the engine cool- ing efficiency Heat can be recovered from the engine both directly or indirectly, by considering the engine coolant as the working fluid for the WHR cycle, or by using a secondary flow loop getting the heat from the main one.

Direct heat recovery from the engine allows to get an higher temperature for the hot source. Thus, a higher Rankine efficiency could be expected. The total weight and volume of the cooling package are also important constraints and those are more likely to be satisfied by a direct heat recovery system. However, engine cool- ing remains a critical issue in the field of ICE development. A proper cooling should ensure the engine durability without affecting performances and fuel con- sumption. Therefore, prior to study the implementation of a Rankine cycle on the coolant cycle of an engine, heat transfer phenomena must be accurately known to guarantee a proper engine working.

Assessment of heat transfer rates within the ICE Engine cooling tech-

nologies are as old as the internal combustion engine itself. However, the latter

(19)

Msc Thesis Report March 27, 2014

has not followed the same development pace as the former did. As a matter of fact, heat transfer predictions are still of low accuracy and much dependent on the engine considered. Typically, the cooling jacket shapes are formed by the space left inside the cylinder head and the cylinder block, once the cylinder, the valves and the ports have been packaged. The liquid mass flow rate is usually adjusted empirically by several tests or fixed by rather conservative company standards.

Recent improvement in simulation and measurements tools as well as a better knowledge of convection heat transfer phenomenon give the possibility to analyze more accurately the heat transfer phenomena within the engine. A better knowl- edge of the heat transfer rates distribution within the engine would allow to tune the coolant flow even more accurately.

Flow boiling heat transfer within the engine in order to recover the heat and increase heat transfer rates Once the heat transfer phenomena within the engine are more accurately known, another solution for engine cooling has to be considered. Indeed, coolant mass flow rates required by the typical liquid convective cooling are too high to be either partially or totally evaporated (see section 2.1.2).

A reduction of the mass flow rate has two major advantages. On the one hand, lower flows are more likely to evaporate and their enthalpy will be higher at the output of the engine. On the other hand, the boiling flow heat transfer coefficients are significantly higher in boiling flows. However, dry out has to be avoided when dealing with boiling flows. As a matter of fact, reaching this point would lead to severe decrease in heat transfer coefficient and be thus critical for the engine operation.

1.3 Objectives

Following the problematic of subject, raised in the previous section, the objectives of the project are as follows:

• Build an engine heat transfer model representative for the common range of operation (in liquid convective cooling).

• Narrow the scope of the engine heat recovery to the most appropriated parts for WHR.

KTH MF205X 4 Vincent CHOQUET

(20)

• Integrate the appropriated flow boiling heat transfer correlations in a heat transfer model and study the influence of flow parameters.

• Integrate the engine heat transfer model in a complete Rankine loop model

in order to estimate the potential recovered power.

(21)

Chapter 2 Frame of reference

This chapter aims at defining the scope of this project. After a brief description of the Rankine cycle principles, the necessity of considering engine as an evaporator is justified. Flow boiling heat transfer properties are then presented. Finally, a review of existing evaporative engine cooling and integrated engine waste heat recovery systems is performed, allowing thus to unveil the potential problems of such a technology.

2.1 Rankine cycle concepts for engine waste heat recovery

Rankine cylce is the considered solution for integrated engine waste heat recovery.

In this section, the Rankine cycle principles are firstly described. Then, two solu- tions for recovering directly engine heat are compared.

2.1.1 Rankine cycle principle

Rankine cycle is an idealized thermodynamic cycle composed of one heat source and one cold source and producing mechanical work. Rankine cycles are commonly used in power plants. Nowadays, Rankine cycles represents the most convenient solution in terms of cost, simplicity, efficiency and packaging for recovering waste heat on long haul trucks. Rankine cycles are typically composed of a pump, an expander, a condenser and at least one boiler. The pressurized working fluid is heated in the preheater. Then, the fluid evolves from the subcooled state to the superheated state, accumulating heat from the hot (waste) heat sources. The fluid

6

(22)

is expanded to the condensation pressure in a turbine or a reciprocating expander (see Figure 2.1). The saturated fluid (or the exhaust vapor) is then condensed to the subcooled state and pumped again to a high pressure.

In the TS diagram (Figure 2.2), P

out

stands for the power produced by the ex- pander, P

in

is the power provided by the pump, ˙Q

in

is the heat transferred to the fluid by the hot sources and ˙Q

out

is the heat rejected by the fluid in the condenser.

The efficiency of the Rankine cycle is as follows:

η

_therm

= P

_exp

− P

_pump

Q ˙

_in

(2.1)

Q ˙

_out

˙

m = h

₄

− h

₁

(2.2)

Q ˙

_in

˙

m = h

₃⁰

− h

₂

(2.3)

P

_in

˙

m = h

₂

− h

₁

(2.4)

P

exp

˙

m = h

₄

− h

₃⁰

(2.5)

where h is the enthalpy of the the point number corresponding to the Figure 2.1.

˙

m is the mass flow rate of the working fluid.

2.1.2 Orientation of the study: role of the engine as a heat source

Engine cooling is a significant source of losses. Integrating the engine cooling in the Rankine cycle and thus consider the engine coolant as a working fluid may allow to recover part of the energy losses. The engine can either be considered as hot source for preheating the fluid or for evaporating it.

Only preheating the fluid in the engine is a less constraining way of recovering

heat. Indeed, liquid convective cooling is the typical way of cooling the engine and

the new system would not require much modification. However, one can wonder

how much heat can be recovered from this architecture. As far as the 11L engine

is concerned, in the case of the highest cooling power, 140kW are conveyed to the

coolant. In this case, in the typical conditions of mass flow rate (see Table 2.1),

the necessary energy to evaporate the whole flow would be 8062kW. It goes with-

out saying that neither the exhaust gas waste power nor a potential EGR cooling

(23)

Msc Thesis Report March 27, 2014

Figure 2.1: Turbine and reciprocating Rankine cycles [12].

Figure 2.2: Rankine cycle in a Temperature-Entropy diagram.

KTH MF205X 8 Vincent CHOQUET

(24)

Enthalpy of vaporization of water-ethylene glycol mixture (50/50 in mass at 3bar)

1.45 10

⁶

kJ/kg

Mass flow rate 5.6 kg/s

Energy conveyed to liquid coolant for preheating 140 kW Necessary energy for evaporating the flow 8062 kW

Table 2.1: Conditions in the case of the highest cooling power, 1800RPM, full load

power could evaporate such a flow. To evaporate part of this fluid with either the EGR of the exhaust gases, the coolant should be by-passed with a maximum ratio of about 3 to 4% (according to heat balance tests on the 11L engine). Therefore, most of the energy conveyed from the engine to the coolant would remain unused.

The inefficiency of recovering the heat from the engine directly and keeping the standard liquid convective cooling has led the author to focus its study on the com- bination of evaporative engine cooling and waste heat recovery. Indeed, boiling flows allow both better heat transfer coefficients and a better potential recovery of heat.

2.2 Two phase flow heat transfer phenomena

Two phase flow is the simplest case of multiphase flow. However, its behavior re- mains very difficult to estimate on account of a set additional of parameters such as flow regimes, thermodynamic and transport properties, etc.

Flow boiling is a subcategory of the two-phase flows as it only implies one fluid.

Boiling is usually classified in two category: Pool boiling and flow boiling. The former happens in the absence of bulk fluid flow and the former happens in the presence of it [6].

Pool boiling happens when a motionless fluid, heated by an external source reaches the saturated state of pressure and temperature. Change of state requires an sup- ply of energy for the fluid, called enthalpy of vaporization. Thus, pool boiling enhances heat transfer to the fluid in comparison to mere conduction.

Flow boiling combines both effects of pool boiling and convection. This phe-

nomenon is hence responsible for high heat transfer and shows great potential

improvements in the field of engine cooling.

(25)

Msc Thesis Report March 27, 2014

Flow boiling phenomenology The clearest way to explain flow boiling phe- nomenon is to consider an uniformly heated vertical tube [19] (Figure 2.3). Relative low flow rate of sub-cooled fluid is imposed at the inlet, in order to see the com- plete evaporation of it, from the sub-cooled to the dry out state. Starting from the early sub-cooled state, where the fluid is totally liquid, various flow patterns can be observed. Those patterns have different heat transfer efficiencies.

• Before evaporation, heat transfer is ensured by convective heat transfer to liquid

• At the very beginning of the evaporation process, small bubbles of vapor nucleate at the wall of the tube. This can happen as the surrounding liquid is sub-cooled ( the heat transfer phenomenon is thus called sub-cooled boiling).

In this case vapor bubbles tend to be condensed by the surrounding liquid.

When the liquid is at the saturation temperature, saturated nucleate boiling is happening.

• Evaporation intensifies and the fluid quality reaches higher level. When the bubbles length exceeds the diameter of the tube. The flow pattern is named slug flow.

• The flow turns to be annular when the liquid aggregates as a film around the wall. Heat transfer is maintained thanks to this liquid layer.

• Dry out occurs when the liquid film has evaporated completely. The flow pattern is thus called mist flow, as the liquid part of the flow forms droplets surrounded by vapor. From this part, heat transfer is ensured by convective heat transfer to vapor.

Heat transfer efficiency is much related to the boiling phenomenology. From the beginning of evaporation till the dry out, heat transfer is enhanced by bubble nucleation along the wall and bubble induced convection phenomenon. It is also enhanced by the convective boiling heat transfer to the liquid phase. After the critical point of dry out, film boiling occurs. As the wall is enveloped by a film of vapor, heat transfer gets reduced significantly being ensured by convection to vapor only. The succession of those various stages strongly depends on the flow parameters as well as the tube geometry. Though not being based on this flow phenomenology, flow boiling correlations allows to estimate the evolution of the heat transfer in flow boiling ( see 3.2.3).

KTH MF205X 10 Vincent CHOQUET

(26)

Figure 2.3: Boiling flow heat transfer regimes in a vertical tube. Source : Collier

& Thome (1994)

(27)

Msc Thesis Report March 27, 2014

2.3 Review of integrated engine waste heat recov- ery systems

The present study is akin to a combination of a Rankine cycle with an evaporative engine cooling system. Though such a combination requires additional constraints to the operating parameters (such as saturation presssure and temperature), evap- orating engine cooling is already a great challenge in itself. Therefore, a review of this cooling concept has been performed prior to a literature study of the actual combination of it with a Rankine cycle.

2.3.1 Evaporative engine cooling

The following section aims at detailing previous investigations on evaporative en- gine cooling. First, a historical set of experiments will be described. At that time, engine were not as compact and heat transfers were not as constraining as it is today. Thus, today’s researchers have been focusing on the modeling and the prediction of two-phase heat transfer, in order to avoid any critical heat transfer drop.

Historical investigations Using evaporative engine cooling instead of forced convective cooling is not a brand new idea. The first investigations on this con- cept are attributed to Harrison, in 1927 [13]. After seven years of work on this concept, he managed to get an efficient evaporative cooling, using a condenser instead of a radiator and designing the cooling circuit so that the pump is fed by liquid only. Maintaining an efficient cooling, the main advantages of this new con- cept were a reduced weight of the cooling system, and a better engine warm-up.

In 1964, Tacchela et al. work focused on the cooling circuit layout of the evapora- tive engine cooling system [3]. Thanks to a separator, saturated liquid is recircu- lated directly to the engine block. Gases are directed through the condenser. This process allows to maintain the fluid under saturated state and to have a homoge- neous temperature of the coolant.

Nissan Motor, Nissan Diesel Motor, Vacor implemented their own evaporative en- gine cooling systems showing good results on their respective prototypes. Keeping a similar cooling efficiency, this concept allows to have better warm-up, to reduce friction, to increase the durability of the engine and get fuel economies, according to those three projects, related in various publications [25, 37, 35].

KTH MF205X 12 Vincent CHOQUET

(28)

Cooling through nucleate boiling has also been investigated both experimentally and theoretically by Valeo in the 90’s. These studies showed good results for getting a simplified and more efficient cooling system which let the engine being cooled by nucleate cooling during the most loaded 5% running time of the engine.

During the idle or low load conditions, which are most frequent, the engine is cooled by liquid convective cooling with a cooling system of appropriate dimen- sions [31, 28, 32, 30].

Finally, an interesting work on a four cylinder engine of Lee et al. shows the diffi- culty to remove air from the circuit. As a matter of fact, air is affecting two-phase heat transfer badly. As a non condensible gas, air does not carry as much energy as vapor does. However, one challenging issue of such evaporative cooling systems was to control the pressure within the cooling circuit. A solution was to keep the coolant reservoir vented to the atmosphere, in order to maintain a constant pressure.

More recent work on flow boiling cooling The previous investigations have shown impressive results and did not point out any critical case of heat transfer drops. However, it would certainly be more complicated to apply such a concept on today’s compact engine. In fact, the engine size has been significantly reduced since those times and heat transfers have become even more intensive. Thus, main- taining the coolant flow at a saturated temperature is not sufficient to guarantee good cooling and avoid dry out. The coolant flow rate has to be thoroughly con- trolled, which requires good nucleate boiling predictions.

Nevertheless, due to its high potential in engine mass reduction, fuel efficiency, and durability, nucleate boiling heat transfer is considered as an outstanding solution for precision cooling [29, 21].

Recent researches have consequently focused on the two-phase flow modeling in order to get more accurate heat transfer predictions. The increase of computers capacity and the development of powerful simulation tools contribute also to a better knowledge of the phenomenon.

Heat transfer correlations with test galleries database for a better

prediction model. Most of the research has thus aimed at calibrating a flow

loop with an ICE in order to get accurate correlations of heat transfer within the

(29)

Msc Thesis Report March 27, 2014

engine. As a matter of fact, the commonly used correlations of Chen, Shah and Rohsenow have been based on flow studies in tube or rectangular ducts. This can- not be applied directly to the ICE galleries whose design is much more complex.

Adaptations to those prediction mathematical tools need to be brought and new models of heat transfer prediction have been set [22, 16, 38, 27].

Robinson and Campbell’s researches on nucleate boiling modeling have focused on a benchmark of two heat transfer approaches [29] (Chen and Rohsenow). As Chen correlation was found to be the most accurate of the two investigated, an adapta- tion of this correlation has been done in [22], in order to match the experimental data using a mixture of water and ethylene glycol (50-50 in volume).

H.S. Lee and A.T. O’neill managed to calibrate the boiling curves of the cooling galleries of an ICE with a test loop [17]. The test apparatus is constituted of a rectangular channel, locally heated at the bottom face. Experiments where pur- sued on the same test loop, aiming at investigating the working fluid evaporating behaviors [18]. This study was oriented toward the phenomenology of boiling as the authors took care of avoiding influence of the boundary wall on the boiling flow. The Chen correlation was then calibrated on the test loop data using a mix- ture of Ethylen glycol and water (50-50 in volume) [16].

Nucleate boiling prediction using dedicated measurements. The ex- perimental investigations of M.Amelio et al. [26] focused on the identification of on- set of nucleate boiling. Coolant pressure pulsations, coolant expansion and engine- in engine-out volumetric flow rate comparison have been the proposed methods for this purpose. As the last method is not accurate enough for low flows, the advised methods remains coolant pressure pulsations and coolant expansion measurements.

2.3.2 Engine integrated WHR system

The fundamental and the applications of evaporative engine cooling have been addressed above. However, the scope of the subject is broader: the actual purpose of the study is to recover heat from the engine partially evaporated coolant. As far as the author knows, few investigations have been led in this field.

Such a combination of two-phase evaporative engine cooling and WHR system using HCFC123 refrigerant has already been investigated in 1993 by Oomori and

KTH MF205X 14 Vincent CHOQUET

(30)

Ogino [10]. Without any modification of the coolant path of the engine, fuel efficiency was improved by approximately 3% on bench test.

Further investigations have been led by K.Katta et al. as well as R.K. Jester et al. [20, 7] , without setting the complete integrated engine WHR system. K.Katta et al. interpolated results obtained from an evaporative engine cooling test bench using mixture of water and tetrafluoroethanol (TFE) in [20]. Their calculations led to an efficiency increase of up to 5%. In the article of R.K. Jester et al. [7], a theoretical analysis of various WHR system architectures with water as refrigerant shows that the use of engine as an evaporator and the exhaust as a superheater is best to recover the waste heat. In this study, the ratio of the recovered energy to the total waste heat goes up to 8% for an automobile engine.

As far as WHR systems using engine cooling as heat source are concerned, a recent

patent from Faurecia shows also an increasing interest in recovering such a waste

heat [8].

(31)

Chapter 3 Description of the heat transfer model

As enlighted in the master thesis objectives, section 1.3, this project has evolved step by step toward the evaluation of the Engine Integrated Waste Heat Recovery system. A sketch of the overall investigations is represented Figure 3.1

First, a complete study of heat transfer was led using the software GT-suite as a simulation tool. Both coolant side and combustion side of the cylinder were considered.

On account of a highly constrained cylinder head water jacket, it was decided to reasonably narrow the scope of the project to the liner wall heat exchange. Then a more adapted model of the liner wall heat exchange was created on Simulink.

This model has been calibrated using data from tests and the study could then be oriented toward flow boiling.

Finally, the previous liner heat exchange model was integrated in a full Rankine cycle model on Matlab, in order to estimate the best operating conditions and the gains of such a WHR system.

3.1 Heat transfer in ICE model using GT software

After a brief description of the software and its attributes, the engine heat transfer model is presented in details.

16

(32)

Figure 3.1: Overall structure of the investigations.

(33)

Msc Thesis Report March 27, 2014

Engine performance modeling (GT-POWER) Analysis of measured engine cylinder pressure

Acoustics of intakes and exhausts Exhaust after-treatment Sil, HiL and Real-Time simulation Vehicle dynamics (drive cycles, drivelines)

Transmission Modeling

Hybrid and electric vehicles, fuel cells Engine cooling

Vehicle thermal management Vehicle energy management

Underhood cooling module analysis (3-D with COOL3D) Air conditioning

Exhaust heat recovery Lubrication circuits and bearings

Fuel injection systems Hydraulics and pneumatics

Valvetrain/Camshaft kinematics and dynamics

Cranktrain kinematics and dynamics (incl. balancing and bearings) Gear, chain and belt drives

Table 3.1: GT-Suite technical Applications

3.1.1 GT suite

GT suite is a product from Gamma Technologies focusing on Engine and Vehi- cle Simulations. In a single software, one can create an integrated vehicle model, taking into account interactions between the various system parts. This software includes the technical applications represented in Table 3.1. The present study has focused on the use of the Engine performance and the Vehicle thermal man- agement applications.

GT suite is not only a cost-cutting and quick tool for engine development. It constitutes also an appropriate tool for studying innovative concepts and estimate data that would not be available by mere engine testing. The underlying purpose of this study is to create a complete cylinder heat transfer model in order to assess the distribution of the heat fluxes within the engine. In the scope of this project, this model has only been used as a reference for studying the engine waste heat recovery concept. However, this model could easily be integrated in an advanced complete thermal management study coupled to an engine model. Indeed, GT is a user friendly platform that allows engineers to get hands on a model quickly.

KTH MF205X 18 Vincent CHOQUET

(34)

GT for thermal management using finite element discretization of the cylinder GT suite library enables its users to model heat transfer fluxes through the walls with a finite element discretization of the cylinder. This can be achieved by using the objects EngCylWallSoln and EngCylStructCond. The former only requires fixed boundary conditions on the coolant side and it is therefore typically used for "stand-alone" engine models. The latter is a more advanced model of the cylinder wall, allowing to couple the combustion process to the coolant and oil circuits models. It can also be part of a stand-alone steady model of the cooling system, by using a fixed in-cylinder boundary conditions. The advanced element has been chosen in order to accurately estimate heat flows within the engine and the interaction between both sides.

The cylinder structure is generated from the geometric parameters and gives a finite element representation of the cylinder liner, head, ports, valves and piston.

3.1.2 Assumptions

In order to have a light model while keeping relevant results, major assumptions have been made.

Assumptions inherent to the GT thermal management application

• 1D-flows in pipe have been considered

• Heat transfer between the FE discretization of the cylinder has been dis- cretized in set of available convective ports

• Engine block, Cylinder head block and oil pan are modeled as thermal masses

Additional assumptions of the implemented model

• Liner and Head water jackets are represented by tubes whose hydraulic di- ameter is equivalent to the reality

• Pressure drop has been neglected in the cooling galleries

• Valve guide and Valve seat heat transfers have been neglected

(35)

Msc Thesis Report March 27, 2014

3.1.3 Model architecture

The model works as a combination of an engine, working with an explicit solution method and the cooling circuits, including oil and engine coolant, solved with an implicit method. The global architecture of the engine is shown Figure 3.2, where the engine cylinder element on the engine side is linked by "multi thermal connections" to the finite element object on the coolant side.

Figure 3.2: Engine and cooling circuits coupled model overview.

Engine Side Model The engine considered for this study is a heavy duty 11L turbocompressed diesel engine satisfying the Euro VI european standards for pol- lutant emissions. The geometrical attributes of this engine are reported in the chart 3.2. The stand alone engine model has been developed by the Combustion and Simulation group of Volvo Powertrain and it is fitted with an user model for the combustion simulation (see Figure A.1, in appendix A). As far as the cylin- der wall temperature solver is concerned, the EngCylStructCond object has been replaced by the EngCylWallSoln one. As this engine model is already calibrated, changes to the initial model have been avoided as much as possible in order not to affect the engine operating conditions.

KTH MF205X 20 Vincent CHOQUET

(36)

No. of cylinders 6 Engine Capacity 11 L Engine Max. Power 460hp

Displacement 10.8 dm3

Stroke 152 mm

Bore 123 mm

Compression ratio 17.0:1

EGR No

Turbocompressed Yes

Table 3.2: Engine Characteristics of the 11L diesel engine considered

Coolant Side Architecture The coolant side of the model has been built from scratch and is constituted of six identical cylinder elements and their own coolant/oil circuits, see Figure A.2 and 3.3.

The elements used and their attributes are detailed in Appendix A.

• EngCylStructCond Element

The finite element discretization of the cylinder allows to model heat transfers on the coolant side. When coupled to an engine model, convective heat transfer can be taken into account by various ports, see Table 3.3. For each of those ports, the convective heat transfer coefficients can be fixed or calculated automatically from the flow characteristics of the flow element it is linked to. The choice of the heat transfer coefficients as well as heat transfer area is explained in appendix A.

• Tube element for modeling galleries

Tube elements are linked by convection connections to the EngCylStruct-

Cond Element. For simplicity’ sake, the coolant galleries has been divided in

two parts, the liner and the head galleries. Though not representative of the

physical parts, the water jackets have been modeled by simple tubes whose

diameter has been to tuned get the right Reynolds number, and thus, the

right convective heat transfer coefficient. As a matter of fact, this simple

approach allows to get an overview of the influence of the coolant flow on

the heat transfer phenomena within the liner and head galleries. However,

it would not allow a local study of the heat fluxes through the wall and a

more detailed approach using CFD should be considered for studying heat

transfer to coolant.

(37)

Msc Thesis Report March 27, 2014

Liner Oil Piston Oil Head Coolant Port1 Coolant Port2 Coolant Port3 Coolant Port4 Coolant Port5 Coolant Valve 1 Guide Coolant Valve 2 Guide Coolant Valve 3 Guide Coolant Valve 4 Guide Coolant Valve 5 Guide Coolant Valve 1 Seat Coolant Valve 2 Seat Coolant Valve 3 Seat Coolant Valve 4 Seat Coolant Valve 5 Seat Coolant

Table 3.3: Convective ports available with EngCylStructCond element (Limited to the use of 2D water jacket model)

• Other Elements

Thermal masses, conduction phenomenon and radiation phenomenon have minor impacts on the engine heat transfer to the atmosphere. However, they should not be neglected when calibrating the engine with global heat balance tests (TC313).

3.1.4 Correlations used for heat transfer model, Cylinder Gas Side

To get heat transfer data from an engine is usually costly and time consuming.

Therefore, one has to use correlations in order to model heat transfer within the cylinder. Those correlations are usually obtained from several experimental data and they allow to estimate with a certain degree of accuracy the extrapolated values. The highest source of heat transferred to the cylinder wall is due to the convective heat transfer [14]. In the case of compression ignition engines, radiation is also a significant source of heat due to the presence of an important amount of soot particles. This phenomenon tends obviously to attenuate for modern engines, which have more stringent emission standards.

KTH MF205X 22 Vincent CHOQUET

(38)

Figure 3.3: Coolant Side of the model, zoom on one cylinder.

(39)

Msc Thesis Report March 27, 2014

Then comes indirect sources of heat to the wall. Those can either be conduction from the piston to the liner through the skirt and the rings. This can also come from conduction between the valves and the cylinder wall.

Finally comes the friction heat, stemming from the rings and the skirt motion along the liner.

In-Cylinder Convection and Radiation Many experimental investigations have been led over the past fifty years in order to build correlations for gas to wall heat transfer coefficients. These work have mostly been based on Nusselt correlations for convection inside circular pipes.

However, these correlations have most often extensively include both convection and radiation phenomena. One must make distinctions between three types of heat transfer correlations [14] :

• Correlation for time averaged heat transfer coefficients

• Correlation for instantaneous spatial averaged coefficients

• Correlation for instantaneous local coefficients

Correlations for time averaged heat transfer coefficients allows to estimate the overall heat transfer rate to the coolant. This correlation is therefore mainly aimed to design the cooling system based on the engine heat loading rather than advanced studies for estimating heat fluxes distribution within the cylinder. Moreover, ac- cording to Sheyler et al. [33], those correlations often neglect heat transfer to oil or friction and always ignore indirect heat transfer.

Instantaneous local coefficient correlation such as Lefeuvre et al. and Dent et al.

ones, have been developed taking into account the swirling flows speed distribution within the cylinder. As far as the considered engine is concerned, swirl is so low that it can be neglected.

Correlations for instantaneous spatial averaged coefficients have thus been consid- ered, being the most appropriated for this study. As a matter of fact, the early work of Eichelberg (1939) has been progressively improved by Annand (1963), Woschni (1967), Sitkei and Ramaniah (1972) and Hohenberg (1979) whose corre- lations give relative accurate results according to Finol’s review on heat transfer correlations [4].

Being based on an experimental dataset using DI diesel engines with dimensions

KTH MF205X 24 Vincent CHOQUET

(40)

similar to the considered engine, Hohenberg is the best suited among these correla- tions. His formulation take into account both radiation and convection phenomena.

As a matter of fact, the former is very likely to increase due to the soot layer on the walls and the latter will tend to decrease due to the same phenomenon. Despite of this effect, Hohenberg pointed out that the heat fluxes stabilized after one hour of engine operation.

Hohenberg Correlation for heat transfer coefficient (h) [15]:

h = C

₁

V

^−0.06

p

^0.8

T

_g^−0.4

(V

_p

+ C

₂

)

^0.8

(3.1) C

₁

and C

2

are calibration constants given as

C

1

= 130 (3.2)

C

₂

= 1.4 (3.3)

This simple formulation gives the heat transfer coefficient h as function of the instantaneous cylinder volume V , the instantaneous cylinder conditions p and T

g

andthe mean speed V

p

:

V

_p

= 2LN

60 (3.4)

where N is the engine speed (RPM) and L is the engine stroke.

As the initial engine model was based on the Woschni Correlation, one must check the impact of changing one for another on the heat transfer coefficients. As a result, Woschni and Hohenberg heat transfer coefficients are plotted Figure 3.4 and show similar results. The peak difference between those two curves is of 2.9%

and the shift is of 13.89 deg. This difference has to be taken into account for the later model comparison.

Although the heat transfer coefficient is a spatially averaged one, the exposition time to the cylinder gases at the top of the liner wall differs from the bottom of liner wall one. Therefore, one has to take into account the piston motion when estimating convection heat transfer to the liner, see Figure 3.5.

More recent correlations for modern diesel engines could also have been used.

However, those correlations have not yet widely been validated by tests. They can

thus not be fully trusted.

(41)

Msc Thesis Report March 27, 2014

Figure 3.4: Evolution of the heat transfer coefficient as function of crank angle for Hohenberg Correlation and Woschni correlation (adapted by GT) at 1800 RPM and full load.

KTH MF205X 26 Vincent CHOQUET

(42)

Figure 3.5: Evolution of the heat transfer coefficient as function of the distance from cylinder top. Hohenberg Correlation.

Ring-Liner and Skirt-Liner Conduction Ring-Liner and Skirt-Liner con- duction accounts for significant heat fluxes due to the high difference of tempera- ture between the piston and the liner wall. As a matter of fact, this difference can reach up to 80 K in the case of the study.

Following the Fourier’s law of heat conduction (3.5), and neglecting the convec- tion heat transfer due to oil motion, Finol [9] estimated heat transfer between the piston and the liner by using Cameron’s correlation for oil thickness 3.1.4. In this case, the oil film is assumed to be the only responsible for conduction between the piston and the liner wall.

Fourier’s law of heat conduction in one dimension : Q = −A ˙

_r

k

_o

T

_p

− T

_w

e

_o

(3.5)

This widely used law of heat conduction depends on the surface of contact

between the cylinder and the ring or the skirt A

r

, the oil conductivity k

o

and the

oil thickness e

o

.

(43)

Msc Thesis Report March 27, 2014

Deduction of the oil film thickness, Cameron [5] :

e

_o

= h(6W

^∗

)

^1/2

( µU πB

W T )

^1/2

(3.6)

Cameron formulation for calculating the oil film thickness e

o

depends on the seg- ment height h or the skirt height respectively and 6W

^∗

which is a non-dimensional load defined by Cameron. 6W

^∗

depends on the segment geometry profile. The value taken is 0.1353, 0.1447 and 0.05979 for the compression rings, the oil control ring and the skirt respectively. The oil viscosity has been calculated at the aver- age operating oil temperature. The lateral thrust W T depends on the ring normal force or the skirt thrust toward the wall.

The ring normal force is due to the ring pretension. As far as the top ring is concerned, cylinder pressure accounts its the normal pressure onto the wall. This normal pressure is calculated as follows:

F

_normal1^st_ring

= π · B · e

₁

· p + 2 · F

_tang₁

(3.7)

F

_normal2ndring

= 2 · F

_tang₂

(3.8)

F

normaloilring

= 2 · F

_tang_o

(3.9)

The skirt thrust F Y

skirt

is calculated from friction force experimental data, see section 3.1.4.

Adapted to the model, the predicted oil film thickness is plotted Figure 3.6, at 1800 RPM and for a given set of ring pretension. The oil film thickness correlation allows then to estimate the heat transfer coefficient between the rings/skirt and the liner. The Figure 3.7 shows the evolution of the Fourier heat transfer coefficient for the separate rings and the skirt. An averaged heat transfer coefficient is then deduced from Figure 3.8 for implementation as a parameter in the GT model.

Ring-Cylinder Friction Friction accounts for important losses in an engine.

This can vary from a 2% at full load in a modern engine up to 100% at idle or no-load. Most of the coolant losses appear as heat conveyed to oil and coolant, it has thus to be taken into account for the engine heat transfer model. The friction work is defined by J.B Heywood [14] as " the difference between the work delivered to the piston while the working fluid is contained within the cylinder and the usable work delivered to the drive shaft" .

KTH MF205X 28 Vincent CHOQUET

(44)

Figure 3.6: Evolution of the oil film thickness as function of the CAD.

Figure 3.7: Evolution of the Heat transfer coefficient as function of the CAD,

1800RPM Full load.

(45)

Msc Thesis Report March 27, 2014

Figure 3.8: Evolution of the average heat transfer coefficient between the piston and the liner.

Friction losses includes thus Pumping, Piston-crank assembly, fuel-injection pump and camshaft losses. Since the implemented model focuses on heat transfer to the cylinder wall, only Piston-Liner friction has been taken into account.

Using previous work on ring/skirt friction for the considered engine, the author has chosen a simple but already calibrated friction model. This model assumes a hydrodynamic lubrication regime, which is relevant according to Stone demon- stration using Stribeck curve [34]. The friction coefficients µ are functions of the piston speed. Those formulas are calibrated with the calibration coefficients µ

k

, µ

_s

, A

s

and A.

µ

_skirt

= A

_s

|v| (3.10)

and

µ

_ring

(v) = µ

_k

+ (µ

_s

− µ

_k

)exp(A|v|) (3.11) .

The friction force of the rings is calculated respectively as a product of the friction coefficient and the ring normal force presented in the previous section 3.1.4.

KTH MF205X 30 Vincent CHOQUET

(46)

F f

_ring

= µ

_ring

F

_normal

(3.12) As far as the skirt friction force is concerned, it is calculated from the skirt thrust F Y

skirt

applied by only one side of the piston. This lateral load is calibrated from experimental data obtained from tests on a 11L liter Volvo engine.

F f

_skirt

= µ

_skirt

F Y

_skirt

(3.13)

The evolution of the friction power losses as function of the CAD is presented in Figure 3.9. However, the focus being placed on heat transfer to the liner wall, this average friction power has been plotted as function of the distance from the cylinder top on Figure 3.10. This represents the average linear friction power for a whole cycle. However, for later investigations, only half of this heat has been con- sidered to be transferred directly to the liner. Finally, GT only takes into account average friction losses into account. Therefore, the average friction losses has been represented Figure 3.11 for various engine speed at full load.

Figure 3.9: Evolution of the friction power losses as function of the CAD [deg].

(47)

Msc Thesis Report March 27, 2014

Figure 3.10: Evolution of the friction losses power as function distance from the cylinder top.

Figure 3.11: Evolution of the averaged friction losses power as function of the engine speed [RPM] at full load.

KTH MF205X 32 Vincent CHOQUET

(48)

3.1.5 GT Model Calibration

The model calibration focused on the set of operating points shown in Table 3.4.

Engine Speed [RPM] Torque [N.m] Power [kW]

2150 1010 227.3

1800 1764 332.3

1500 2040 320.3

1300 2144 291.7

1200 2125 266.9

950 2169 245.7

Table 3.4: Operating points considered for the calibration

The complete model has been calibrated using the experimental data of the 11L engine considerd as listed below.

• Cylinder head thermal mapping

• Piston thermal load simulation study

• Thermal measurements on liner

Those data come from various experimental investigation and their initial purposes were not to build such a model. However, such experimentations are very costly and time-consuming. It goes without saying that this model would become more robust with an appropriate single experimentation.

Calibration of the model using the cylinder head temperatures Cylinder head thermal mapping tests aim at measuring cylinder head temperatures at the valve decks using thermocouples (see comparison points on Figure 3.12). The temperature at the surface of the valve decks is compared with the GT power results in Figure 3.13. The difference between the actual tests and the simulation is plotted on Figure 3.14.

Calibration of the model using the liner temperatures Liner tempera-

tures tests on the 11L engine have been performed at Lyon for the US7, Eu5 and

Eu4 engines. Unfortunately, no data are available for the Eu6 engine. However,

temperature measurement differences between those engines do not differ more

than 5K. Moreover, it has been noticed that the parameters most influencing the

(49)

Msc Thesis Report March 27, 2014

Figure 3.12: Nodes for GT FE analysis and the thermocouples postition for tests.

Figure 3.13: Simulated temperatures and measured temperatures as function of the operating point (full load).

KTH MF205X 34 Vincent CHOQUET

(50)

Figure 3.14: Difference between simulated temperatures and measured tempera-

tures for various operating points (full load). E=Exhaust, I=Intake.

(51)

Msc Thesis Report March 27, 2014

liner temperature are the coolant flow rate, the coolant temperature, the fuel in- jected rate and, to a lower extent the start of injection. Relying on the influence study of those parameters led in those experiments, the Eu6 engine liner temper- atures have been estimated for the operating points considered. They have been next compared to the results obtained with the simulated model average wall tem- peratures, Figure A.3. Thermocouples measurement cannot practically be limited to one point but rather to a local domain of the material whose temperature is measured. Therefore, it has been preferred to compare the simulated temperature averaged over the thickness with the test results, see Figure 3.15.

One can notice a local peak of wall temperature at a distance of approximately 18 cm from the cylinder top. This peak is likely to be not relevant in comparison to literature results from Woschni [36] and Finol [9]. In fact, this peak lies below the bottom dead center and below the end of the liner water jacket. Thus the wall is only cooled by oil impingement cooling, which is hardly modeled in GT. Liner oil cooling is actually modeled by a uniform heat transfer coefficient.

Calibration of the piston temperature The piston geometry attributes is defined by a set of parameters in GT such as the piston cup thickness, the piston height, the ring thickness, the skirt thickness and so on. However, this geometry definition is quite limited and does not allow to predict accurately the piston tem- perature distribution. Indeed, major features such as the oil cooling gallery cannot be modeled on GT. The possibility to implement its own piston geometry may be available in the future versions

¹

.

Piston temperature distribution results on GT are thus not representative. As a matter of fact, the peak temperature is typically reached at the bowl rim, due to the fuel injection jet orientations. As this feature is complex to take into account, the highest piston temperature is reached at the furthest point from the cooling interfaces of the piston, see Figure 3.16, in the case of 1800RPM, full load.

Thermal experiments on piston for the 11L engine provide the temperature values on the bowl thanks to measurement with thermocouples. They provide also tem- perature measurement on the piston second land (between the first ring and the second ring) and on the skirt top.

The second land temperatures have thus been used for piston model calibration on GT. A steady value as been considered for it. Unfortunately, the skirt tem- peratures were impossible to match due to this bad representation of the piston