Decoupled Design of Auxiliary Systems for Internal Combustion Engines

(1)

Link¨oping University | Department of Management and Engineering Master Thesis, 30 credits | Aeronautical Engineering Autumn 2015 | LIU-IEI-TEK-A–15/02410–SE

Decoupled Design of Auxiliary

Systems for Internal Combustion

Engines

Author:

Niklas Blomgren

Supervisors: Fredrik Henriksson

Johan Renner

Andreas Thomasson

Examiner:

Liselott Ericson

(2)

(3)

Copyright

The publishers will keep this document online on the Internet – or its possible re-placement – from the date of publication barring exceptional circumstances.

The online availability of the document implies permanent permission for anyone to read, to download, or to print out single copies for his/hers own use and to use it unchanged for non-commercial research and educational purpose. Subsequent trans-fers of copyright cannot revoke this permission. All other uses of the document are conditional upon the consent of the copyright owner. The publisher has taken tech-nical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against in-fringement.

For additional information about the Link¨oping University Electronic Press and its

procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/

c

(4)

Abstract

This thesis investigated if decoupled design of the air intake and exhaust systems for four-stroke internal combustion engines is possible. Using the information found design guidelines were set up for the formula student team ELiTH Racing.

The literature study revealed that the systems are not uncoupled, and the influence of exhaust geometry on air intake behavior needed more thorough investigation.

Experiments were designed, using a single cylinder engine with simple intake and exhaust geometries. The tests were attempted, but had to be abandoned due to time constraints. Successful tests would have yielded results in the form of pressure measurements, from a Prandtl-tube, in the air intake, and footage of smoke tests.

As a secondary task the potential of computer simulations during the design process was investigated, which yielded a suggestion on how to set up a complete reasonable computational model of the systems.

This also resulted in that the design guidelines included how to use computer simulations for the design process.

(5)

Acknowledgments

During this thesis I have had a lot of help from other people. I wish to thank Hugo Johansson for lending me his scooter, Thomas Larsson, Peter Karlsson and Per Johansson for all their help with manufacturing and other support, Anders Bj¨ork for his help with the rig for the hammer drill, Aevan Nadjib Danial for helping me with the test runs. Fredrik Henriksson, Johan Renner and Andreas Thomasson are thanked for all their support. Lastly I wish to thank ELiTH Racing for setting up this project.

(6)

List of Figures

2.1 Schematic of the energy transfer from an engine to the driving wheels on

a car. . . 6

2.2 The main events of the four stroke cycle. ( Wikimedia commonsCC with added text). . . 7

2.3 The ideal thermodynamic cycles. The upper graphs show the Otto cycle, and the two lower the diesel cycle. [5] . . . 8

2.4 Schematic of the air intake system. . . 10

2.5 The air intake of ER15. The non-labeled parts in the background are fuel injectors. (Source: ER15 CAD-Model, with permission.) . . . 10

2.6 Schematic of the exhaust system. . . 11

2.7 The exhaust system of ER15. (Source: ER15 CAD-Model, with permis-sion.) . . . 11

2.8 The effect of a decrease in duct diameter. . . 12

2.9 The basic look of the velocity profile of the flow, close to the wall. . . 13

2.10 Difference between inviscid and viscous flow around a cylinder. (Inspired by Figure 3.7 in Flight Physics, Torenbeek E. and Wittenberg H.). . . 13

2.11 Illustration of flow through a bend. (Inspired by Idelchik, 1986). . . 15

2.12 The principle of a manometer. (Inspired by wikimedia commons). . . 15

2.13 The principle of a pitot-tube. . . 16

2.14 The principle of Prandtl-tubes. . . 16

2.15 Photograph of an airflow visualized using smoke. [12] . . . 17

3.1 Photographs of the spur gears. The pinion, shown in the two upper photos, was placed on the crank shaft, while the gear was used in the hammer drill. . . 23

3.2 Photograph of the pinion mounting. . . 24

3.3 Photograph of the hammer drill mounting. . . 24

3.4 Top, bottom and side view of the adapter plate. . . 24

3.5 Left: The chainsaw centrifugal clutch. Middle and right: The adapter plate mounted on the chainsaw. . . 25

3.6 The centrifugal clutch on the scooter. Note the three holes, in which the screws on the adapter plate were inserted. . . 25

3.7 Photographs of the PVC-tubes. On the left is the tube to the intake port, and on the right is the tube from the exhaust port. . . 26

3.8 Photographs of the air intake and test section. . . 27

3.9 Photograph of the reference tube. . . 28

3.10 Photograph of the step geometries. . . 28

(9)

3.13 Barometer. . . 30

5.1 Schematic of suggestion for computational model. . . 39

A.1 The vacuum former. . . 44

A.2 The CNC wood mill. . . 45

A.3 The mold for the air intake. The piece of paper on the side was used to cut out the shape from the plastic sheets. . . 45

A.4 The mold for straight tubes. . . 45

A.5 The molds for the step geometries. . . 45

A.6 The molds for the bends. . . 46

A.7 Photo of one end of the reference geometry. It shows the cross-section of the tubes is not perfectly round. . . 47

B.1 Image of the drawing for the gear assembly used on the hammer drill. Image scale 0.5. . . 48

B.2 Image of the drawing for the gear used on the hammer drill. Image scale 0.5. . . 49

B.3 Image of the drawing for the gear pin used on the hammer drill. Image scale 0.45. . . 49

B.4 Image of the drawing for the pinion assembly. Image scale 0.5. . . 50

B.5 Image of the drawing for the pinion. Image scale 0.5. . . 50

B.6 Image of the drawing for the socket. Image scale 0.5. . . 51

B.7 Image of the drawing for the disc with splines. Note that this is not a drawing of the actual plate which was used. Image scale 0.45. . . 51

C.1 Image of the assembly drawing for the chainsaw adapter plate. Image scale 0.5. . . 52

C.2 Image of the drawing for the plate mounted on to the scooter rear wheel drive shaft. Image scale 0.7. . . 53

C.3 Image of the drawing for the plate screwed on to the chainsaw drive shaft. Image scale 0.7. . . 54

C.4 Image of the drawing for the threaded pins used to screw the two plates together. Image scale 0.7. . . 55

C.5 Image of the drawing for the distances used to stiffen the construction. Image scale 0.7. . . 56

(10)

List of Tables

1.1 Point distribution of the competition events [1]. . . 1

3.1 Geometrical properties of the gears. . . 23

(11)

Nomenclature

Quantity Explanation Unit

q Dynamic pressure Pa ρ Density kg/m3 u Flow speed m/s pa Ambient pressure Pa ps Stagnation pressure Pa p Pressure Pa h Height m g Gravitational acceleration 9.81 m/s2

µ Dynamic viscosity Pa·s

L Typical length m ν Kinematic viscosity m2/s ω frequency rad/s Ψ Amplitude m k Wave number -x Distance m t Time s Abbreviation Explanation FR Functional Requirement DP Design Parameter M Mach number Re Reynold’s number

CFD Computational Fluid Dynamics

DNS Direct Numerical Simulation

LEE Linearized Euler Equations

RANS Reynold’s Averaged Navier-Stokes equations

CAA Computational AeroAcoustics

TDC Top Dead Center

BDC Bottom Dead Center

rpm revolutions per minute

N Number of cylinders

cc cubic centimeters

1D 1-dimensional

3D 3-dimensional

(12)

Chapter 1 Introduction

LiU Formula Student is a student association at Link¨oping university. The association

builds one formula car per year and competes in formula student contests. During the scope of this thesis the association was named ELiTH Racing, and is therefore be re-ferred to by that name.

Formula student is a design competition, where the cars are evaluated on design solu-tions, manufacturing, presentation, economy and car performance. The competitions are divided into events, each of which can give a certain amount of points, as seen in table 1.1. The events are divided into two types, static and dynamic events. Static events includes presentation of the car, its design and an idea of how to mass produce it. The dynamic events tests how well the car performs when it is driven.

The numbers presented are maximum points for each event. Only the winner of an event receives maximum points. All other competitors receive points in relation to how well their designs stand up to the winner.

Static events: Presentation 75 Engineering design 150 Cost analysis 100 Dynamic events: Acceleration 75 Skid-pad 50 Autocross 150 Efficiency 100 Endurance 300 Total: 1000

Table 1.1: Point distribution of the competition events [1].

As can be seen in table 1.1 the static events make up about one third of all points to be earned, and almost half of those points are earned through the engineering design. The scoring of this event is not only determined by how well a certain design works, but also by how well the design choices are motivated. This means it is worth spending time to think about how the various parts of the car are designed and integrated.

(13)

The cars are designed and produced by the students during the course of approximately one and a half year, which gives rise to many challenges due to the limited amount of time.

When designing air intake and exhaust systems for an internal combustion engine, nor-mally the air intake is designed first and then the exhaust system. The two systems are then tweaked to increase performance. For an organization like Elith racing this is very time consuming, and therefore designing them simultaneously would be preferable. However, functionality of the car must be prioritized over having a decoupled design. This report aims at explaining the design rules set up for Elith racing, in such a way that no previous knowledge of the systems or engines should be required. However, some understanding of mathematics and physics, in particular mechanics, is needed.

1.1 Problem description

The problem consisted of setting up design rules for designing air intake and exhaust systems, in such a way that they may be designed separately and simultaneously, without affecting each other.

The main question for this study was: For each system are there any changes that can be done, which do not have significant negative effect on neither the other system nor the engine performance?

System refers to air intake and exhaust respectively.

• What changes to the air intake have impact on the downstream systems? • Are the upstream systems affected by the geometry of the exhaust? • How should computer simulations be used in the design process?

1.2 Objectives

The objective was to develop a set of design rules, appropriate for Elith racing, with primary focus on decoupled design of the air intake and exhaust system.

Regardless of whether decoupled design is possible or not, the study was going to result in a document containing design rules.

As a secondary objective the use of computer simulations for the design process was investigated.

(14)

1.3 Limitations and delimitations

• No electronic control devices are discussed. • Catalytic converters are disregarded.

• Experiments focused on changes in exhaust geometry only. • Plastic tubes were to be used for experiments.

• Engine was never started.

• Only pressure would be measured.

1.4 Outline

The thesis begins by introducing Elith racing and formula student, to clarify the pur-pose, and then the problem and objectives. The theory chapter includes what decoupled design is, the physics of the systems and basics of how computers are used to simulate the physics. This chapter’s structure is similar to the results chapter. In the method chapter it is explained how the literature study was performed and how experiments were designed. More specific information on how the systems and computer simulations work, and what the experiments showed is presented in the results chapter. Following is a discussion about the theory, results, possibilities of achieving decoupled design and the potential of computer simulations. The report finishes with some conclusions and suggestions for future works. The appendices contain detailed information on manufac-turing processes, safety concerns for the experiments and the conclusions in the form of design guidelines.

(15)

Chapter 2 Theory

To investigate the possibility of decoupled design it is necessary to understand the physics of the air intake and exhaust, when attached to a four-stroke engine. In turn it becomes necessary to understand how a four-stroke engine works, how air flow in the ducts work, thermodynamics and, for reasons explained below, acoustics. However, first an explanation of what is meant by a decoupled design is required.

2.1 Axiomatic design

Axiomatic design is a mathematical methodology used to obtain a design framework. The goal is to define a design matrix and two vectors, satisfying two axioms. The vectors are functional requirements, FR, and design parameters, DP. One of the central ideas of axiomatic design is to distinguish between what is to be achieved and how it is to be achieved. [2]    F R1 F R2 F R3    =   A11 A12 A13 A21 A22 A23 A31 A32 A33      DP1 DP2 DP3    Framework of axiomatic design.

Axiom I, the independence axiom: Maintain the independence of the functional require-ments.

This axiom means that the design parameters should be uncoupled. In case this is not possible the design matrix should be made lower triangular, by partitioning. A lower triangular matrix means there is no backwards influence from changes in the design pa-rameters or functional requirements, as long as the changes are performed in a certain order. A coupled matrix which cannot be partitioned this way is called a full matrix, and should be avoided. [3]

Axiom II, the information axiom: Minimize the information content of the design. This axiom tells us that if axiom I is satisfied the complexity of the design should be kept to a minimum.

(16)

There are three design modes which can be distinguished:

Uncoupled - One design parameter and one functional requirement can affect each other, in only one way.

Decoupled - The design matrix is lower triangular.

Coupled - All functional requirements are influenced by all design parameters. [2]

Uncoupled: F R1 F R2 =x 0 0 x DP₁ DP2 Decoupled: F R1 F R2 =x 0 x x DP1 DP2 Coupled: F R1 F R2 =x x x x DP₁ DP2

Basic frameworks of the axiomatic design modes.

The methodology is a way of keeping things as simple as possible. However, it is not always possible or appropriate. [3]

For example in this thesis the functional requirements could be pressure in the air intake and pressure in the exhaust. The design parameters could be air intake and exhaust duct dimensions, e.g. length. The sought after framework is then

pA pE =A11 0 A21 A22 lA lE

In order to see if it is possible to create a design, which satisfies this framework, it would be necessary to investigate the effects of duct lengths on the pressure in both the air intake and exhaust.

(17)

2.2 Internal combustion engines

2.2.1 Basics

The purpose of the engine is to transform chemical energy into mechanical energy, through combustion of fuels. This is achieved by igniting the fuel in such a way that the explosion causes a piston to move up and down inside a cylinder. The piston is attached to a crank shaft, which transfers the mechanical energy to the drive shaft which in turn rotates the wheels on the car. This is show in figure 2.1

Figure 2.1: Schematic of the energy transfer from an engine to the driving wheels on a car.

In order to obtain an optimized explosion it is necessary to control the amount of fuel and the amount of air that mixes before ignition.

The amount of air is controlled by intake valves on the cylinders, which in turn are controlled by a camshaft, and the amount of fuel is controlled by a fuel injector. The fuel-air mixture is ignited by spark plugs, glow plugs, or in some cases laser.

The timing of the valves relative to the pistons is normally controlled using a timing belt or chain. On newer engines the spark is often electronically controlled, but on older engines it is controlled by the timing belt and a distributor.

The hot gases formed by the explosion are pushed out through the exhaust port. The exhaust valve is controlled by a camshaft.

2.2.2 The four stroke cycle

The transfer of energy from burning fuel to rotating the crank shaft is described by a four stroke cycle.

In an engine cylinder a piston moves up and down, while rotating the crank shaft. The motions are powered by the explosion, except for when starting the engine. In order to start an external cranking device is needed. This device is called a starter.

The process includes intake of air, compression, igniting the fuel-air mix, expansion and exhaust. The cycle is considered to start from when the piston is at the highest point in the cylinder. This is known as Top Dead Center (TDC), which is seen in the first image in the upper row in figure 2.2. The cycle starts with the air intake stroke, second image in the upper row in figure 2.2. Regardless of which type of engine is used air is injected into the cylinder. For Otto-engines fuel is also injected and mixes with the air during the stroke. The second stroke begins when the piston has reached its lowest position, known as Bottom Dead Center (BDC), and starts moving up. The air, or fuel-air mix,

(18)

image in the upper row in figure 2.2. For a diesel engine the fuel is injected during the end of this stroke. Once the piston is back at TDC the fuel-air mix is ignited, causing the start of the third stroke, called expansion or power stroke. See the two first images in the bottom row of figure 2.2. The last image in figure 2.2 shows the exhaust stroke. When the piston is at BDC the exhaust valve opens, and the exhaust gases are pushed out of the cylinder by the piston.

Figure 2.2: The main events of the four stroke cycle. ( Wikimedia commons withCC

added text).

Between the end of each cycle and the beginning of the next cycle there is a short overlap period, during which both the exhaust- and intake valves are open at the same time. This causes the two systems to affect each other, as well as the engine performance, in different ways. One effect is reduced pressure from the intake, caused by the leakage into the exhaust. This effect must not be too large, or some fuel is pushed out of the cylinder instead of being ignited.

The overlap is necessary, because with no valve overlap the exhaust flow would be reduced, causing some of the combustion residuals to remain in the cylinder. Also, at the time for exhaust valve closing more residuals may be blown back into the cylinder, and this blow back may continue to the air intake when the intake valve opens. With valve overlap this could also occur if the valve timing is improperly done. [4]

Another effect is pressure blow-back, also known as back-pressure, from the exhaust. It is caused by formation of acoustic pressure waves reflected at the exhaust pipe ends see 4.2.2.

(19)

2.2.3 The thermodynamic four-stroke cycle

In terms of thermodynamics two different four-stroke cycles exists. The Otto cycle, which is used for gasoline and E85, and the diesel cycle. The difference is that in the Otto cycle fuel is injected into the air before entering the cylinder, while in the diesel cycle it is injected just before maximum compression. This means that the combustion in the ideal Otto cycle is isochoric, i.e. the volume is constant, while for the ideal diesel cycle it is isobaric, i.e. constant pressure.

The ideal cycles are adiabatic, which means there is no increase in heat due to the com-pression. The cycles can be seen in figure 2.3. [5]

Figure 2.3: The ideal thermodynamic cycles. The upper graphs show the Otto cycle, and the two lower the diesel cycle. [5]

The efficiency of the ideal four-stroke engine is much higher than what is achieved in reality. The reason is that the ideal cycles do not account for entropy effects, which is realized by the diagrams to the right in figure 2.3. [5]

(20)

2.2.4 Cylinder geometry

The geometry of the cylinder is given by its displacement, bore and stroke. The bore is the diameter of the piston head and the stroke is the distance between TDC and BDC. The engine displacement is the swept volume of all the pistons, in the cylinders, in the engine. It is thus given by:

Displacement = π

4 · Bore

2_{· Stroke · N} _(2.1)

where N is the number of cylinders.

The clearance volume is the free space in the cylinder when the piston is at TDC. The size of this volume is normally given by a compression ratio, which is the ratio between the volume of the free space in the cylinder at BDC and the clearance volume. This value normally is around 10:1 for Otto engines, while it is usually 15:1 or higher for diesel engines.

2.2.5 Volumetric efficiency

Volumetric efficiency can be considered a measure of how efficiently the engine cylinders are filled with fresh air. It is defined as the ratio between the volume of air entering the cylinder and the volume of air that would fill the swept volume of the cylinder, at intake manifold pressure and temperature. The volumetric efficiency is expressed as a percentage.

This value is greatly affected by the air intake and exhaust, as well as the geometries in the cylinder and valve timing.

In a regular car a volumetric efficiency of about 80-90% is usually achieved, but it is possible to get as high as 110%. The reason why it is possible to get above 100% is the valve overlap period, which causes some leakage of fresh air into the exhaust. In order to get above 100% all systems must be thoroughly designed to work together.[6]

(21)

2.3 Auxiliary systems

2.3.1 Air intake

The purpose of the air intake is to deliver sufficient amounts of air needed for combustion and the valve overlap. This requires the system to be designed in such a way that as much air as possible get to the cylinder.

As both the air intake and exhaust systems resonate when acoustic waves are generated by the piston movement, the total duct length needs to be carefully chosen.

The different parts of the system are inlet, air filter and manifold, shown schematically in figure 2.4. The inlet is designed to lead as much air as possible into the duct. The air is cleansed in the filter, to get rid of elements which should not take part in the combustion. The manifold is basically a box which alters the pressure to a desired level. From the manifold there are ducts to the air intake valves on the cylinders.

Figure 2.4: Schematic of the air intake system.

The throttle is drawn to the side, because it is a control valve positioned on the air intake, usually between the air filter and the manifold. It is used to operate the engine in different situations and sometimes to help start the engine. If it is not present the engine speed is not controlled.

Formula student cars are also fitted with what is called a restrictor. This is a regulation defined by a smallest duct diameter, through which all air used for combustion must pass [1]. This way all cars will have a maximum power output, because the airflow is choked.

An example of a formula student car air intake is shown in figure 2.5.

Figure 2.5: The air intake of ER15. The non-labeled parts in the background are fuel injectors. (Source: ER15 CAD-Model, with permission.)

(22)

2.3.2 Exhaust

The exhaust has one primary purpose; to transport hot gases to a place where they do not inflict any direct harm. Secondary objectives of the system includes noise reduction and gas cleaning.

The typical exhaust system for a formula student car, schematically shown in figure 2.6, consists of various tubes and a silencer. The first part of the tube mapping is called the manifold, unless the tubes are made from sheet metal then they are referred to as headers. This is where all tubes from each cylinder are connected to the main tube of the exhaust. An example can be seen in figure 2.7.

Figure 2.6: Schematic of the exhaust system.

Figure 2.7: The exhaust system of ER15. (Source: ER15 CAD-Model, with permission.)

The path of the tubes must be mapped in such a way that they fit in the car, while not being in a place where high temperatures are harmful. For regular cars they also should be easy to replace, as the high temperatures and elements in the gases wears on the tubes, while for a formula student car this is not critical. The dimensions of the tubes affect noise and engine performance.

(23)

The engine may have more than one cylinder. This requires several exhaust tubes to be merged into one, which makes the design process more complex than with just one tube. The main issue is to make sure back-pressure is kept to a minimum.

2.4 Internal flows

Flow through a duct is called internal flow, note that the duct does not have to be fully enclosed. The three most important properties to know about, to understand flow behavior, are massflow, flow speed and pressure. In this section the flow mechanical effects from geometrical changes to a duct is discussed.

Flow speed is often referred to as the Mach number, M, which is the ratio between the flow speed and the speed of sound. This is used because it is often more practical to use dimensionless values. Note that the speed of sound is dependent on temperature and pressure. [7]

Pressure consists of two parts. Static and dynamic, or kinematic, where the dynamic pressure is caused by the motion of the fluid. It is given by: [8]

q = 1

2ρ · u

2 _(2.2)

which gives stagnation pressure as:

ps = pa+ q = pa+ 1 2ρ · u

2 _(2.3)

Massflow is the amount of mass of the fluid, which passes through a cross-section of the duct. It is considered constant because the flow is statistically steady, i.e. the flow behavior is periodic and can be regarded as constant by time averaging.

Bernoulli’s theorem states that the rate of change of the flow speed is inversely propor-tional to the rate of change in pressure. [9]

If the duct diameter increases the flow speed will decrease, because of the larger duct volume. Consequently the pressure rises. The opposite will happen if duct diameter is decreased, as shown in figure 2.8.

(24)

If there is an obstacle in the duct, such as a screen or filter, part of the flow is completely blocked. This can be regarded as a decrease in duct diameter for the length of the ob-stacle, which results in a pressure drop across the whole cross-section, though locally the pressure may increase.

Viscous effects has significant impact on the flow.

At the wall of a bounding surface the flow speed is zero. This is known as the no-slip condition, and as a consequence the flow close to the wall will have a parabolic velocity profile, see figure 2.9. [7]

Figure 2.9: The basic look of the velocity profile of the flow, close to the wall.

This is referred to as a boundary layer, and its thickness is defined by the flow speed as a function of the wall distance, u = u(y).

The normal definition used is that the top of the boundary layer is found where the flow speed is 99% of the free stream flow speed, u(y) = 0.99 · u∞[8].

Boundary layers can be laminar, transitional or turbulent. A boundary layer becomes turbulent due to an adverse pressure gradient, meaning the static pressure increases in the direction of the flow. As the static pressure increases the boundary layer thickness grows, and if the pressure increases enough the flow will separate.

For a laminar boundary layer the thickness is usually in the order of 1 mm while a turbulent boundary layer can be in the order of 10 cm. [7] [10]

Because the duct diameters of air intakes and exhausts are in the region of 5-10 cm and the flow is turbulent, the viscous effects cannot be ignored.

In figure 2.10 the effects of viscous flow becomes apparent.

(25)

Though figure 2.10 is an example of external flow, the flow close to the wall in internal flows will have similar effects due to surface roughness and manufacturing errors. [7]

A quantity describing flow behavior is the Reynold’s number, Re. It is the ratio between inertial forces and viscous forces, and is used to help predict similar flow patterns in different flow situations.

Flows with high Re are more likely to become turbulent, and thus relatively small dis-turbances will have a greater effect on the flow. The number is calculated by

Re = ρuL

µ =

uL

ν (2.4)

where ρ is the density of the fluid, u is flow speed, L is a typical length, µ is dynamic viscosity and ν is the kinematic viscosity. For air at atmospheric pressure and 288 K

the density is 1.225 kg/m3, the dynamic viscosity is about 2·10−5 Pa·s and thus Re is

60,000·u·L. As comparison honey has a density of 1440 kg/m3 _{and a dynamic viscosity}

of 10 Pa·s which gives a Re of 144·u·L.

If a bend is introduced there will be a radial pressure gradient. Because the massflow is constant the flow speed along the outer wall is higher than along the inner wall. Thus the pressure on the inside of the bend will be higher.

As the pressure along the outer wall begins to drop the flow moving along the inner wall will start to move towards the outer wall. Due to the centrifugal forces the flow will swirl.

At some point the pressure will be at a minimum on the outer wall, which means that there is an adverse pressure gradient after that. This causes the boundary layer to become turbulent and, assuming the bend curvature is sufficiently high, separate causing an eddy to form in the corner.

Just after the pressure peaks along the inner wall, it will begin to drop, while at the same time the pressure is increased on the other side of the duct, causing an adverse pressure gradient along the inner wall. Because of this an eddy forms just behind the bend, assuming high enough bend curvature. The two eddies can be regarded as flow obstacles and because of this the bend causes pressure drop. Even without the eddies the bend will cause pressure drop, because frictional losses are greater than for an equivalent straight duct.[11]

(26)

Figure 2.11: Illustration of flow through a bend. (Inspired by Idelchik, 1986).

2.5 Flow measurement techniques

Common properties measured are pressure, flow speed, mass flow, fluid density and/or temperature. Much information is often given from visually examining the flow, e.g. by using smoke or fluorescent substances. The more properties that are measured, the more complete the image of flow behavior is, and at the same time it is easy to check calculations and/or calibration of instruments. However, it depends on the situation which properties are measured.

This thesis focuses on measuring pressure and performing smoke tests.

2.5.1 Manometers

A manometer is a U-shaped pipe filled with liquid. When applying force in one end of the tube the liquid will rise in the other, as shown in figure 2.12. From the difference in height of the columns dynamic pressure is given by equation 2.5.

(27)

2.5.2 Pitot-tubes

Pitot-tubes, seen in figure 2.13, are bent tubes and use the effects visible in a manometer to measure stagnation pressure.

When one end is placed in a flow the stagnation pressure can be measured in the other end of the tube.

Figure 2.13: The principle of a pitot-tube.

The problem with a pitot-tube is that either the ambient or dynamic pressure needs to measured by some other method, if the flow speed is to be calculated. Otherwise anemometry measurements should be used to get a more complete comprehension of the flow.

2.5.3 Prandtl-tubes

Prandtl-tubes, seen in figure 2.14 are basically pitot-tubes with an extra tube and valve, which allows for direct measurement of the ambient- and stagnation pressures.

Figure 2.14: The principle of Prandtl-tubes.

Before placing the tube in the flow the stagnation pressure will be equal to the ambient pressure. This pressure is kept in the shorter tube, due to the valve. The dynamic pressure is then given by the difference between the stagnation pressure and ambient pressure, when the nozzle is placed in the flow.

(28)

2.5.4 Visual testing

Visual investigation of a flow is a simple way of detecting events of interest in the flow. The most common way of observing airflow is to introduce smoke. If the medium is a liquid, coloring agents are used instead, e.g. fluorescent gel.

Visual testing can often help explain unexpected measurements and provide a good understanding of the flow behavior, which also shows how well a simulation compares to the real flow.

Figure 2.15: Photograph of an airflow visualized using smoke. [12]

In figure 2.15 the flow direction is downwards. This particular test was used to measure angles of the flow. That type of information could not be obtained by measuring the pressure or flow speed. The smoke was generated by a smoke flare. [12]

(29)

2.6 Acoustics

Acoustics is an interdisciplinary science. Originally it was only considering sound, but now deals with all types of mechanical wave propagation, including the waves caused by piston motion in a reciprocating engine. This thesis will focus on the acoustics in ducts.

Acoustic effects in ducts arise from the flow, viscosity and vibrations in the duct itself. Acoustic waves are pressure waves. This means they affect the pressure of the flow in the duct.

The acoustic pressure, for a certain frequency, in a duct can be considered constant across the cross-section if the width of the duct is less than half of the acoustic wave-length. This means the only modes of this frequency are modes which are only dependent on the duct length. The range of frequencies with this property is known as the plane wave range. Frequencies above the plane wave range behave in a more complex manner, with cross-sectional pressure gradients.[13]

The wave equation for flow in a duct is [13]

∇2p − 1 c2( ∂ ∂t+ U ∂ ∂x) 2 p = 0 (2.6)

where p is the pressure, c is the speed of sound and U is the particle velocity. By applying a separation technique

p = ˆpΨ(y, z)eikxx_eiωt _(2.7)

where k is the wave number, defined as angular velocity per unit distance, ω the fre-quency and Ψ(y, z) describes the amplitude. The form of the wave equation becomes

( ∂ 2 ∂y2 + ∂2 ∂z2)Ψ + (k 2_{+ 2M kk} x+ M2kx2− kx2)Ψ = 0 (2.8)

which if Ψ is constant and M=0 yields kx = ±k. By setting the expression inside the

right parenthesis equal to the absorption factor, α, squared this gives

( ∂ 2 ∂y2 + ∂2 ∂z2)Ψ + α 2_{Ψ = 0} _(2.9)

(30)

For a rectangular duct with cross-section b x h they are k+_x,nm= 1 1 − M2(kM − p k2_{− (1 − M}2_)α2 nm) (2.10a) k−_x,nm= 1 1 − M2(kM + p k2_{− (1 − M}2_)α2 nm) (2.10b)

From these expressions it can be realized that there exist a frequency below which the longitudinal wave number becomes complex, and the mode is damped exponentially. This frequency is called the mode’s cut-on frequency, and shows that at sufficiently low frequencies only plane wave propagation is possible.[13]

Another important factor for the acoustics in a duct is the acoustic boundary layer. It is similar to flow mechanical boundary layers.

It has been shown that the viscous effects affect the acoustics both inside and outside of the boundary layer.

The effects become stronger at low Re and high frequencies.

The velocity gradient of a wave inside the boundary layer is higher at high Re and low frequency.[14]

2.7 Computer simulation

Softwares for simulating fluid flows, thermodynamics and acoustics are available. How-ever, they all include various kinds of simplified models to reduce computing time. This section describes how these softwares work on a basic level.

2.7.1 Computational Fluid Dynamics

Computational Fluid Dynamics, CFD, is mainly used to simulate fluid flow, but can also be used to simulate thermodynamics. It is done by defining a computational domain, which is divided into smaller finite elements. The elements are connected to each other. Equations are solved numerically within the elements. [15]

The most precise method is direct numerical simulation, DNS. The equations to be solved are, for inviscid flow, the Euler equations or, for viscous flow, the Navier-Stokes equations. However, DNS requires a lot of computing time and instead the equations solved are the Linearized Euler Equations, LEE, and the Reynolds Averaged Navier-Stokes equations, RANS, respectively.[15]

The computational domain is a geometrical model, and the more details it contains the better the result will be. A high level of detail will require more computing time. [8] If the geometry is symmetrical it is split through its symmetry plane. For a straight tube, with constant cross-section, there are two symmetry planes and it is thus only necessary

(31)

Division of the computational domain is known as meshing. A mesh can be crude or fine, and the mesh elements can have varying quality. The more mesh elements the more details of the flow will be captured, but costs time and computational power.[8]

One significant objective when meshing is to capture the boundary layer accurately, in order to take viscous effects into account. This can be done by adding an inflation layer at walls. These mesh elements tend to be small, and with high quality. However, the quality of mesh elements outside of the inflation layer may be reduced.[15]

It is difficult to find a working compromise between inflation layer and the rest of the mesh, without increasing the number of mesh elements to such an extent that it costs unreasonable amounts of computing time.

In order for the solver to know how to perform the calculations it requires boundary conditions. [15] Types of boundary conditions include Inlet, Outlet, Opening, No slip wall, Free slip wall and Symmetry. At the inlet initial conditions are set. Inlet and outlet are constrained to only allow flow in one direction, while openings allows flow in both directions through it.

Simulations are not possible without first defining the inlet, outlet and walls.

2.7.2 Acoustic modeling

There are various types of acoustic models.

1D-simulation softwares use DNS of plane wave propagation. These models are not very accurate, but can capture the general acoustic behaviors in e.g. a ventilation duct. In order to get more accurate simulations it is necessary to use 3D-models. The two relevant for this thesis are Computational AeroAcoustics, CAA, and thermoacoustics. They are done in a similar way to CFD.

Computational AeroAcoustics

CAA is used to model the acoustic behavior in a flow field, for example in the turbulent region behind a cylinder. This type of model requires some CFD to be carried out, but in order to reduce the computing time needed the RANS equations are often replaced by LEE. The flow is thus considered an ideal gas with no viscous effects.

The mesh is, compared to CFD, crude with a minimum mesh element size of 2-3 cm. CAA is often used to simulate the wave propagation in ventilation ducts. [16]

Thermoacoustic models

Thermoacoustic models are used to simulate acoustic effects, influenced by thermody-namic events. The thermodythermody-namics affect the density and viscosity of the fluid and thus these models require a finer mesh in order to capture the acoustic boundary layer. This type of model like CAA often solves LEE to simulate fluid flow.

(32)

Chapter 3 Method

3.1 Information gathering

Most of the information was found in the Society of Automotive Engineers database. Other places used was DIVA, Google and Google scholar. Some information was found in text books [13] [15].

The first search term used was ”air intake and exhaust design internal combustion en-gines”. This generated many hits on how to tune an engine when using Google. For the SAE database the search term was too broad, as many hits were regarding jet engines, two-stroke engines and six-stroke engines. The other search terms are results of what was found in manuals and studies.

All search terms used are listed below.

• ”internal combustion engine” • ”four stroke engine”

• ”four stroke internal combustion engine” • ”air intake”

• ”air intake design” • ”four stroke air intake”

• ”four stroke air intake design” • ”exhaust”

• ”exhaust design” • ”four stroke exhaust”

• ”four stroke exhaust design” • ”air intake exhaust design”

(33)

• ”four stroke air intake fluid flow” • ”four stroke exhaust fluid flow”

• ”four stroke internal combustion engine cfd” • ”four stroke internal combustion engine modeling” • ”four stroke air intake cfd”

• ”four stroke air intake acoustics”

• ”four stroke air intake acoustic modeling” • ”four stroke exhaust cfd”

• ”four stroke exhaust acoustics”

• ”four stroke exhaust acoustic modeling” • ”acoustic 3D modeling” • ”computational aeroacoustics” • ”computational aeroacoustics cfd” • ”thermoacoustics” • ”thermoacoustics cfd” • ”thermoacoustics cfd hybrid”

• ”thermodynamic modeling four stroke engine intake” • ”thermodynamic modeling four stroke engine exhaust” • ”axiomatic design”

• ”cfd sine wave boundary condition engine”

Search results which could be relevant, judging from only the titles, were saved. The saved results were then screened by reading the abstracts, and only those which were both relevant to the subject and used a relevant method were kept. For example several studies, containing only CFD-simulations, with only one theoretical validation case were discarded. All experimental studies were kept, as well as all relevant information on acoustic modeling.

In the case of contradicting studies, the methods were examined more closely. Often studies could be discarded because they used simulation software combined with only one, or at least few, theoretical validation cases.

(34)

3.2 Experimental setup

During the literature study none of the studies found provided information on how changes in the exhaust affects the air intake were found. In theory there are acous-tic effects, but this needs verification. Experiments were designed for this purpose, and the experimental setup is described in this section.

Results from the experiments can easily be analyzed using Matlab or similar software.

3.2.1 Flow generation

When testing engines they are placed in a test bench, where they are running. The flow is generated by the valve and piston motion.

To generate the flow without starting the engine, the engine is cranked. To generate enough flow for realistic results the engine speed should be around what it is when the engine is running.

Two methods were tried for this: a hammer drill with spur gear connection directly on the crankshaft, and a chainsaw, with dismounted sword and a custom made adapter plate, connected to the drive shaft of the rear wheel. The chainsaw would not have been tried if the attempt to use a hammer drill had been successful.

The hammer drill rotates at about half of what the engine speed should be, which was dealt with by the spur gears. The gear properties can be seen in table 3.1 and the gears can be seen in figure 3.1.

Pinion Gear Pitch diameter (mm) 30 64.29 No. of teeth 14 30 Width (mm) 8 8 Pressure angle (◦) 20 20 Module 2.14 2.14

Table 3.1: Geometrical properties of the gears.

Figure 3.1: Photographs of the spur gears. The pinion, shown in the two upper photos, was placed on the crank shaft, while the gear was used in the hammer drill.

(35)

More on the gear design can be found in Appendix B.

A wooden rig was built for the hammer drill. It was screwed on to the crank house. The setup is shown in figures 3.2 and 3.3.

Figure 3.2: Photograph of the pinion mounting.

Figure 3.3: Photograph of the hammer drill mounting.

The chainsaw normally runs at about twice the speed of the scooter engine. Because of the extra resistance, from the scooter engine, the engine speed would have been somewhat reduced. In order to lower the engine speed further the chainsaw was driving the drive shaft for the rear wheel, and using the drive belt to drive the crank shaft. The gear ratio would then be automated by the centrifugal clutch. The setup is shown in figures 3.4 - 3.6.

A failure mode effect analysis was done for this method. It is an analysis of the risks involved, and can be found in Appendix C.

(36)

Figure 3.5: Left: The chainsaw centrifugal clutch. Middle and right: The adapter plate mounted on the chainsaw.

Figure 3.6: The centrifugal clutch on the scooter. Note the three holes, in which the screws on the adapter plate were inserted.

(37)

3.2.2 Equipment

The equipment used was a scooter with engine, plastic tubes, plastic hoses, a Prandtl-tube with receiver, a hammer drill with spur gear connection, a chainsaw with the sword replaced by an adapter plate, a smoke machine and a camera.

Engine

The engine used was a single cylinder engine from a scooter. It was kept in the scooter, in order to use the scooter as test rig. The engine properties are shown in table 3.2.

Engine type: Scooter engine

Displacement: 50 cc

Bore x stroke: 38x44 mm

Max power output: 2.98 kW at 7500 rpm

Max torque: 4.5 Nm at 5000 rpm

Table 3.2: Properties of the engine.

The original air intake, throttle and exhaust were dismounted.

Access to the drive shafts is gained by removing the kick starter and crank housing cover.

Air intake and exhaust

Various plastic tubes was used to create the geometries. In order to connect the systems to the engine a vinyl hose was used, to allow for easy access where the tested geometries were attached. This can be seen in figure 3.7.

Figure 3.7: Photographs of the PVC-tubes. On the left is the tube to the intake port, and on the right is the tube from the exhaust port.

(38)

The tubes which made up the intake and exhaust geometries were manufactured by vacuum forming plastic PET-G sheets. CAD-files for wooden molds were created, which were used to mill the molds in a CNC-machine. A more detailed description of the manufacturing process is found in Appendix A.

Intake duct

The inlet was a funnel, with a minimum diameter of approximately one inch. The tube then continues to a straight tube, which is the test section, attached to the PVC-tube. this can be seen in figures 3.8 (a) and 3.8 (b).

(a) The air intake. (b) The test section.

Figure 3.8: Photographs of the air intake and test section.

The extra tube used as test section was necessary, because the blockage from the Prandtl-tube would have been too large in the inlet duct.

The reason for having the hose was practical. It did not generate an optimal solution for running the engine, but for the experiments it was more important to get consistency. Consistency is easier to achieve if assembly of the parts is easy.

Exhaust ducts

The exhaust duct was made from the PVC-tube, which was intended for use as a refer-ence in order to see how much differrefer-ence the straight tubes make, and several PET-G geometries.

Reference geometry

The reference geometry was a straight tube. In order to speed up the manufacturing process only the longest tube needed was produced. It could then be cut to the shorter reference lengths needed. The tube can be seen in figure 3.9.

(39)

Figure 3.9: Photograph of the reference tube.

Step geometries

Step geometries are ducts with abrupt changes in diameter. An increase in duct diameter is called step up, while a decrease is called step down.

These geometries was the reason for vacuum forming plastic. This type of geometry, for the purpose of measuring flow, is difficult to manufacture from metal. Relatively small parts would be welded together, and the welds form irregularities on the inside of the duct. These irregularities cause disturbances to the flow, and the effects cannot be foreseen. Vacuum forming was an attempt to avoid this problem.

Three different step sizes were planned, but only two could be manufactured. The steps which would have been used was one inch and one half inch in size. These can be seen in figure 3.10.

Figure 3.10: Photograph of the step geometries.

Map testing geometries

The experiments required for straight tubes and two bends. The geometries can be seen in figure 3.11.

(40)

Figure 3.11: Photographs of the map testing geometries.

Measuring instruments

For the pressure measurements a Prandtl-tube was to be used. It connects to an elec-tronic receiver, which displays the dynamic pressure and stores maximum and minimum values. It can be seen in figure 3.12.

Figure 3.12: Photograph of the Prandtl-tube connected to the electronic receiver.

The receiver for the Prandtl-tube requires calibration to the ambient pressure when started. This is easy to forget when doing many measurements, and thus it was in-tended to use a regular barometer, seen in figure 3.13, to check the ambient pressure. During the data analysis, if the values seemed strange, it would have been easy to sub-tract the ambient pressure.

(41)

Figure 3.13: Barometer.

The black arm shows the ambient pressure, while the red arm is rotated manually. Often the red arm is set at 1 atm. The units displayed are mbar and mmHg.

3.2.3 Testing

There are two main things which were meant to be tested. Effects from step geometry and exhaust mapping, on air intake flow. As a secondary effect these tests may also have shown the influence of introducing bends in the exhaust.

Step geometry testing

To get a good idea of what influence a step geometry has, two different step sizes were planned to be tested. Each would then be compared with a regular straight tube with the same length, to check if it is the step or the length difference that has eventual effect.

Map testing

This test was set up to be done using two assemblies of one bend and two straight tubes. Both setups had the same length, 417 mm, and would be compared to a straight tube of the same length. The purpose of the test was to check if the positioning of the tubes has any influence.

Smoke testing

The smoke tests need to be filmed, in order to be analyzed. To achieve this, a small film shooting area would have been set up.

It was planned to use a small smoke machine, normally used to create ”party fog”. Its

capacity was 200 m3/min, which is too much. Quickly turning it on and off, while using

(42)

Procedures

In order to get an idea of how large the user error is, several persons measure the same thing a few times each. From the averages the error can be estimated. The error occurs because the Prandtl-tube must be held perpendicular to the flow direction, and each person will have a different way of holding the instrument.

All ducts are relatively short and a fully developed flow cannot be expected. This means the values shown on the display fluctuates, but stays within some interval. What would have been searched for is changes to that interval.

(43)

Chapter 4 Results

4.1 Experiments

No measurement values were obtained from the experiments, as flow could not be gen-erated.

The hammer drill was not powerful enough to overcome the compression pressure at sufficiently high engine speeds.

The chainsaw could not be started due to the extra resistance from the scooter.

There are four possible outcomes of the experiments:

1. No changes to the exhaust geometry affect the pressure in the air intake. 2. All changes to the exhaust geometry affect the pressure in the air intake. 3. Only the step geometries affect the pressure in the air intake.

4. Only the map changes affect the pressure in the air intake.

4.2 Data gathering

4.2.1 Engine performance

The amount of air entering the cylinder will have a large effect on the energy released from the combustion. Thus the torque is affected, and in general the engine speed for peak volumetric efficiency coincides with engine speed for peak torque. The curves for volumetric efficiency over engine speed and torque over engine speed are similar to each other. With a known volumetric efficiency it is possible to calculate how large amount of fuel can be ignited, for best performance, as well as if the amount of fuel injected is known it is possible to calculate the required volumetric efficiency.[6]

(44)

4.2.2 Auxiliary systems

Many manuals and online tools for the purpose of designing air intake and exhaust systems were found. They use empirical formulas to calculate duct dimensions.

However, these formulas should be used very carefully because they could not be verified in experimental studies, though it was confirmed that for mid and upper ranges of engine speed, the tuned exhaust length was independent of the exact engine speed and only dependent of tube diameter. [17]

Air intake

Intake runner length will influence the volumetric efficiency [18]. The reason is that with shorter ducts the turbulence intensity is higher, which will give a better mixing of the air and fuel in the cylinder, but lowers the volume of air entering the cylinder[19]. This will in turn affect the temperature of the exhaust gases, as the more air that enters the cylinder, the more oxygen is provided for combustion, which increases the heat.

The length of the duct is designed with respect to the speed of the engine. Lower speeds require longer ducts, in order to get more torque [20].

Exhaust

The back-pressure forms because the exhaust pipe works like a church organ, only with hotter and denser gas than air. Acoustic pressure waves inside the tubes are reflected at tube ends. These acoustic waves are generated by the motion of the piston [21]. The reflected waves can cause an increase in pressure closer to the engine. If this pressure gets too high the exhaust gases will not exit the cylinder, which reduces the engine’s performance and lifespan. An increase in back-pressure of about 3.4 kPa causes a 1% decrease in power, for engine speeds between 1000-4000 rpm [17].

The geometries of the air intake and exhaust system must not have the same eigenfre-quencies, because they should not resonate at the same engine speed. The reason is both to keep noise level down and to keep engine performance up. The exhaust piping is designed to resonate at a specific engine speed and, from the pressure waves reflected at the pipe ends, change the pressure at the exhaust valve to drop below ambient pressure during the overlap period. The air intake will form another resonating system, which must not cancel the effect on the pressure from the exhaust.[22]

The frequency of the exhaust is determined by its length, which means the higher the design engine speed is the shorter the exhaust should be, until tuned exhaust length becomes independent of exact engine speed [22].

(45)

4.2.3 CFD

Pure CFD-analysis have been successfully used for optimization of the air intake manifold for a diesel engine.[23]

It has also been used for in-cylinder flow simulation, for the purpose of optimizing the cylinder geometry.[24]

CFD-simulations of combustion is possible and numerical methods exist, but it requires knowledge about the chemical reactions involved and it is often necessary to perform several simulations to capture different stages in the combustion process. [15]

Using pure CFD-simulation for all systems, i.e. air intake, engine and exhaust, simulta-neously has been investigated. The model was fitted to measurement data, in order to simulate wave effects accurately. The model did not work well to simulate flow behavior of the compression stroke. The main difficulty lies in simulation of the vortex breakdown during this stroke.[25]

4.2.4 Acoustics and acoustic modeling

For acoustics in ducts it is important to take viscous effects into account.[14]

When considering a four stroke engine the main frequency is given by the engine speed and the scavenging of the cylinders. For example a four cylinder engine at 3000 rpm will empty two cylinders in every revolution, which gives the main frequency as 2·3000₆₀ = 100 Hz. However, there will be higher frequencies generated by other events in the four stroke cycle. [13]

Decoupled hybrids of CFD and CAA have been done. It was validated by more accu-rately simulating the acoustics in cavities and in the flow field behind a cylinder, than pure CAA simulations. [26]

It is possible to use CAA in confined spaces and in combination with CFD. An attempt to simulate a real application including a ducted diaphragm at low Mach number was tried. However, the acoustic model was oversimplified and the suggested approach is in need of a more thorough validation. [27]

(46)

4.2.5 Thermodynamic modeling

The numerical methods used to simulate the combustion process of Otto-engines need refining, but have potential of being very accurate. [29]

1D thermodynamic modeling has been developed and correlates well with experimental values. The model uses fractal geometry to deal with the wrinkling process of the flame, which is used to calculate the increase in surface area. [30] [31]

(47)

Chapter 5 Discussion

5.1 Limitations

Most of the other studies found were focused on the air intake and individual parts of the exhaust system. One was found on the overall exhaust system design. That is why this study focused on how changes in exhaust geometry affects the air intake.

The catalytic converter is disregarded because it is not used on formula student cars, nor was it going to be used in the experiments of this study.

One possible way to manufacture step geometries was thought to be by vacuum-forming plastic. Hence, it was decided that plastic tubes should be used.

This also enabled smoke testing. However, plastic tubes cannot take the heat from a running engine, and therefore the engine had to be cranked without being started.

For this thesis a Prandtl-tube was available, whereas an anemometer was not. Thus, flow speed was not measured.

5.2 Experimental results

The experiments could not be conducted because flow could not be generated.

In the first try, using a hammer drill and spur gears, the hammer drill was simply not powerful enough to overcome the compression pressure. The drill had a varying rota-tional speed of between 50-500 rpm, depending on the resistance encountered from the cylinder. This speed is too low to generate any noticeable flow, and to keep trying would burn the electronics in the hammer drill.

The second try, using a chainsaw with a custom made adapter plate, would probably work. However, it would require a different cranking device for the chainsaw.

(48)

The starter rope is attached to a plastic roll, with two small metal pins coming out of it. The metal pins lock on to the flywheel and by pulling the rope the engine is cranked. Since the chainsaw’s crank shaft was directly connected to the scooter’s rear drive shaft, the resistance when cranking was increased a lot.

When pulling the starter rope the rotation encountered heavy resistance abruptly. Pulling harder on the rope would likely have caused the plastic to crack, as the rope was less likely to yield.

While designing a cranking device for the purpose of starting the chainsaw might not be difficult, it would take too long time for the scope of this thesis.

Clearly the compression pressure should not be underestimated, even for a small engine. In order to get these experiments to work, a method for measuring flow properties, a camera, an engine, air intake and exhaust geometries, and a method for cranking the engine are needed. The cranking method must to be able to overcome the compression pressure, at the intended engine speed and without damaging the engine.

If the tests were to show that none of the exhaust geometries influence the air intake, additional testing with metallic tubes is needed. There is a possibility that the reflected acoustic waves become insignificantly small in comparison to the source, but it is more likely that the plastic tubes behave different from metallic tubes.

Should all geometries have influence, this means flow mechanic and acoustic effects are important, as the only possible difference in the map test is that if the bend is placed further downstream the flow has had more time to stabilize. However, if flow mechanics or acoustics is dominating can only be determined by analyzing the results more thor-oughly, and probably by performing more tests.

The most likely result is that only the step geometries cause changes in the air intake. It would mean there is no influence from flow mechanics upstream. Only acoustic waves need to be considered in this way. Considering the fact that flow mechanically the sys-tems are only connected during the overlap period, and a downstream flow is expected, and at the same time acoustic waves are traveling in both directions is what makes this the most probable case.

The situation where only the map tests have influence should not happen unless some-thing has gone wrong during the tests. For example if the map tests are done first, and when changing to step geometries a leak is created. For this case it is more likely that all geometries have influence.

This type of tests may further increase understanding of how the systems work.

5.3 Theory

From 2.4 and 4.2.2 it is clearly not possible to change the geometry of the air intake without affecting the exhaust.

The effects of intake geometry can to some extent be taken care of by the intake man-ifold design, but there is limited space on the car for the manman-ifold and all ducts to the cylinders must fit. This limits how much difference there can be between the incoming

(49)

therefore the total length of the air intake ducts will influence the length of the exhaust ducts.

This does not necessarily mean that the air intake cannot be designed separately, as long as the design choices are constantly updated and communicated to the exhaust designer, though it means the systems are not uncoupled.

The exhaust design does influence the upstream flow within itself. However, it is inter-esting to investigate if the effects are noticed even further upstream, and if it matters in which order different tubes are assembled. If it can be shown that the order of the tubes is insignificant, the positioning of the parts in the exhaust can be done in a decoupled way, even if the systems are coupled assuming their dimensions are set.

Adding tubes to the exhaust will affect the back-pressure. Exactly how depends on each specific case, and it has not been determined whether or not it will influence the air intake design.

From the existing literature it is apparent that the first step of the design process should be to set a engine speed range to design for.

The range should be quite high, since a formula student car is small and thus the tube length will be short. This also means the exact engine speed is not important for tuned exhaust length, see 4.2.2.

5.4 Computer simulation

For Elith racing the best current method is to design their systems in the conventional way. That is to use 1D-acoustic models and tweak them, using CFD and bench tests. Starting with the air intake, continuing with the exhaust and iterating. The reason why only a 1D-model is insufficient is due to the fact that it does not take frequencies above the cut-on frequency into account.

The method is time consuming, but there currently does not exist a complete model to take everything into account.

The reasons for designing the air intake first is because it is easier to alter the design of the exhaust than trying to redesign the air intake. Especially since ELiTH Racing manufactures the air intakes from carbon fiber. Changing the design after it has been manufactured is therefore problematic, making it better to tune the exhaust.

A reasonable complete calculation model would have to take fluid flow, acoustics and thermodynamics into account, while requiring as little computer power as possible.

The first idea was that it may be possible to do a pure CFD-simulation with some har-monic wave perturbations as boundary conditions. The method is theoretically possible, but it is extremely difficult to get accurate results if the exact wave effects are unknown. Since those values cannot be measured during the design phase, the model needs to be expanded to include a large amount of the surrounding air in the computational domain.

(50)

However, the mesh becomes very large and computationally heavy, and there is likely no way of simplifying this kind of model without losing accuracy.

Pure CFD-models can be used for optimization purposes, because the modeling process starts with fitting the model to an existing situation, and then altering the design.

Combining CFD with CAA for the air intake and CFD with thermoacoustics for the exhaust and engine are possibilities.

A different approach for the exhaust and engine may be to use CAA and CFD with a thermodynamic model, e.g. in Ansys it is possible to create both a CFD-model and a thermodynamic model and make them communicate with each other. Using this and a CAA-model can reduce the needed computer power, since 1D simulation of the thermo-dynamics may suffice.

The whole system should be divided into two or three subsystems: air intake, engine and exhaust. The air intake model provides information about pressure, massflow, air-to-fuel ratio and, maybe, level of mixing to the engine. It could also be better to keep the level of mixing within the engine model.

The engine model provides information about the acoustic sources to the air intake, while information about pressure, thermodynamic effects, massflow and acoustic effects are given to the exhaust model.

The exhaust model returns information on back-pressure to the engine.

The air intake and exhaust models also need to provide each other with information on their overall acoustic resonance. This may be possible to achieve with an external CAA-model, though it may be necessary to investigate other acoustic models for this purpose. This information may also be useful when designing fasteners for the systems. This suggestion is shown in figure 5.1.

Figure 5.1: Schematic of suggestion for computational model.

A problem with this approach is that a detailed model of the internal geometries of the engine is required, to produce accurate results. Even a small change in e.g. corner angles can have a large impact on the acoustic behavior. Initially the approach is likely

(51)

Chapter 6 Conclusions

• The systems are not uncoupled. Changes in the air intake will influence the exhaust design in some way.

• Whether the systems are decoupled or coupled is not determined, as influence of exhaust geometry changes on the air intake could not be tested.

• A design process for Elith racing is suggested

– First set a range of engine speeds to optimize the design for. All other engine speeds should then be mapped in such a way that the engine is running, but not at its best.

– It is currently best to use a 1D-Wave simulation, to get a crude design concept.

– The design concept should be refined using CFD and bench testing.

– The air intake should be designed first, and then the exhaust. The design process needs to be iterated.

• A suggestion of how a complete computational model may be constructed is pre-sented.

– The model must account for flow mechanical, acoustic and thermodynamic effects, while requiring as little computing time as possible.

– The computational domain of the model should be divided into air intake, engine and exhaust.

– The air intake is likely possible to model accurately using a decoupled hybrid of CFD and CAA.

– The engine and exhaust are likely possible to model using a decoupled hybrid of CFD and thermoacoustics, or using a hybrid of CFD, CAA and thermody-namic models.

– A model for the overall resonance of the systems is also needed. A CAA-model might work.

Decoupled Design of Auxiliary Systems for Internal Combustion Engines