CFD simulations of mixing processes

in Direct Injection SI engines

Master’s thesis in Automotive Engineering - MPAUT

Nithin Revadal
Subhash Ravi Kumar

DEPARTMENT OF MECHANICS AND MARITIME SCIENCES

C HALMERS U NIVERSITY OF T ECHNOLOGY
Gothenburg, Sweden 2021
www.chalmers.se
Master’s thesis in Automotive engineering 2021

CFD simulations of mixing processes in

Direct Injection SI engines

Nithin Revadal
Subhash Ravi Kumar

Department of Mechanics and Maritime Sciences

Division of Combustion and Propulsion Systems
Chalmers University of Technology
Gothenburg, Sweden 2021
CFD simulations of mixing processes in Direct Injection SI engines

© Nithin Revadal, Subhash Ravi Kumar, 2021.

Supervisor and Examiner: Petter Dahlander

Professor at division of Combustion and Propulsion Systems
Mechanics and Maritime Sciences, Chalmers University of Technology

Master’s Thesis 2021

Department of Mechanics and Maritime Sciences
Division of Combustion and Propulsion systems
Chalmers University of Technology
SE-412 96 Gothenburg
Telephone +46 31 772 1000

Cover: Equivalence ratio of iso-octane at -180CAD, obtained from Tecplot for CON-
VERGE.

Typeset in LATEX, template by Magnus Gustaver

Printed by Chalmers Reproservice
Gothenburg, Sweden 2021

iv
Abstract
To keep up with stringent emission standards and Internal Combustion Engines
(ICE) having major market share of all the motor vehicles in the world, it is necessary
to explore the avenues to decrease emissions and improve efficiency of the ICE
engines.
In this thesis the effects of varying piston and fuel temperatures on mixing process
and fuel film mass on the piston surface were studied using CONVERGE CFD soft-
ware. The simulations were performed by selecting a single component (iso-octane)
and a multi-component fuel blend (iso-octane, n-hexane, n-decane) as surrogates for
gasoline, in a Volvo single cylinder medium tumble 4 valve direct injection engine.
Simulations were performed with piston temperatures of 380K,430K, 450K and with
fuel temperatures 335K and 363K.
The piston temperature played an important role in the quality of mixing, as
the higher piston temperature led to faster vaporisation, better mixing. Whereas
lower piston temperature caused leaner mixture as some the fuel failed to vaporise.
A higher fuel temperature resulted in better mixing with single component fuel,
whereas with multi-component fuel and its wide boiling temperature range showed
poorer mixing.

Keywords: Internal combustion engines, CFD, simulation, mixing, fuel film mass.

v
Acknowledgements
First and foremost we would like to offer our sincere gratitude to Petter Dahlander,
who guided us in every step and provided his valuable time for this thesis. Without
which, finishing this thesis would have been impossible.

We would also like to thank the PhD and Postdoc researchers, Sreelekha Etikyala
and Rafig Babayev in the Combustion and Propulsion Systems division, for their
diligent efforts in answering our many questions.

Convergent Science provided CONVERGE licenses and technical support for this
work.

We are eternally grateful for our family for their continuous support and, never
ending love and belief in us.

Gothenburg, June 2021

Nithin Revadal and Subhash Ravi Kumar

vii
 Contents

List of Figures xi

List of Tables xiii

1 Introduction 1
1.1 Aim and Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Background work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Fixing the geometric defects . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Intersecting Triangles . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Non-Manifold Edges . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.3 Open Edges . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.4 Normal Orientation . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.5 Isolated Triangles . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Method 7
2.1 Case setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Spray modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.3 Drop evaporation . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.4 Drop wall interaction . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4.1 Wall film model . . . . . . . . . . . . . . . . . . . . . 12
2.1.4.2 Drop/Film rebound . . . . . . . . . . . . . . . . . . 12
2.1.4.3 Drop/film splashing . . . . . . . . . . . . . . . . . . 12
2.1.5 Spray breakup . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.5.1 Kelvin-Helmholtz model . . . . . . . . . . . . . . . . 14
2.1.5.2 Rayleigh-Taylor model . . . . . . . . . . . . . . . . . 14
2.1.6 Regions and Initialization . . . . . . . . . . . . . . . . . . . . 15
2.1.7 Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.8 Grid control . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.8.1 Fixed Embedding . . . . . . . . . . . . . . . . . . . . 17
2.1.8.2 Adaptive Mesh Refinement (AMR) . . . . . . . . . . 18
2.2 Operating parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Multi component fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3 Results and discussion 23

3.1 Reverse flow into intake manifold . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Reverse flow with single component fuel . . . . . . . . . . . . 24

ix
Contents

3.1.2 Reverse flow with multi-component fuel . . . . . . . . . . . . . 25

3.2 Single component fuel . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Baseline simulation . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1.1 Pressure . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1.2 Fuel film mass on piston . . . . . . . . . . . . . . . . 29
3.2.1.3 Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1.4 Temperature . . . . . . . . . . . . . . . . . . . . . . 31
3.2.2 Fuel film mass on piston . . . . . . . . . . . . . . . . . . . . . 32
3.2.2.1 Effects of varying surface piston temperature . . . . 32
3.2.2.2 Effects of increasing fuel temperature . . . . . . . . . 33
3.2.3 Mixing process . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.3.1 Piston temperature = 380K . . . . . . . . . . . . . . 34
3.2.3.2 Piston temperature = 450K . . . . . . . . . . . . . . 35
3.2.3.3 Fuel temperature = 363K . . . . . . . . . . . . . . . 36
3.3 Multi component fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1 Baseline simulation . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1.1 Pressure . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1.2 Fuel film mass on piston . . . . . . . . . . . . . . . . 38
3.3.1.3 Mixing Process . . . . . . . . . . . . . . . . . . . . . 39
3.3.1.4 Temperature . . . . . . . . . . . . . . . . . . . . . . 40
3.3.2 Fuel film mass on piston . . . . . . . . . . . . . . . . . . . . . 41
3.3.2.1 Effects of varying surface piston temperature . . . . 41
3.3.2.2 Effects of increasing fuel temperature . . . . . . . . . 42
3.3.3 Mixing process . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.3.1 Piston temperature = 380K . . . . . . . . . . . . . . 43
3.3.3.2 Piston temperature = 450K . . . . . . . . . . . . . . 44
3.3.3.3 Fuel temperature = 363K . . . . . . . . . . . . . . . 45

4 Conclusion 47

5 Future work 49

Bibliography 51

x
 List of Figures

1.1 Intersecting triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 Open edge error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Geometric defects upon running diagnosis in Converge [3] . . . . . . . 4

2.1 Front view of the Cylinder . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Top view of the Cylinder . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Boundary conditions [3] . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Possible parcel wall interaction outcomes in Kuhnke model [2]. . . . . 13
2.5 Schematic of the KH-RT spray breakup model [2] . . . . . . . . . . . 15
2.6 A visualisation of fixed embedding around the valve [2] . . . . . . . . 18
2.7 Vaporisation profile [5] . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.1 Reverse flow of iso-octane . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2 Fuel back flow with iso-octane . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Fuel back flow with iso-octane [Tw 380 vs 430K] . . . . . . . . . . . . 25
3.4 Fuel back flow of iso-octane in MCF . . . . . . . . . . . . . . . . . . . 25
3.5 Fuel back flow of n-hexane in MCF . . . . . . . . . . . . . . . . . . . 26
3.6 Fuel back flow of n-decane in MCF . . . . . . . . . . . . . . . . . . . 26
3.7 Pressure inside the cylinder . . . . . . . . . . . . . . . . . . . . . . . 28
3.8 Fuel film mass on piston . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.9 Equivalence ratio distribution with Piston temperature at 430K . . . 30
3.10 Temperature inside the cylinder for SCF . . . . . . . . . . . . . . . . 31
3.11 Fuel film mass on piston with varying piston temperatures . . . . . . 32
3.12 Fuel film mass on piston with Fuel temperature at 363K . . . . . . . 33
3.13 Equivalence ratio distribution with Piston temperature at 380K . . . 34
3.14 Equivalence ratio distribution with Piston temperature at 450K . . . 35
3.15 Equivalence ratio distribution with Fuel temperature at 363K . . . . 36
3.16 Pressure inside the cylinder for MCF . . . . . . . . . . . . . . . . . . 37
3.17 Fuel film mass on piston for MCF with Piston temperature at 430K . 38
3.18 Equivalence ratio distribution with Piston temperature at 430K . . . 39
3.19 Temperature inside the cylinder for MCF . . . . . . . . . . . . . . . . 40
3.20 Fuel film mass on piston with varying piston temperatures for MCF . 41
3.21 Fuel film mass on piston for MCF with Fuel temperature at 363K . . 42
3.22 Equivalence ratio distribution for MCF with Piston temperature at
380K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.23 Equivalence ratio distribution for MCF with Piston temperature at
450K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

xi
List of Figures

3.24 Equivalence ratio distribution for MCF with Fuel temperature at 363K 45

xii
 List of Tables

2.1 Engine specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Spray modeling parameters . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Regions and initialisation parameter setup . . . . . . . . . . . . . . . 16
2.4 Boundary temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Simulation parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Fuel properties of Gasoline and multi-component fuel . . . . . . . . . 20

xiii
List of Tables

xiv
 1
Introduction

With an eye on the global warming and increasing stringent emission standards, it is
imperative to increase the efficiency, as well as reduce emissions of Internal Combus-
tion Engines (ICE), who account for majority of all the motor vehicles in the world
today. Computational Fluid Dynamics (CFD) simulations are an effective tool to
study the behaviour of the fluids interacting with various surfaces or geometries. In
the field of automotive engineering, CFD simulations are extensively employed to
study the behaviour and characteristics of ICE under different operating conditions.

With its autonomous meshing, various and efficient chemistry solvers, and the abil-
ity to accommodate complex moving geometry, CONVERGE CFD software was
the best option to run our simulations on. It has robust example models and an
extensive user manual to understand both the user interface and the inner workings
i.e. physical and chemical models employed in it.

In this thesis we implemented the geometry of a cylinder, belonging to an engine

used in an experimental setup at the department of Combustion and Propulsion
Systems, in Chalmers University of Technology, into the CONVERGE. This gave
us an exciting opportunity to run our simulations at practical operating conditions,
and to comprehend, explain the results better.

1.1 Aim and Objective

The principle aim of this thesis is to study the mixing processes in a Gasoline Direct
Injection (GDI) spark ignition (SI) engine, along with the liquid fuel film on the
piston surface. The thesis was carried out with an extensive use and exploration
of the CONVERGE CFD software, which aided in implementing the geometry and
interpret the results effectively.

A good mixing process is one of the significant gateways for clean combustion and
better fuel efficiency of the engine. To keep up with the emission standards it is
important to study the mixing process of air and vaporised fuel in an ICE, as it
contributes to the better extrapolation of parameters affecting the combustion and
thus the emissions.

1
1. Introduction

The objectives of the thesis are listed below,

• Implement the geometry of the cylinder from experimental setup into the sim-
ulation software.

• Simulate by introducing a single component and multi-component fuel, as sur-

rogates for the gasoline into the setup.

• Running simulations on different scenarios, by altering the initial operating

conditions of the case setup through,
– Varying the temperature of the piston.
– Increasing the fuel temperature.

1.2 Background work

As the aim of the thesis is to study the fuel film mass on piston surface and mixing
process at the end of compression stroke under different parameters, an extensive
literature survey was conducted to understand the causes and results better.

CONVERGE CFD being relatively new to the department of Combustion and

Propulsion Systems, posed a few challenges when implementing a new geometry
into the software. To implement the new Computer-aided design (CAD) geometry,
a STL file generated from a CAD software was initially introduced into the Converge
environment using a python script.
Once the CAD geometry was successfully imported into CONVERGE CFD, the soft-
ware automatically meshes the geometry assessing the different boundaries present.
But, it contains numerous geometric defects which needs to be rectified. The most
commonly encountered geometric defects are:

1. Intersecting triangles.
2. Non-manifold edges.
3. Open edges.
4. Normal orientation inconsistencies.
5. Isolated triangles.

1. Intersecting triangles: Intersecting triangles essentially means that two are

more triangles are traversing through each other in the present geometry. It
must be noted that, even if the error does not exist at one time frame in the
geometry, it could appear when the motion is assigned to the geometry that is
being simulated which could lead to fatal error of the simulation. Figure [1.1]
shows a two dimensional example.

2
 1. Introduction

Figure 1.1: Intersecting triangles

2. Non-manifold edges: Any edge of an element must connect two triangles

to it. Non-manifold error occurs when an edge is shared by more than two
triangles to it.

3. Open edges: Open edge defects occur when at least one of the triangles of
the geometry does not have an adjoining triangle on each of its edges.

Figure 1.2: Open edge error

4. Normal orientation inconsistencies: Normal Orientation inconsistency

occurs when all the connected triangles’ normal vectors of a boundary are not
pointing towards one of the two directions (axially inwards or outwards). This
helps define which surface interacts with the fluid and the surface that does
not come in contact with the fluid.
It is preferred to have the normal vector of all the triangles pointing in the
direction where the fluid flows

5. Isolated triangles: Isolated Triangle defect occurs when a triangle does not
share any of its three edges with the adjoining triangle of the same boundary,
essentially encompassed by triangles of a different boundary on all sides.

3
1. Introduction

Upon running diagnostics for the first time following surface defects were
encountered,

Figure 1.3: Geometric defects upon running diagnosis in Converge [3]

1.3 Fixing the geometric defects

Rectifying the above defects becomes simpler when clearly defined Boundary lines
and Boundary fences are present.

• Boundary Line : The line consisting of edges which connects two adjacent
boundaries.
• Boundary Fence : The line consisting of edges marking clear distinction
between start/end of a complex geometric shape in a boundary.

1.3.1 Intersecting Triangles

• Delete and Re-create: One or an array of triangles intersecting are deleted
and new triangles are created in it’s place. If a clear boundary line/fence is
defined, these triangles are assigned to their respective boundaries automati-
cally.
• Re-Align and Stitch : If the boundary lines/fence of the array of triangles
intersecting each other are clearly defined. Re-aligning them to the required
axis and stitching the boundary lines together is the quick and easy way to fix
the issue.

4
 1. Introduction

1.3.2 Non-Manifold Edges

Non-Manifold edge surface defects are comparatively a rare occurrence. But when
it appears, it is usually a singled-out element.

• Delete and Re-create : The element whose edge is connected to more than
two triangle is to be deleted and two are more triangles need to be created in
its place, so that further non-manifold errors are not shown.

1.3.3 Open Edges

• Patch: This solution is used to close an open surface when there is a clear
boundary fence present, around the missing triangles.

• Create: If the boundary fences are not clearly defined, they are defined man-
ually. Then triangles are created at required positions and assigned to corre-
sponding boundary.

1.3.4 Normal Orientation

• Attempt to fix all normal: In CONVERGE studio, Transform > Normal
> Attempt to fix all normals is the best fix, where the software automatically
recognizes and fixes all the normal orientation defects of the geometry.

• Manually change normal: If the above method does not work, the normals
of individual triangle or the triangles of a boundary are changed under Trans-
form > Normal > Propagate change for single triangle/Reverse all normals in
a boundary.

1.3.5 Isolated Triangles

• Select individual triangle and assign : If there are very few isolated
triangles, they are individually selected and assigned to the corresponding
boundary.

• Select boundary and assign : If there are numerous isolated triangles inside
a boundary. All the triangles of that boundary are selected and assigned to
the corresponding boundary instead of selecting the triangles individually.

5
1. Introduction

6
 2
Method

2.1 Case setup

The geometry selected for this simulation was a Volvo single cylinder medium tumble
4 valve direct injection engine. The major reason for choosing this engine was its
relevance in today’s market and the availability of data i.e. an experimental setup
with the same engine is used in the department of Combustion and propulsion
Systems at Chalmers University of Technology. Hence obtaining realistic data for
the simulations, which usually is a difficult and time consuming task, was effortless
in this case and thus helped in the better time management of the project.

Figure 2.1: Front view of the Cylinder

7
2. Method

Figure 2.2: Top view of the Cylinder

The specifications of the selected single cylinder engine,

Cylinder volume 475cc

Bore 82mm
Stroke 90mm
Engine head 4 valve SGDI
Spark plug Single electrode
Injector type 6 hole solenoid
Injection pressure ≈ 200 bar
Intake air temperature 310K
Coolant temperature 15-90◦ C
Con rod length 139.5mm
RPM 2000
Inlet valve opening (IVO) 356 CAD aTDC
Intake valve closing (IVC) 578 CAD aTDC
Exhaust valve opening (EVO) 145 CAD aTDC
Exhaust valve closing (EVC) 357 CAD aTDC

Table 2.1: Engine specifications

2.1.1 Boundary conditions

Since some parts of the cylinder are not stationary, while some do not come in con-
tact with fuel and have different physical properties (temperature, etc) due to being
in contact with cooling liquid etc, the cylinder geometry is divided into 20 individ-
ual boundaries in boundary option of the geometry menu in the software. This gave
the freedom to define the properties of each boundary individually, and further define
their
behaviour upon contact with fluids appropriately, which was accomplished using
boundary conditions menu under case setup.

8
 2. Method

Compared to others, some of the regions such as inlet valves, exhaust valves are
divided into a number of smaller boundaries. This is due to the translating motion
of some of its parts and to achieve fine meshing (achieved by using fixed embedding)
as the air passes through/around them.

Figure 2.3: Boundary conditions [3]

The boundary conditions can be used to predominantly define,

• Boundary temperature.
• Wall motion type.
• Surface movement.
• Turbulent kinetic energy.
• Profile, etc.

• Boundary temperature: The individual boundaries defined in the cylin-

der have temperatures ranging from 310-600 K. Although different boundaries
have different temperatures, but once defined it is constant throughout that
boundary i.e. in reality the temperature in that boundary region (for eg: Pis-
ton) may vary over the surface, the software assumes it to be a constant and
thus bringing some discrepancy to the results in the simulation compared to
that of experimental setup with similar operating conditions.

• Wall motion type: There are several types of motion like translating, rotat-
ing, arbitrary etc. Even though the simulation doesn’t start at 0 seconds or 0

9
2. Method

CAD, the position of the constant speed translating boundaries are adjusted
as though the simulation started at 0 CAD, and moved to that position of
starting CAD.

• Profile: Position of the translating boundaries in the cylinder geometry such

as piston, inlet, exhaust valves need to be defined with respect to the global
coordinates for each CAD. These profiles were obtained from the experimental
setup and added into the case setup in the simulation.

2.1.2 Spray modeling

The liquid fuel and its properties in the simulation are defined in spray modeling
menu, under physical models in the case setup.
Some of the important parameters defined in spray modeling are,

Parcel species iso-octane, n-hexane, n-decane

Mass fraction in 3 component fuel 0.45, 0.35, 0.2
Start of injection -310 CAD
Fuel temperature 335K
Injection duration 1.65ms or 19.8 CAD
Total injected mass 29.28 mg
Number of nozzles 6
Nozzle diameter 1.77 mm

Table 2.2: Spray modeling parameters

• Parcel species and mass fraction: A parcel is what the software solves in-
stead of a drop. A parcel usually undergoes various processes such as breakup,
coalescence, evaporation, etc. In a multi-component fuel mass fraction of each
parcel is defined to represent the liquid fuel injected.

• Injection duration: The value for injection duration (1.65ms) obtained from
experimental setup was converted to CAD as follows,

2π
ω(rad.s −1 ) = · N(rpm)
60
In this case RPM = 2000,

ω = 209.44(rad.s −1 ) or 12000(deg.s −1 )

Injection duration(CAD) = 12000 ∗ 0.00165

⇒ 19.8

10
 2. Method

• Total injected mass: The mass of fuel injected in each cycle was calculated
using the RPM and mass flow rate as follows,

M ass f low rate = 0.488(g.s−1 )

M ass f low rate ∗ 720 ∗ 60

M ass per cycle =
2000 ∗ 60 (kg.cycle−1 )

• Nozzle diameter: To achieve the same injection pressure as the experimental

setup (≈ 200 bar), the nozzle diameter was decreased from 2mm to 1.77mm

2.1.3 Drop evaporation

Drop evaporation, which is an immense part of the mixing process is governed by
Frossling correlation in the CONVERGE CFD modeling.
More than 50% of injected fuel comes in contact with the high temperature
(400-450K) surfaces of piston and liner. Although some of it is rebounded or splashed
upon contact, the amount remaining on those surfaces is still significant and plays
an important role in the mixture concentration.

The Frossling correlation is given by,

dr0 αspry ρg D
=− Bd Shd [2]
dt 2ρl r0
where,
• αspray is the vapor mass fraction at the drop surface

• D is mass diffusivity of liquid vapour in air.

Y1∗ − Y1
Bd = [2]
1 − Y1∗

• Y1∗ is vapour mass fraction at the drop surface

• Y1 is vapour mass fraction

• Sh d is Sherwood number given by,


1/2
 ln (1 + Bd )
Shd = 2.0 + 0.6Red Sc1/3 [2]
Bd

11
2. Method

2.1.4 Drop wall interaction

2.1.4.1 Wall film model
For drop wall interaction the CONVERGE software uses particle based wall film for
modeling. It usually uses a hybrid technique by assuming particle based quantities
for few calculations and film based quantities for others [2].

P
pVp
hα = [2]
|Aα,i |

where,

• ha is the film thickness on wall face a.

• Vp is the parcel volume of the parcel p.

• Aa,i is the area projection vector for face a.

2.1.4.2 Drop/Film rebound

When the weber number of the droplets is lower than wrebound specified in spray
modeling, the drops are assumed to rebound. The weber number defined in spray
modeling is given by, [2]

ρl Vn2 d
W ei = [2]
σ
where,

• ρ is the density of the liquid.

• Vn is the component of drop velocity normal to the surface.

• d is the drop diameter.

• σ is the surface tension.

2.1.4.3 Drop/film splashing

Drop/film splashing is defined by using Kuhnke model in simulations.

Splashing of a parcel in Kuhnke model is performed by equating the K number of

the parcel, to the splash critical K number[2].

12
 2. Method

Figure 2.4: Possible parcel wall interaction outcomes in Kuhnke model [2].

The K number of liquid parcel is given by,

(ρd)3/4 U 5/4
K= [2]
σ 1/2 µ1/4
Which can also be written as,

K = W e5/8 La1/8 [2]

where, W e is the weber number. La is the Laplace number given by,

σρL
La = [2]
µ2
• L is the diameter of the parcel.

• µ is the fluid viscocity.

• ρ is the density.

• σ is the surface tension.

2.1.5 Spray breakup

The CONVERGE software uses KH-RT (Kelvin-Helmholtz, Rayleigh-Taylor) model
to solve the spray breakup.

13
2. Method

2.1.5.1 Kelvin-Helmholtz model

The drop breakup size is determined by,

σ2
rc = [2]
2ρ2g U 3 vg

where,
• rc is breakup drop radius.

• r0 is liquid jet radius.

• ρg - density of inviscid gas.

• U liquid jet radius.

• ρl liquid density.

• µl viscosity of the liquid.

2.1.5.2 Rayleigh-Taylor model

Along with KH model the RT instability also causes spray breakup. The model(RT)
ignores both liquid and gas viscosities, the unstable RT waves are caused by quick
deceleration of drops from drag force and its given by the equation,

v !
ρl − ρg k3 σ
u
a − RT
u
ωRT = tk
RT [2]
ρl + ρg ρl + ρg

• ωRT - RT growth number.

• ρl - liquid density.

• ρg - Density of inviscid gas.

• KRT - wave number

In the converge CFD software both models can be ran concurrently,

14
 2. Method

Figure 2.5: Schematic of the KH-RT spray breakup model [2]

s
ρl
Lb = Cbl d0 [2]
ρg

As shown in figure [2.5] the model initially checks if RT model is responsible for
breaking up the droplets, if not the KH model is employed.
In general the KH model is responsible for droplet breakup within the characteristic
breakup distance Lb , beyond that distance both KH and RT models are activated
[2]

2.1.6 Regions and Initialization

Converge CFD defines collection of boundaries as a region. Each region can be given
a set of parameters apart from the parameter setup individual boundaries already
have.

The geometry of the engine used is divided into three following regions:
• Cylinder
• Intake System
• Exhaust System

Each of the above regions are initialised with a set of parameters, among which
Temperature, Pressure and the mass fraction of gaseous elements that are present
inside the region prior to starting the simulation were primarily focused upon.

The parameter setup with which simulations were run is shown in the Table [2.3],

15
2. Method

Region Cylinder Intake System Exhaust System

Temperature(K) 400 310 500
Pressure(Pa) 101325 101325 101325
Residual Gases N2 N2 N2
O2 O2 O2
CO2 CO2
H2O H2O

Table 2.3: Regions and initialisation parameter setup

2.1.7 Events
Events include instances such as controlling the motion of fluid between two bound-
aries. The opening and closing of the inlet valves and outlet values is to be included
in this section.
"CONVERGE has an option called ’Automatic Valve events’ where it creates OPEN
and CLOSE events based on specific valve lift profiles"[2] The valve lift profile which
determines the movement of inlet and outlet valves with respect to CAD are up-
loaded.

The geometry’s initial position prior to starting the simulation is important, which
includes the exact positions of both inlet and outlet values which needs to be in a
closed state in this case. But, if the valve’s upper surface comes in contact with
the cylinder head which physically happens in real life, it triggers the intersecting
triangles error. Hence in this case, there must be a minimum gap between the two
boundaries involved in the event, which needs to be specified.
When the minimum gap is reached, Converge automatically creates a set of tem-
porary triangles connecting the edges of both the boundaries involved, thus sealing
the area which blocks the flow of fluid. And during the event of opening the valves,
the triangles sealing the gap disappears once the valve moves beyond the specified
minimum value, allowing the flow of fluid.

2.1.8 Grid control

A base grid size specifies the size of the three dimensional blocks that are defined as
empirical elements of the geometry. The grid size has more to do with the simulation
run-time and the accuracy of the results. Smaller grid size provides a more accurate
result but increases the simulation run-time and vice versa.
There are certain options in Converge to reduce the simulation run-time by control-
ling the grid size by region or by the timing of events happening.
The following were used in the simulation:

16
 2. Method

2.1.8.1 Fixed Embedding

At the locations where it is critical to get accurate results, a finer resolution of grids
is applied to the geometry using fixed embedding. While using fixed embedding
an embed scale(positive integer) is defined to instruct the software regarding the
refinement of grid at the specified location [2].

basegrid
embed = [2]
2(embed scale)

A huge gain in saving the simulation run-time can be done by specifying the specific
time period for which the fixed embedding must exist, which essentially refines the
grids into finer elements only during a particular event at that specific region which
is otherwise at a base grid size. This helps is extracting the accurate results, cutting
down a huge part of simulation run-time. The region of a fixed embedding could
be assigned to any space in the domain, where the initial setup for each embedding
needs to be manually assigned.

The regions and specifics of the fixed embedding assigned under the geometry using
which simulations were run, is explained below:

• Cylinder Region
The cylinder region where the most of fluid interaction takes place is an impor-
tant region to capture the dynamics of the fluid, which influences the results
such as air-fuel ratio, spray, temperature and pressure variations, etc. So, an
embed scale of 2 is allotted throughout the whole simulation for the cylinder
region.

• Intake and Exhaust valve gaps

The gaps between the intake/exhaust valves and the cylinder head through
which the movement of gases (air and exhaust gases) occurs, is a crucial part
which defines the mixing process that takes place after the intake valves open
and escape of remnant gases after the combustion, since they affect the result
of the simulation significantly.
Hence, the grids corresponding to this region was given an embed scale of of
3 to refine the grids to a smaller size for the duration from the opening until
closing of the respective valves.This helps in capturing an accurate movement
of the gases along with reducing the simulation-run time by choosing to acti-
vate the embedding layer for a specific period of time, rather than to embed
the area for the whole simulation run-time.

17
2. Method

Figure 2.6: A visualisation of fixed embedding around the valve [2]

• Spray area
The fuel sprayed using an injector is designed to produce a spray cone. Dur-
ing the spray, thousands of fuel parcels (discrete packets of fuel) are ejected
through the nozzles of the injector. There are a lot of possibilities for each
parcel of fuel as to how and where they cease to exist depending on the environ-
ment it is interacting with. Most of the liquid parcels instantaneously vaporize
into gaseous form of the fuel, due to the existing high temperature and pres-
sure inside the cylinder, some end up on the piston and cylinder walls(liner) as
liquid droplets. Thus it is crucial to capture dynamics of the liquid fuel spray.
Hence, an embed layer in the shape of the cone adequate to capture the fuel
spray is introduced for the duration of injection.

2.1.8.2 Adaptive Mesh Refinement (AMR)

AMR is used to define grid at the regions with unsteady flows such as temperature
or velocity, automatically. This is helpful as it refines grid at the regions which are
of interest and at the same time, it does not slow down the simulation [2].

As mentioned above, a fixed embedding of scale 2 was applied covering the cylinder
region. But with the dynamics of fluid flow coming into the picture, it is critical
to capture the finer details of changes in temperature, pressure and velocity of the
flow of vaporized fuel-air mixture, which would require a more finer grid size than
procured by the embed scale ’2’.
Thus intake system and cylinder regions with an embed scale of ’3’ were selected,
since this work concentrates on the pre-combustion events, AMR automatically re-
fines the grids identifying quick changes in the required parameters such as temper-
ature, velocity of fluid, etc, for the specific period of time saving a huge amount of

18
 2. Method

simulation run-time and giving detailed output of the required parameter particu-
larly during the fuel spray and inlet of air into the cylinder.

2.2 Operating parameters

The initial case setup for the Volvo single cylinder medium tumble 4 valve GDI
engine, which will be used as a reference to compare further simulations in the
report was conducted with the following operating parameters,

Boundary Temperature [K]

Piston 430
Liner 400
Inflow 310
Outflow 600
Cylinder head 400
Intake valve 380
Exhaust valve 380

Table 2.4: Boundary temperatures

These parameters were obtained from the experimental data, as they showed better
results with good injection timing.
Except for that of piston, the boundary temperatures of all the individual bound-
aries remain unchanged i.e. in simulations the boundary temperature of the piston
is varied to investigate its effect on mixing process.

Injection temp 335 K

SOI -310 CAD
Simulation start -380 CAD
Simulation end -20 CAD

Table 2.5: Simulation parameters

The simulation was conducted from the start of intake stroke till the end of com-
pression stroke (just before ignition timing), as the focus of the thesis is to study
mixing processes and liquid film inside the cylinder.
The simulation time parameters i.e. start/end time will remain unchanged over en-
tire simulations with different operating points.

19
2. Method

2.3 Multi component fuel

Gasoline has boiling range of 307-470K, whereas iso-octane has a boiling temperature
of 372K, below which it would not evaporate unlike gasoline. And similarly when
the temperature inside the cylinder goes beyond 373K all the fuel (iso-octane) is
vaporised. Thus, to accommodate the vaporisation profile [2.7] of gasoline, 2 fuels
(n-hexane, n-decane) with lower and higher boiling points respectively, than that of
iso-octane were selected for multi-component fuel (MCF) blend.

Figure 2.7: Vaporisation profile [5]

Fuel Density [kg/m3] Saturation temp[K] LHV[MJ/kg]

Gasoline 740 304-470 43.40
iso-octane (IC8H18) 699 372 44.65
n-hexane (nC6H14) 661 342 45.09
n-decane (nC10H22) 722 447 44.59

Table 2.6: Fuel properties of Gasoline and multi-component fuel

This blend was selected based on their Lower heating value (LHV) and density
compatibility with the gasoline.

X
LHV(gasoline) = LHVi ∗ mi
Where,

• LHVi - LHV of individual surrogate fuel.

• mi - Mass fraction.

20
 2. Method

⇒ (0.45 ∗ 44.65) + (0.35 ∗ 45.09) + (0.2 ∗ 44.59)

LHV(blend) = 44.72

Which is close to that of gasoline (43.40 MJ/kg).

Since the focus of the thesis is mixing, density also plays an important role in selec-
tion and composition of the blend. As the density of the components in the blend
are close to that of gasoline [2.6], this blend was selected for simulations.

21
2. Method

22
 3
Results and discussion

For convenience the results section is divided into,

1. Reverse flow into intake manifold.

2. Single component fuel.

3. Multi component fuel.

In the latter two sections, results of following parameters are discussed,

• Baseline simulation: A simulation was conducted with the piston temper-

ature of 430K and fuel temperature 335K as a baseline for comparing further
simulations.

• Piston film mass: Amount of fuel reaching piston and rate of vaporisation.

• Mixing process: The equivalence ratio which is calculated using the formula,

F AR
φ=
F AR (Stoic)
which is a ratio of actual fuel to air ratio(FAR), to stoichiometric fuel to air
ratio.

• Pressure and temperature: The temperature and pressure profiles and

trend inside the cylinder, over the course of simulation.

3.1 Reverse flow into intake manifold

As the intake valves close after the start of compression stroke, a reverse flow of the
fuel-air mixture was encountered in all the cases.

RAM effect is a phenomenon which occurs when the piston is slowing down at BDC,
the inertia of air inside the intake manifold increases the pressure, and continues
charging until the inlet valves are closed when the engine speed is high. This effect
is taken into advantage by closing the intake valves 40-60°after BDC[4].

23
3. Results and discussion

The engine used in this thesis has a valve lift profile designed to take advantage
of RAM effect at high speeds. As the operating speed is 2000RPM(lower engine
speed) in this study, the reverse flow of fuel is more pronounced due to the low
speed, pressure difference, when the piston starts moving towards TDC.

3.1.1 Reverse flow with single component fuel

Figure 3.1: Reverse flow of iso-octane

Figure [3.1] shows equivalence ratio distribution inside the cylinder, it is observed
that the reverse flow of fuel starts from ≈ -180 CAD, when the piston is beginning
to move to TDC.

3.5e-05
Fuel temp - 363K
3e-05 Piston temp - 450K
2.5e-05 Piston temp - 380K
IC8H18 (kg)

Piston temp - 430K

2e-05
1.5e-05
1e-05
5e-06
0
-400 -350 -300 -250 -200 -150 -100 -50 0
Crank (DEG)

Figure 3.2: Fuel back flow with iso-octane

In the plot [3.2] it can be observed that the fuel starts escaping into intake manifold
around -180CAD, due to the pressure difference as the piston starts moving from
BDC to TDC. This trend of reverse flow into the intake manifold continues as long
as the inlet ports are open i.e until the CAD -126. After which the amount of fuel
in the cylinder stays constant.

For the case when the piston temperature is 380K the evaporation takes longer,thus
a slow trend in the increase of fuel mass inside the cylinder is observed [3.2].

24
 3. Results and discussion

Figure 3.3: Fuel back flow with iso-octane [Tw 380 vs 430K]

It was noted that the highest fuel back flow with single component fuel takes place
at the lowest piston temperature (380K)[3.2], this is due to the concentration of rich
fuel-air mixture in the vicinity of the inlet ports, compared to the case when the
piston temperature is 430K as shown in figure [3.3]

3.1.2 Reverse flow with multi-component fuel

Similar to the Single component fuel, simulations with Multi-component fuel also
exhibited reverse flow into the intake manifold when the piston started moving to
TDC from BDC.

1.4e-05
Fuel temp = 363K
1.2e-05 Piston temp = 380K
1e-05 Piston temp = 430K
IC8H18 (kg)

Piston temp = 450K

8e-06
6e-06
4e-06
2e-06
0
-400 -350 -300 -250 -200 -150 -100 -50 0
Crank (DEG)

Figure 3.4: Fuel back flow of iso-octane in MCF

As seen from the plot [3.4], which shows the iso-octane mass in the cylinder over
the simulation period, the reverse flow trend of fuel follows that of single component

25
3. Results and discussion

fuel as shown in [3.2].

1.2e-05
Fuel temp = 363K
1e-05 Piston temp = 380K
Piston temp = 430K
NC6H14 (kg)

8e-06
Piston temp = 450K
6e-06

4e-06

2e-06

0
-400 -350 -300 -250 -200 -150 -100 -50 0
Crank (DEG)

Figure 3.5: Fuel back flow of n-hexane in MCF

n-hexane with a lower boiling temperature of 342K evaporates quicker compared

to iso-octane, therefore the slow evaporation curve (when Tw=380K) is not en-
countered. Similar to iso-octane the reverse backflow starts when the piston starts
moving back to TDC from BDC (≈ -180 CAD), and the escape continues until the
intake valves are closed at -126CAD.

6e-06
Fuel temp = 363K
5e-06 Piston temp = 380K
Piston temp = 430K
NC10H22 (kg)

4e-06
Piston temp = 450K
3e-06

2e-06

1e-06

0
-400 -350 -300 -250 -200 -150 -100 -50 0
Crank (DEG)

Figure 3.6: Fuel back flow of n-decane in MCF

As seen in the plot [3.6] n-decane with higher boiling temperature evaporates slower
than the other 2 components of the blend. For all the cases except when the piston
temperature is 380K, the fuel evaporation tends to the follow trend similar to that
of single component fuel.

When the piston temperature is 380K which is much lower than the boiling tem-
perature of n-decane (447K), the curve is seen to gradually increase, in contrast to
other cases which see a rapid rise in n-decane mass inside the cylinder, and the re-
verse flow is also encountered later (between -150 & -126 CAD), than the other cases .

Unlike the other 2 components of the blend which have completely evaporated, due
to the increase in pressure and temperature, n-decane continues to evaporate after
the intake valve closes, and it continues until the end of simulation.

26
 3. Results and discussion

The maximum fuel escaped in the simulations was ≈5%, all of which will re-enter
as a pre-mixed, lean, homogeneous mixture when the inlet valves open in the next
cycle. Therefore this phenomenon is assumed to have minimal impact in this cycle,
and its effect is ignored as it is beyond the scope of study of the thesis.

27
3. Results and discussion

3.2 Single component fuel

3.2.1 Baseline simulation

An initial simulation with operating parameters described in [2.4][2.5] was conducted
as a baseline to compare the results from further simulations, which were performed
with changes made in the operating parameters.

The results of the baseline simulation are collated with that of further simulations,
to get a better understanding of the effects of changes made in operating parameters.

3.2.1.1 Pressure

Figure 3.7: Pressure inside the cylinder

The pressure profile inside the cylinder during the simulation is shown in [3.7], it is
identical for all the cases with different operating parameters (change in piston and
fuel temperature).
The Pressure inside the cylinder is seen to increase when the piston starts from
to TDC around -180CAD, at the end of compression stroke (-20 CAD) the peak
pressure is 1.25 Mpa.
The further simulations with single component fuel with varying operating param-
eters as mentioned above, showed the identical pressure profiles with no consider-
able deviation compared to the baseline simulation with pressure peaking around
1.25 MPa at -20 CAD.

28
 3. Results and discussion

3.2.1.2 Fuel film mass on piston

Figure 3.8: Fuel film mass on piston

The injection of fuel into the cylinder starts at -310 CAD, but the actual increase
in the mass of the fuel on the piston is seen after few CAD i.e. at -306 CAD [3.8].
The amount of fuel on the piston returns to zero at -263CAD i.e. after 43CAD from
the initial spike.
The peak fuel mass sitting on the piston is 5.5mg which is at the -285 CAD.

3.2.1.3 Mixing

To study the mixing process the equivalence ratio plot is obtained from Tecplot for
CONVERGE, by inserting a slice in the Y axis of the cylinder and equivalence ratio
distribution histogram is obtained from CONVERGE.

Figure [3.9] represents the equivalence ratio at -20 CAD (end of compression stroke).
The image above in the figure visually represents the distribution of fuel-air mixture
in the cylinder at a two-dimensional plane passing through the centre of the cylin-
der. The image below consisting the histogram shows the complete distribution of
fuel-air mixture inside the cylinder.

29
3. Results and discussion

Figure 3.9: Equivalence ratio distribution with Piston temperature at 430K

The image shows the existence of fuel rich mixture of equivalence ratio around 0.95
(Orange region) in the central part of the cylinder and the charge gets leaner grad-
ually moving away from the centre with equivalence ratio ranging from 0.89 (Yellow
region) to 0.80 (Green region) as we move towards the cylinder wall.

As represented by the histogram, a total of 55% of mixture is at an equivalence ratio

between 0.9 and 1.0, 37% between 0.8 and 0.9, 8% between 1.0 and 1.1 and the rest
of the mixture being leaner with equivalence ratio below 0.8. A high percentage of
mixture with an equivalence ratio in a single band shows us that the mixture present
in the cylinder at the end of compression stoke with piston temperature at 430K is
a fairly homogeneous mixture and indicates that a good mixing has taken place.

30
 3. Results and discussion

3.2.1.4 Temperature

Figure 3.10: Temperature inside the cylinder for SCF

The temperature profile for all the cases is identical. As seen in the plot [3.10], at
the start of intake stroke the temperature in the cylinder decreases for few CAD,
as the incoming air is substantially cooler (310K) than the temperature of cylin-
der boundaries(400-430K). Hence the initial defined temperature of 400K inside the
cylinder, goes down to 324K. The temperature inside the cylinder slightly increases
until the start of compression stroke, after which we see a quick increase in its value,
and the temperature at the end of compression stroke was 616K.

31
3. Results and discussion

3.2.2 Fuel film mass on piston

In this section the piston fuel film mass i.e. the amount of fuel hitting the piston
upon injection and its subsequent evaporation are discussed.

3.2.2.1 Effects of varying surface piston temperature

To study the effects of piston fuel film mass on mixing at different surface temper-
atures, the piston temperature was changed from the initial value of 430K to 450K
and 380K.

Figure 3.11: Fuel film mass on piston with varying piston temperatures

As shown in figure [3.11] amount of fuel, and the duration for film evaporation varies
drastically for different piston temperature.

• At the piston temperature 450K, which is slightly higher than that of baseline
temperature 430K and substantially higher than the boiling point of iso-octane
(372K). The peak piston film mass seen at -285CAD decreases from 5.5 mg to
4.5 mg.

• At the same piston temperature the evaporation duration decreased from 43

CAD to 37 CAD.

• When the piston temperature was decreased to 380K, which is much lower
than the earlier 2 cases. The peak piston mass increases from 5.5mg to 8.46
mg at -285CAD. This is due to slow evaporation rate i.e. the decrease in piston

32
 3. Results and discussion

temperature causes higher mass to accumulate before complete evaporation.

• At the same temperature [Tw=380K] the complete evaporation duration sub-

stantially increases from 43CAD to 130CAD i.e. starting from initial spike at
-306CAD, the fuel on the piston completely evaporates at -176CAD.

3.2.2.2 Effects of increasing fuel temperature

In the further cycles as the fuel passes through the tubes surrounded by components
which are increasing in temperature, the fuel temperature also increases. Taking this
into account, the fuel temperature was increased from the initial value of 335K to
363K.

Figure 3.12: Fuel film mass on piston with Fuel temperature at 363K

As the boiling temperature of fuel is 372K, increasing the fuel temperature to 363K
brought about quicker evaporation after coming in contact with the air inside the
cylinder [400K] and the peak piston film mass reduced from 5.5mg to 3.5mg. The
evaporation duration also decreased from 43 CAD to 35 CAD.

33
3. Results and discussion

3.2.3 Mixing process

Mixing process is an essential step in the ICE operation. A good mixing not only
gives better fuel efficiency but also reduces the particulate emissions. Thus achiev-
ing stoichiometric mixture is of high importance in this case.

3.2.3.1 Piston temperature = 380K

Figure [3.13] represents the equivalence ratio at the end of compression stroke when
the piston temperature is 380K.

Figure 3.13: Equivalence ratio distribution with Piston temperature at 380K

The image shows the existence of fuel rich mixture of equivalence ratio around 1.0
(Red region) in the right-side of the cylinder which continues towards the centre with
ER around 0.95 (Orange region) surrounded by ER around 0.87 (Yellow region). A
leaner mixture pockets of ER at 0.78 (Green region) and 0.5 (Blue region) exist
on the top right and left bottom of the cylinder, the clear distinction of different
coloured bands represent the mixture to be stratified.

As represented by the histogram, a total of 34% of mixture is at an equivalence

ratio between 0.7 and 0.8, 20% between 0.8 and 0.9, 16% between 0.9 and 1.0, 18%
between 1.0 and 1.1, 9% between 1.1 and 1.2 and majority of the rest of the fuel be-
ing leaner with ER between 0.5 to 0.7. Considerable mixture of air-fuel distributed
among the band varying from 0.6 and 1.2 confirms the presence of stratified mixture
inside the whole cylinder with a majority of air-fuel mixture being leaner when the

34
 3. Results and discussion

piston temperature is at 380K compared to 430K as shown in figure [3.9], represent-

ing a poor mixing of the charge.

3.2.3.2 Piston temperature = 450K

Figure [3.14] represents the equivalence ratio at the end of compression stroke when
the piston temperature is 450K.

Figure 3.14: Equivalence ratio distribution with Piston temperature at 450K

The image shows the existence of fuel rich mixture of equivalence ratio around 0.95
(Orange region) in the central part of the cylinder surrounded by a region of mixture
with ER around 0.89 (Yellow region). Leaner mixture pockets of ER 0.75 and 0.55
exist on the left side of the cylinder while the right side is comparatively has fuel
rich mixture.

Histogram shows a total of 45% of mixture is at an equivalence ratio between 0.9

and 1.0, 30% between 0.8 and 0.9, 19% between 1.0 and 1.1, and majority of the
rest of the fuel being leaner with ER between 0.5 to 0.8. A high percentage (94%) of
fuel sits in the range between 0.8 and 1.1 representing a fairly homogeneous mixture
with the presence of leaner charge pockets shows the mixing taken place to be better
with piston temperature at 450K than 380K but poorer than 430K.

35
3. Results and discussion

3.2.3.3 Fuel temperature = 363K

Figure [3.15] represents the equivalence ratio at the end of compression stroke when
the piston temperature is 450K.

Figure 3.15: Equivalence ratio distribution with Fuel temperature at 363K

The image shows the existence of fuel rich mixture of equivalence ratio around 0.95
(Orange region) in the central part of the cylinder surrounded by a region of mixture
with ER around 0.89 (Yellow region). Leaner mixture pockets of ER 0.75 and 0.55
exist on the left side of the cylinder while the right side is comparatively has fuel
rich mixture.

Histogram shows a total of 55% of mixture is at an equivalence ratio between 0.9 and
1.0, 26% between 0.8 and 0.9, 16% between 1.0 and 1.1, and rest of the fuel being
leaner with ER between 0.5 to 0.8. A high percentage (97%) of fuel ranging between
0.8 and 1.1 representing a fairly homogeneous mixture with the presence of leaner
charge pockets shows the mixing taken place similar to that of fuel temperature at
335K.

36
 3. Results and discussion

3.3 Multi component fuel

Similar to single component fuel a baseline simulation was performed for MCF as
well. This was used to compare and discuss the results of further simulations.
A total of four simulations were conducted among which one simulation with a
parameter setup up as mentioned in [2.4][2.5] to be a baseline simulation. Further
three simulations were conducted with varying individual operating parameters as
mentioned below,
• Piston Temperature = 380 K
• Piston Temperature = 450 K
• Fuel Temperature = 363 K

The results from these simulations were compared with the baseline simulation, as
well as to the Single Fuel Component simulation results of the corresponding pa-
rameter setup.

3.3.1 Baseline simulation

An initial simulation with operating parameter setup as described in [2.4][2.5] was
conducted as a baseline simulation whose results have been presented below.

3.3.1.1 Pressure

Figure 3.16: Pressure inside the cylinder for MCF

37
3. Results and discussion

Figure [3.16] shows in-cylinder pressure profile over the simulation period, it is iden-
tical for all the cases with different operating parameters. As seen in the graph, the
pressure starts increasing from -180 CAD from when the piston starts moving from
BDC towards the TDC, the pressure peaks at the end of the compression stroke
around 1.25 MPa at -20 CAD.
The further simulations using Multi component fuel with varying parameters as
mentioned above, showed the identical pressure profile with no considerable devia-
tion compared to the baseline simulation with pressure peaking around 1.25 MPa
at -20 CAD.

3.3.1.2 Fuel film mass on piston

Figure 3.17: Fuel film mass on piston for MCF with Piston temperature at 430K

Figure [3.17] represents the fuel film mass deposited on the piston. The fuel is
injected at -310 CAD, among which most of the fuel parcels get vaporized instan-
taneously due to the higher temperature that exists inside the cylinder. But, some
amount of liquid parcels which is directly shot towards the piston which has not
undergone vaporization settles on the piston. As it can be seen from the figure the
fuel film mass on the piston increases quickly from the start of injection (-310 CAD)
from 0 mg to 5.22 mg at -286 CAD which is just after the end of injection at -291
CAD. Further, a steep decline in the fuel film mass on piston is seen due to the heat
being transferred from the piston surface which is at 430K through conduction as
well as the surrounding high temperature that exists in the cylinder.

38
 3. Results and discussion

The fuel considered in this simulation include three components which are Iso-
Octane, n-Hexane and n-Decane. The decline in fuel film mass on piston can be
broken down into two stages with the differing slopes, one from CAD -286 to -269,
and the second from CAD -269 to -146. The first stage has a steep decline com-
pared to the second stage because it is observed that majority of the Iso-Octane and
n-Hexane settled on the piston gets evaporated during this stage since they have a
lower boiling point than that of n-Decane. The second stage sees a gradual decrease
in the mount of fuel film on the piston surface which majorly consists of the com-
ponent n-Decane that evaporates slower than Iso-Octane and n-hexane while the
temperature and pressure in the cylinder gradually increases.

3.3.1.3 Mixing Process

Figure [3.18] represents the equivalence ratio at -20 CAD (End of compression
stroke). The image above in the figure visually represents the distribution of fuel-air
mixture in the cylinder at a two-dimensional plane passing through the centre of the
cylinder. The image below consisting the histogram shows the complete distribution
of fuel-air mixture inside the cylinder.

Figure 3.18: Equivalence ratio distribution with Piston temperature at 430K

The image shows the existence of fuel rich mixture of equivalence ratio around 0.9
(Orange region) in the central part of the cylinder and the charge gets leaner grad-
ually moving away from the centre with equivalence ratio ranging from 0.85 (Yellow

39
3. Results and discussion

region) to 0.70 (Green region) as we move towards the cylinder wall.

As represented by the histogram, a total of 54% of mixture is at an equivalence

ratio between 0.8 and 0.9, 23% between 0.9 and 1.0, 17% between 0.7 and 0.8, 5%
between 1.0 and 1.1 and the rest of the mixture being leaner between 0.6 and 0.7. A
high percentage of mixture (about 95%) with equivalence ratio between 0.7 and 1.0
shows us that the mixture present in the cylinder at the end of compression stoke
with piston temperature at 430K is a fairly homogeneous mixture and indicates that
a good mixing has taken place.

3.3.1.4 Temperature

Figure 3.19: Temperature inside the cylinder for MCF

Figure [3.19] shows the Temperature profile inside the cylinder over the simulation
period, it is identical for all the cases. As seen in the graph, the initial temperature
inside the cylinder is 400K, then it gradually decreases due to the inlet valves open-
ing at -360CAD allowing the relatively cold air (310 K) inside the cylinder dropping
the in-cylinder temperature to around 325K. A steep increase in temperature is seen
from -180 CAD when the piston starts moving from BDC towards TDC and it peaks
at the end of compression stroke with temperature 617K.
Further simulations with varying parameters showed an identical temperature profile
with no considerable deviations compared to the baseline simulation temperature
profile.

40
 3. Results and discussion

3.3.2 Fuel film mass on piston

3.3.2.1 Effects of varying surface piston temperature
Similar to single component fuel, to study its effect on mixing processes and film
mass on piston, the initial piston temperature of 430K was changed to 380K and
450K for multi component fuel as well.

Figure 3.20: Fuel film mass on piston with varying piston temperatures for MCF

• For piston temperatures 380K and 450K the plot follows the same trend as
that of the baseline simulation (430K).
• The fuel film deposition on piston for all the three simulations reach the peak
around -286CAD. Showing a higher peak for 380K with 7.61mg and lower peak
for 450K with 4.48mg.
• For the piston temperature at 450K, the first stage of drop in fuel film starts at
-286CAD (peak) to -275CAD where majority of the Iso-Ocatane and n-Hexane
is vaporized. The second stage immediately begins where the remaining fuel
film containing n-Decane component starts evaporating. It takes a longer
duration than the first stage, and gets completely evaporated at -236CAD.
• For the piston temperature at 380K, the peak which is slightly later at -284
CAD compared to the other two simulations, the first stage of evaporation
lasts until -220 CAD and the second stage which begins at the same CAD
does not see the complete evaporation of the fuel film which gradually de-
creases in amount until the end of simulation (end of compression stroke) with
a fuel film mass of 1.07 mg still deposited on the piston surface. This residual
fuel film mass on the piston will result in particulates upon combustion.

41
3. Results and discussion

3.3.2.2 Effects of increasing fuel temperature

Figure [3.21] represents the fuel film mass deposited on the piston surface for fuel
temperature at 335K and 363K.
It can be observed that, the fuel film mass plot for fuel temperature 335K, follows
the exact same trend when the piston temperature is 430K. When the fuel temper-
ature is 363K we see a lower peak of 3.86mg, compared to 5.2 mg of 335K, at the
same CAD of -286. The first stage of evaporation lasting from the peak to -273
CAD and the second stage ending at -182 CAD.

Figure 3.21: Fuel film mass on piston for MCF with Fuel temperature at 363K

42
 3. Results and discussion

3.3.3 Mixing process

3.3.3.1 Piston temperature = 380K

Figure 3.22: Equivalence ratio distribution for MCF with Piston temperature at
380K

In the Figure [3.22] the two-dimensional slice of cylinder shows the mixture to be
leaner (mostly covered in green) and a blue pocket towards the left bottom, near
the connection of piston and cylinder wall. This is confirmed by histogram with a
majority 40% of mixture between E.R of 0.7 and 0.8, 22% between 0.8 and 0.9, 26%
between 0.9 and 1.0. and the rest are rich and lean pockets near the piston.
Since the deposition of the fuel film on the piston surface is highest when the piston
temperature is at 380K, and the fuel does not completely evaporate,it deprives the
charge inside the cylinder of a considerable amount of fuel giving rise to a leaner
mixture of charge as seen the figure.

43
3. Results and discussion

3.3.3.2 Piston temperature = 450K

Figure 3.23: Equivalence ratio distribution for MCF with Piston temperature at
450K

Equivalence ratio distribution for MCF at Tw=450K In the figure [3.23] the two-
dimensional slice of the cylinder shows the mixture to be stratified with multiple
regions of colour with a rich mixture at the centre and leaner mixture on either sides
and a very lean mixture (blue pocket) at the left bottom near the piston surface.
This is confirmed by histogram, which shows a majority 50% of mixture between
equivalence ratio of 0.8 and 0.9, 23% between 0.9 and 1.0, 13% between 0.7 and 0.8
and about 9% between 1.0 and 1.1 and the rest being very lean with ER around 0.6.

44
 3. Results and discussion

3.3.3.3 Fuel temperature = 363K

Figure 3.24: Equivalence ratio distribution for MCF with Fuel temperature at
363K

The image shows the existence of fuel rich mixture of equivalence ratio around 0.88
(Orange region) in the centre-right part of the cylinder surrounded by a region of
mixture with ER around 0.85 (Yellow region). Leaner mixture pocket of ER 0.75
exists on the left side of the cylinder while the right side is comparatively has fuel
rich mixture. Also, a fuel rich pocket of ER 1.0 near the piston on the left side,
indicating the better homogeneous mixing on the right side but distinct pockets of
different ER on the left side of the cylinder.

Histogram shows a total of 40% of mixture is at an equivalence ratio between 0.7

and 0.8, 26% between 0.9 and 1.0, 22% between 0.8 and 0.9, and majority of the
rest of fuel being leaner with ER between 0.5 to 0.7. Considerably higher and equal
percentage of air-fuel mixture spread across different ER bands and the presence
of distinct leaner and richer pockets identifies the stratified nature of the charge in
the cylinder which results in poor combustion compared to fuel temperature at 335K.

45
3. Results and discussion

46
 4
Conclusion

• Higher piston and fuel temperatures result in less fuel film mass on the surface
and faster evaporation.

• Lower fuel film mass on the piston and faster evaporation of the fuel results
in better mixing, as it provides more time for evaporated fuel and the air to
mix well.

• Multi-component fuel which has a realistic evaporation profile, provides better

and more realistic results.

• Fuel back flow into the intake manifold was observed when the piston started
moving from BDC to TDC.

47
4. Conclusion

48
 5
Future work

As the CFD software takes the defined surface temperature to be constant through-
out the surface, including a temporal and spatial temperature profile for the surfaces
is beneficial for obtaining better and realistic results.
Changing the Start of Injection (SOI) to study its effects on fuel film mass, mixing
and the fuel escape into the intake manifold. As the late injection might cause less
fuel film mass on piston surface and earlier injection time more film.
Performing ignition in CFD simulations, with the optimal conditions based on the
results of this thesis and study emissions.

49
5. Future work

50
 Bibliography

[1] Köpple, F., Seboldt, D., Jochmann, P., Hettinger, A. et al., "Experimental
Investigation of Fuel Impingement and Spray-Cooling on the Piston of a GDI
Engine via Instantaneous Surface Temperature Measurements," SAE Int. J.
Engines7(3):2014, doi:10.4271/2014-01-1447
[2] Richards, K. J., Senecal, P. K., and Pomraning, E., CONVERGE 3.0 Manual,
Convergent Science, Madison, WI (2020).
[3] Richards, K. J., Senecal, P. K., and Pomraning, E., CONVERGE 3.0, Conver-
gent Science, Madison, WI (2020).
[4] J.B. Heywood. Internal Combustion Engine Fundamentals. McGraw-Hill, 1988.
[5] Köpple, F., Jochmann, P., Kufferath, A. and Bargende, M., "Investigation of the
Parameters Influencing theSpray-Wall Interaction in a GDI Engine - Prerequi-
site for the Prediction of Particulate Emissions by Numerical Simulation,"SAE
Int. J. Engines 6(2):2013, doi:10.4271/2013-01-1089.

51
Bibliography

52
Department of Mechanics and Maritime Sciences
CHALMERS UNIVERSITY OF TECHNOLOGY
Gothenburg, Sweden
www.chalmers.se