Effect of Stainless Steel Additive Manufacturing On Heat Conductivity and Urea Deposition

IN

DEGREE PROJECT MATERIALS DESIGN AND ENGINEERING, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020 ,

Effect of Stainless Steel Additive Manufacturing On Heat

Conductivity and Urea Deposition

SABRINA EL MOUHIB

Abstract

Hydroforming is the manufacturing process that Scania uses to produce ex- haust pipes with complex shape and high durability. Selective Laser Melting is the process used by designers to print prototype pipes and perform emissions tests before mass production. Results from previous tests at Scania showed su- perior performance of 3D printed pipes compared to hydroformed components during emissions test as the 3D printed pipes were able to transfer heat faster than hydroformed pipes. To understand the reason behind this mismatch, the effect of selective laser melting parameters on energy density, relative density, grain size and thermal conductivity are investigated. These properties have direct impact on heat transfer.

Ten samples were fabricated using the same laser power and layer thickness but different combinations of scanning speed and hatch distance. Samples were then subject to microstructural analysis using an optical microscope and average grain size measurement using image analysis software called Imagej.

The density of each sample was measured using the Archimedes method.

Moderate correlation is found between energy density and relative density. No

ranking of the selective laser melting parameters with respect to forming the

highest density was achieved because of the high uncertainties involved with

the density measurement technique. Thermal conductivity was measured us-

ing the one dimensional heat flow equation with an appropriate experimental

set up. Thermal conductivity seems to be more influenced by relative density

and direction of printing layers than the energy density and grain size. This

conclusion is not statistically significant due to high uncertainty involved in

the measurement of thermal conductivity. More advanced and accurate tech-

nologies need to be used in the future to measure both density and thermal

conductivity in order to find the most suitable selective laser melting parame-

ters for Scania’s prototype pipes. The findings of this research can be used as

a foundation for future research related to urea deposition on 3D printed pipes.

Sammanfattning

Hydroforming är den tillverkningsprocess Scania använder för att producera avgasrör som har en komplex form och hög hållbarhet. Selektiv lasersmältning är den process som används av konstruktörer för att skriva ut prototyprör och utföra utsläppstester före massproduktion. Resultat från tidigare utsläppstes- ter på Scania visade en överlägsen prestanda för 3D-tryckta rör jämfört med hydroformade komponenter, eftersom 3D-tryckta rör kunde överföra värme snabbare än hydroformade rör. För att förstå orsaken bakom denna skillnad undersöks effekten av selektiva lasersmältningsparametrar som energitäthet, relativ densitet, kornstorlek och värmeledningsförmåga. Dessa egenskaper har direkt inverkan på värmeöverföringen.

10 prover tillverkades med samma laserkraft och skikttjocklek, men med oli- ka kombinationer av skanningshastighet och kläckavstånd. Proverna utsat- tes sedan för en mikrostrukturell analys med hjälp av ett optiskt mikroskop, samt genomsnittlig kornstorleksmätning med hjälp av bildanalysprogramva- ran Imagej. Densiteten för varje prov mättes med Archimedesmetoden. Mått- lig korrelation kunde identifieras mellan energitätheten och relativ densitet.

Ingen rangordning av de selektiva lasersmältningsparametrarna med avseen-

de på bildning av den högsta densiteten uppnåddes på grund av de höga osä-

kerhetsfaktorer som är involverade i densitetsmättekniken. Värmeledningsför-

mågan mättes med hjälp av den endimensionella värmeflödesekvationen, med

en lämplig experimentell uppställning. Värmeledningsförmågan tycks påver-

kas mer av tryckskiktens relativa densitet och riktning än energidensiteten och

kornstorleken. Denna slutsats är inte statistiskt signifikant på grund av hög osä-

kerhet i mätningen av värmeledningsförmåga. Mer avancerade och noggranna

teknologier måste användas i framtiden för att mäta både densitet och värme-

ledningsförmåga, för att hitta de mest lämpliga selektiva lasersmältningspara-

metrarna för Scanias prototyprör.

Preface

I would like to thank my supervisor Christopher Hulme-Smith, from KTH Royal Institute of Technology in Stockholm, for his help and support during this journey. His invaluable guidance, thoughtful feedback and continuous encouragement during the difficult times in which this project was conducted is only a proof of his exceptional supervision.

My biggest thanks to my parents and my two brothers for their uncon- ditional support in this intense academic year. Thank you for believing in me and giving me the strength to reach for the stars. I could not have done this without you. I love you more than words can describe.

Stockholm, October 2020

Sabrina El Mouhib

Contents

1 Introduction 1

1.1 Nitrogen Oxide . . . . 1

1.2 Selective Catalytic Reduction . . . . 2

1.2.1 Urea Deposition . . . . 3

1.3 Hydroformed and 3D Printed Exhaust Pipes . . . . 3

1.4 Metal Additive Manufacturing . . . . 4

1.4.1 Selective Laser Melting . . . . 5

1.4.2 Selective laser melting parameters . . . . 5

1.5 Problem Description . . . . 6

1.6 Goal of This Thesis . . . . 7

1.7 Scope . . . . 7

1.8 Limitations . . . . 7

1.9 Methodology . . . . 8

1.10 Ethical, Environmental and Economic Impact . . . . 8

2 Background 9 2.1 Influence of Selective Laser Melting Parameters . . . . 9

2.2 Heat Transfer . . . . 10

2.2.1 Thermal Conductivity Measurement Techniques . . . 10

2.2.2 Heat Equation . . . . 11

3 Methods 14 3.1 SLM samples . . . . 14

3.2 Microstructural Characterization and Grain Size Measure- ment . . . . 15

3.3 Relative and Energy Density Calculation . . . . 16

3.4 Thermal Conductivity Measurement . . . . 16

4 Results 19

4.1 Relative and Energy Density Results . . . . 19

4.2 Optical Micrographs of SLM samples and Grain Size Calcu- lation . . . . 19

4.3 Thermal Conductivity Results . . . . 24

5 Discussion 25 5.1 Relative and Energy Density . . . . 25

5.2 Microstructure and Grain Size . . . . 26

5.3 Thermal Conductivity . . . . 28

5.4 Propagation of Error . . . . 30

6 Conclusion 33 7 Future Work 34 8 References 35 A Appendix 39 A.1 Build Layout of SLM samples . . . . 39

A.2 CAD File Screenshot of Build Layout . . . . 40

A.3 Picture of SLM Samples Build Platform . . . . 41

Chapter 1 Introduction

The Diesel engine is widely used in the automotive industry. Its high absolute power, cost-effectiveness and high energy per unit mass make it a favourable choice for heavy-duty transport systems. However, its usage faces a big chal- lenge in terms of air pollutants. Diesel combustion is the source of several toxic emissions such as nitrogen oxide, carbon monoxide, hydrocarbon and particulate matter. Nitrogen oxide (NOx) are toxic gases that cause severe damages to the environment, mainly eutrophication and acidification. The long exposure to NOx harms the human respiratory system and lung function.

It is therefore necessary for the automotive industry to reduce the amount of NOx released to the environment in order to meet wordwide and European reg- ulations on emissions control and safety. To achieve that goal, the efficiency of exhaust gases after-treatment needs to be increased [1].

1.1 Nitrogen Oxide

The source of NOx emissions for heavy-duty vehicles is a result of diesel com-

bustion. Nitrogen oxide (NOx) consists of two main compounds, nitrogen

dioxide (NO 2 ) and nitric oxide (NO). Diesel engines only produce NO which

subsequently oxidizes to NO 2 . It also produces N 2 O which has very little pol-

luting effect. There are three mechanisms by which nitrogen oxide is formed

in diesel engine: Zeldovich, Fenimore and Fuel. Zeldovich is the formation of

NO through chemical reactions between N 2 and O 2 at high temperature dur-

ing the premixed (where the fuel vapour spontaneously ignites and the reaction

gets faster continuously) and diffusion(where the rate of combustion is limited

by the diffusion of reagents and products) phases of diesel combustion. The

high temperature rise during the premixed phase, and the reaction between ni-

trogen and oxygen occurring during the diffusion phase, favor the formation of NOx [2, 3].

The Fenimore mechanism, also known as prompt NOx mechanism, consists of several reactions between N 2 and hydrogen cyanide (HCN) to form NO. The reactions mainly occur at the flame front where the oxygen level is low due the reaction of oxygen in the atmosphere with the fuel. The Fuel mechanism occurs when fuel, during combustion, releases nitrogen bound radicals that decompose to NO [1].

The most dominant NOx mechanism in diesel engine is Zeldovich. It highly depends on the temperature and air-fuel ratio within the reaction region. The higher the combustion temperature, the higher nitrogen oxide formation. How- ever, the higher the combustion temperature, the higher the engine’s thermo- dynamic efficiency. Thus, reducing combustion temperature to reduce NOx emissions is not an option for truck manufacturers. Instead, after-treatment technologies such as Selective Catalytic Reduction (SCR) is implemented to treat exhaust gases before released to the environment [1].

1.2 Selective Catalytic Reduction

Selective Catalytic Reduction (SCR) is an exhaust aftertreatment technology used in heavy-duty diesel engines to reduce NOx emissions. In SCR a catalyst responsible for reducing NOx to nitrogen gas and water using ADBlue, a mix- ture of urea and water. Urea is an organic compound with the chemical formula (NH 2 ) 2 CO and accounts for 33 percent of the ADBlue. When injected to ex- haust gases at 500°C, the urea decomposes into carbon dioxide and ammonia (NH 3 ). In fact, a high temperature in the exhaust gases is extremely impor- tant for urea decomposition as it allows the formation of ammonia (NH 3 ). The latter is important during the after-treatment of exhaust gases because NH 3 is the main reducing agent during the break down and reduction of NOx (Figure 1.1). The main chemical reactions in selective catalytic in selective catalytic reduction are presented in equations (1.1-1.3).

4 NO + 4 NH ₃ + O ₂ −−→ 4 N ₂ + 6 H ₂ O (1.1)

2 NO ₂ + 4 NH ₃ + O ₂ −−→ 3 N ₂ + 6 H ₂ O (1.2)

NO + NO ₂ + 2 NH ₃ −−→ 2 N ₂ + 3 H ₂ O (1.3) The higher the temperature in the exhaust pipes, the higher the evaporation of urea, and ultimately the higher NH 3 concentration and the lower NOx emis- sions. Therefore, high temperature in exhaust pipes is extremely important for efficient NOx reduction in SCR. A high temperature in the pipe is achieved via conduction of heat from the exhaust gases into the pipe wall. Therefore, a high thermal conductivity in the pipe material is essential.

Figure 1.1: Schematic representation of selective catalytic reduction (SCR). The num- bers of symbols representing each species are not necessary representative of the sto- ichiometry.

1.2.1 Urea Deposition

Urea deposition or urea build-up occurs when droplets of urea from the AD- Blue solution don’t evaporate. Instead, the urea droplets crystallize on top of the exhaust pipe surface, forming a white solid (Figure 1.2). The urea build- up forms an obstacle for the flow of exhaust gases which affects the efficiency of nitrogen oxide reduction process. In severe cases, the urea deposition can lead to the exhaust pipe becoming clogged, which will prevent the engine from working properly. Thus, urea deposition in exhaust pipes should be avoided.

1.3 Hydroformed and 3D Printed Exhaust Pipes

Hydroforming is the manufacturing process that Scania uses to produce ex-

haust pipes with complex shape and high durability. Metal additive manufac-

Urea Build-up

5 cm

Figure 1.2: Urea deposition in 3D printed pipe (left) and hydroformed pipe (right).

turing is the process used by designers to print prototype pipes and perform emissions tests before mass production. Results from previous experiments showed superior performance of 3D printed prototypes compered to compo- nents hydroformed. Less urea deposit was formed in 3D printed parts, mean- ing higher amount of urea was evaporated. This will lead to a more efficient reduction of NOx (Figure 1.2). One possible explanation is higher heat transfer in 3D printed pipes. In fact, higher amount of heat was transferred to exhaust gas flow causing a higher evaporation of urea within the 3D printed exhaust pipes. Therefore, it is important to investigate the effect of additive manufac- turing on heat transfer.

1.4 Metal Additive Manufacturing

Additive Manufacturing consists of manufacturing parts by adding new layer

of material on previously completed layer, then fusing or melting particles

together using energy input such as laser, electron beam or a binder [4]. The

process is repeated until the parts are completed. Additive manufacturing has

shown great potential in building complex design components without the

need of tooling. This leads to shorter lead-time compared to conventional

methods. For this reason, rapid prototyping using metal additive manufactur-

ing is a huge interest for designers that need to test their designs in the fastest

and most efficient way. This process has also several drawbacks that limits

its usage in large scale. For example, high energy consumption, anisotropy

of mechanical properties, pores and defects of final products, and limited

material choice.

Metal additive manufacturing can be divided into two categories: pow- der bed and directed energy deposition:

• Powder bed systems consist of spreading metallic powder across the sur- face of the bed build followed by sintering or melting of the metallic particles with an energy source: laser, electron beam or binder (Binder Jetting). The powder is spread again and the process is repeated until the part is created.

• Directed energy deposition consists of melting or fusing materials as it is deposited. Such techniques are divided into two types of system, powder feed and wire feed system:

- Powder feed systems allow metallic powder to be conveyed through a nozzle and then melted or sintered simultaneously by a laser to create the desired shape. This system is suitable for building mod- erate build volumes.

- Wire feed systems use wire as feedstock instead of powder. The energy source, either laser electron, beam or plasma arc, melt the wire at the right deposition in order to build the printed structure.

This type is suitable for large build volumes and generally requires more machining [5].

For sake of simplicity, only powder bed systems are studied in this research, more specifically, selective laser melting (SLM). It is also the only technology available for this project.

1.4.1 Selective Laser Melting

SLM is a powder bed system that allows the fabrication of structural compo- nents thought the complete melting of the metallic powders using laser as an energy source. The SLM process is strongly affected by three main factors:

the properties of the metallic powder, the printing machine and the printing parameters. In this study, the influence of the SLM parameters is analyzed.

1.4.2 Selective laser melting parameters

SLM setting parameters are divided into four categories: base, up-skin, core

and contour scanning parameters [6]. Base refers to the parameters of the first

two layers of the printed parts, up-skin refers to the last three layers and core

parameters controls the layers in between. The contour parameters refer to

Figure 1.3: Selective laser melting parameters.

the contouring of printed samples (Figure 1.3). In this research, the following parameters are investigated:

• Core parameter - scan speed: is the speed at which the laser is moved across the surface of the metal powder.

• Core parameter - hatch distance, is the distance between two consecutive tacks of the laser.

1.5 Problem Description

Designers at Scania rely heavily on SLM prototypes to test their designs.

Shorter lead times and lower costs are the main incentives for choosing SLM

over hydroforming. However, the discrepancies in results of urea deposition

tests for hydroformed and printed pipes makes SLM unreliable as a manu-

facturing process. In addition, there is a lack of knowledge on how to adjust

SLM parameters to efficiently print prototypes with similar properties of hy-

droformed ones. A lot of research has been done on a fundamental study level

to investigate the effect of SLM parameters on surface quality and mechanical

properties [7–10]. However, little has been done in terms of the effect of SLM

defects on thermal conductivity.

1.6 Goal of This Thesis

The goal of this project is to investigate the effect of various values of scanning speed and hatch distance on the density, grain size and thermal conductivity of 316L stainless steel samples. These properties are chosen because they have direct impact on the heat transfer within the pipes and thus an impact on the NOx reduction. This will allow designers to optimize the build time of the samples while achieving close properties to hydroformed 309 stainless steel, meaning hight density, minimum porosity and a thermal conductivity of 15.6 W m ^-1 K ^-1 (the value of the hydroformed components). The correlation between SLM parameters and thermal conductivity, density and grain size is calculated to find which parameters are highly affected by SLM process.

1.7 Scope

The properties of 3D printed stainless steel is affected by several SLM parameters. To achieve a throughout understanding of metal additive manu- facturing, previous research on selective laser melting process and its effect on fatigue, strength and microsturcture were studied.

Considerable time was devoted to understand the exhaust system of Scania’s truck since this project is revolved around the manifacturing of prototypes for the exhaust pipes.

This work is considered a foundation for the investigation of urea depo- sition and heat transfer within exhaust pipes.

1.8 Limitations

Several limitations are considered within this project:

• All experiments were conducted at KTH Royal Institute of Technology using only the equipment available at the university. At the start of the project, it was intended to perform experiments at labs in Scania AB, but this was not possible due to restrictions imposed as a result of the Covid-19 pandemic.

• Only selective laser melting is investigated and 316L stainless steel was

used. No other material or printing process was studied. This is to rep- resent the prototyping procedure in place at Scania AB.

1.9 Methodology

Thirteen SLM samples with different printing parameters are evaluated. Vari- ant combinations of scan speed (742, 928, 1114 mm s ^-1 ) and hatch distance (0.08, 0.12, 0.10 mm) is used to print the samples. These specific values repre- sent ( ± 20 % ) away from the default value for the hatch distance and the scan speed parameters in the 3d printer at Scania AB. No larger variance is permit- ted by the software that controls the machine. The thermal conductivity coef- ficient is measured for all the samples. Microstructural analysis is performed using optical microscopy to study different defects of the SLM samples. Each sample’s density is compered to a metric representing the printing parame- ters. Finally, all experiment results are analyzed to find the most suitable SLM parameters for Scania’s prototype pipes.

1.10 Ethical, Environmental and Economic Impact

There are no significant ethical considerations about this research, but it has indirect effects on sustainable development. This project contributes to the following United Nations Sustainability Goals (SDGs):

• SDG Number 9: industry innovation and infrastructure. The project aims to improve selective laser melting process which is considered an innovative manufacturing process.

• SDG Number 12: responsible consumption and production. Selective laser melting additive manufacturing allows the optimizations of mate- rials. Thus, it contributes to a responsible production in the automotive industry.

• SDG Number 13: climate action. Selective laser melting can replace

outsourcing from big industrial countries such as China. Local manu-

facturing decreases emissions related to shipping leading to less green-

house gas emissions.

Chapter 2 Background

2.1 Influence of Selective Laser Melting Pa- rameters

Several studies have been conducted previously to improve the SLM process and to optimize the properties of printed parts. Increasing laser power results in wider melt pools, that improved the surface texture of the samples [4]. Low laser energy results in incomplete melting of powder which leads to the cre- ation of metallic spherical particles and irregular cavities. Using a laser with a power that is too high results in a large and wide melt pool that creates un- even surfaces. Thus, it is concluded that very high and very low laser power leads to low surface quality. Similarly, increasing scanning speed leads to an increase in defects such as cracks and pores due to the unmelted particles [11].

In the same context, tensile and fatigue strength increases with an increase in the component density and a decrease in porosity [12]. Changes induced by heat treatment at different temperature (573, 873, 1273, 1373 and 1673 K) has no effect on phase formation as 316L is austenitic from room temperature to the temperature at which it begins to melt and no precipitates are expected to form [12]. An alternative to increase the strength for the SLM 316L samples is the generation of high dislocation density at grains and cell boundaries. This can be achieved for conventional 316L stainless steel. However, SLM materi- als are already produced in a state of high density dislocation due to the rapid solidification during the SLM process. Therefore, increasing the strength of SLM 316L materials is very difficult.

Heat treatment leads to deteriorated tensile strength due to the decrease in

barriers that hinder the movement of dislocations [13]. However, at high tem-

perature, heat treatment leads to a decrease in anisotropy, meaning the yield

strength becomes identical in all orientations [14]. Thus, there is a trade-off between anisotropy and the strengh of SLM samples during heat treatment.

2.2 Heat Transfer

Heat transfer occurs whenever there exists a temperature difference within a medium [15]. Heat transfer can take many forms and various mechanisms depending on where the temperature gradient occurs. There are three modes of heat transfer:

• Conduction: The atomic or molecular interactions within the medium allows a transfer of kinetic energy from high temperature to low temper- ature eventually causing a transfer of heat across the medium.

• Convection: The motion of fluid from one location to another along with the flow of the matter within that fluid.

• Radiation: Electromagnetic waves.

In this study, conduction of heat within the 3D printed samples is investigated because there is direct physical contact between the heat source and the ma- terial. Also temperature gradient occurs within the stainless steel. Thus, heat transfer through the exhaust gases is neglected, meaning convection heat trans- fer is not covered in this research. Similarly, radiation is neglected as the tem- perature of exhaust gas are 500°C which will result in a very low amount of radiative heat transfer though electromagnetic waves. The heat transfer within stainless steel is highly affected by its properties, especially porosity, density and microstructure. Therefore, it is important to investigate the effect of SLM process on these properties.

2.2.1 Thermal Conductivity Measurement Techniques

There are several ways to measure thermal conductivity, depending on mate-

rials properties and median measurement temperature. The methods to mea-

sure thermal conductivity are classified into two classes: steady-state and non-

steady state methods. Steady-state methods measure the thermal conductivity

when the temperature difference is constant with respect to time. This tech-

nique is mostly used for materials with low thermal conductivity such as in-

sulation materials, polymers and composites [16, 17]. An example of steady-

be placed between a hot plate and heat sink. The drop in temperature across a known length of sample allows the calculation of the thermal conductivity using Fourier’s Law. The goal of this project is to calculate the thermal con- ductivity of stainless steel samples, a material with high thermal conductivity.

Thus, the steady-state methods are deemed unsuitable. A non-steady method must be used to calculate thermal conductivity of the different SLM stainless steel samples. Non-steady or transient methods are usually measuring time- dependant energy dissipation on a sample [16, 19]. There are several transient techniques to measure the thermal conductivity. However, these methods re- quire very specific equipment not available at KTH Royal Institute of Tech- nology. For this reason, the calculation of the thermal conductivity of the samples is done using the heat equation and temperature change within each sample when exposed to heat. More details about this experiment is discussed in the Methods section.

2.2.2 Heat Equation

The law of heat conduction, Fourier’s Law, states that the rate of energy trans- ferred per unit area and time is proportional to temperature gradient ^∂T _∂x and thermal conductivity, k (equation 2.1):

q _x = −k ∂T

∂x (2.1)

the heat flux, q x , is the rate of energy transferred per unit area per unit time in the direction parallel to the direction of the heat transfer. Thermal conductiv- ity is transport property that indicates the rate at which energy is transferred by diffusion process [15, 20]. The minus in the equation is to express the di- rection of the heat transfer, towards the lowest temperature. Thus, Fourier’s law determines the heat flux in terms of the characteristics of the material and the manner by which the temperature varies within the studied medium [15].

Note, heat flux q x is different than Q, the heat rate of conduction. Q is a prop- erty that measures the heat transfer through an area A perpendicular to the direction of transfer [15].

Q = −Aq _x Thus, Fourier’s law is also expressed as follows:

Q = −Ak ∂T

∂x

The effect of the heat flow is described by the following equation:

q _n = cm∂T

q _n is the net heat flowing through the material, c is the specific heat capac- ity and m is the mass the material. m can be expressed as the product of volume,V , and density, ρ.

q _n = cρV ∂T

Volume is the product of the cross-section through which the heat flows, A, and some small distance parallel to the direction of heat flow, ∂x:

q _n = cρ∂xA∂T

Since the heat, q n , is transferred within the time ∂t, then this equation is also valid for the heat flow rate ^∂q _∂t

and temporal change in temperature ^∂T _∂t [21]:

∂q _n

∂t = cρ∂xA ∂T

∂t Q n = cρ∂xA ∂T

∂t

The heat rate by conduction, Q n , is determined by the difference between heat flow, Q ⁱⁿ , entering the system and the heat flow leaving the system, Q ^out [21]. Therefore, Q ⁿ can be written as follows:

Q _n = Q _out − Q _in = −dQ Thus,

−∂Q = Aρc∂x ∂T

∂t

∂T

∂t = − 1 Aρc

∂Q

∂x replacing with the Fourier’s Law equation:

∂T

∂t = − 1 Aρc

∂(−kA ^∂T _∂x )

∂x

∂T

∂t = k ρc

∂( ^∂T _∂x )

∂x

∂T

∂t = k ρc

∂ ² T

∂x ²

Temperature, T , is generally a function of time, t, and distance, x. Thus the above equation becomes a partial differential equation with respect of time and space [21]:

∂T (x, t)

∂t = k

ρc

∂ ² T (x, t)

∂x ² This may be solved as follow:

∂T (x, t)

∂t ' T (x, t + ∆t) − T (x, t)

∆t

∂ ² T (x, t)

∂x ² ' T (x + ∆x, t) − 2T (x, t) + T (x − ∆x, t) (∆x) ²

Substituting these approximations into the heat equation:

T (x, t + ∆t) − T (x, t)

∆t = k

ρc

T (x + ∆x, t) − 2T (x, t) + T (x − ∆x, t) (∆x) ²

Thus, thermal conductivity K is expressed as follow:

k = T (x, t + ∆t) − T (x, t)

T (x + ∆x, t) − 2T (x, t) + T (x − ∆x, t) (∆x) ²

∆t cρ (2.2)

For all SLM samples, the temperature is measured at three positions as

a function of time, then substituted in equation (2.2) to calculate the thermal

conductivity of each sample, k, at the center point.

Chapter 3 Methods

3.1 SLM samples

Laser speed and hatch distance are the major SLM influencing factors for this research. Samples are fabricated as ten rectangular plates (50 × 30 × 50 mm) and three square plates (30×30×10 mm)) mm) from 316L stainless steel pow- der with the nominal composition given in table 3.1 and maximum particle size of 63 µm. 316L stainless steel is chosen for this study because of its similar properties (corrosion resistance, thermal properties, strength and fatigue) to the 309 stainless steel used for conventional exhaust pipes and is widely avail- able as a powder for additive manufacturing. The 316L samples are printed using EOS M 290 powder bed 3D printer equipped with a fiber laser, argon gas environment, (250 × 250 × 325 mm) mm building volume and automatic powder-layering apparatus with a layer thickness of 40 µm. The samples are fabricated using a standard pre-defined scanning strategy in which layers are not rotated and have the same orientation. No post-processing treatment is done to the 316L samples. The samples are printed with the same laser power (214 W) and layer thickness (0.04 mm) but different values of scan speed and hatch distance (table 3.2). Figure A.3 illustrates the final dimensions of the samples used for this project.

Table 3.1: Nominal composition of 316L. All values are in wt% [22].

Fe Cr Ni Mo C Mn Cu P S Si N

Bal. 17.00- 13.00- 2.25- 0.030 2.001 0.500 0.025 0.010 0.750 0.100

Table 3.2: SLM samples printing parameters.

Samples Scan Speed (mm/s) Hatch Distance (mm)

1 742 0.08

2 742 0.10

3 742 0.12

4 928 0.08

5 928 0.10

6 928 0.12

7 1114 0.08

8 1114 0.10

9 928 0.10

1 1114 0.12

3.2 Microstructural Characterization and Grain Size Measurement

The samples for microstructural analysis are cut, ground and polished accord- ing to standard metallographic techniques recommended for 316L stainless steel [23]. To reveal the microstructure, samples are etched for 10s using Car- penter’s stainless steel etchant composed of FeCl 3 , CuCl 2 , hydrochloric acid, nitric acid and ethanol. The microstructural characterization is then carried out using optical microscopy. The section chosen for the microstructure anal- ysis is the one perpendicular to the build direction and parallel to the scanning direction since this section allows better observation of the shape of pores elongated along the layer boundary. High-aspect-ratios pores have preferred alignment with layer boundaries in the scanning direction rather than the build direction [14]. In addition the pores resulting from the lack of fusion of parti- cles, lack-of-fusion pores, the balling effect and cracks are investigated in this study. Grain size measurement is carried out using the software ImageJ us- ing the mean linear intercept method [24]. This consists of drawing randomly several lines on the micrograph and calculating the average grain size using the following equations:

d = n

l

where d is the grain size, n is the number of times each line intersects with the grain boundaries and l is the line length. Then, the average grain size ( d) is: ¯

d = ¯

i=n

X

i=1

d _i n

3.3 Relative and Energy Density Calculation

The density of the samples is measured using Archimedes method. Samples are measured in air and water using an analytical balance with ±0.01g precision. The equation used to calculate the specific density for each sample is :

ρ _s

ρ _w = M _a M _a − M _w

with ρ s the density of the sample, ρ w the density of water, M a weight of sample in air and M w weight of sample in water. The relative density is calcu- lated as the ratio of specific density of the sample ρ s and the theoretical density of 316L (7.9 g/cm ³ ):

ρ _r = 100 ρ _s 7.9

Another factor that affects the properties of the SLM printed samples is energy density,"E". It is a value that groups the most influential processing parameters for SLM. It allows the characterization of porosity and melt pool.

The energy density is calculated as follows [25]:

E = P

V DL

where E is the energy density (J/mm ³ ), P is the laser power (W), V is the scanning speed (mm s ^-1 ), D is the hatch distance (mm) and L is the layer thickness (mm). To analyze the extent to which the energy density can predict the relative density, determination coefficient R ² is calculated to determine the correlation between energy and relative density of the printed 316L samples.

3.4 Thermal Conductivity Measurement

There are many techniques that measure accurately the thermal conductivity

method was derived to rank the thermal conductivity of the samples. The re- sults are considered scientifically valid given the fact that all samples have the exact same experimental set-up. The experiment is conducted at room tem- perature. Each sample (50 ×10 ×10 mm) is fixed using a stand and heated on the other end with a hot air gun that reaches 500°C ( Figure 3.1). The distance between thermocouples is 12.56 mm and is measured using a micrometer with a precision of ±0.005 mm. Three thermocouples are spot welded in those po- sitions. An insulation panel from foam is placed above the thermocouples to avoid air blowing directly on thermocouples. This set-up will measure ther- mal conductivity along the metal and not the temperature of the air itself. The temperature in each thermocouples is measured every two minutes for a du- ration of six minutes as the change of temperature after six minutes was very low. This process is repeated for all samples. No boundary or initial condi- tions are needed for this experiment as only measured temperatures will be substituted into the heat equation. Thus, the heat transfer within the samples depends solely on the physical and mathematical law represented by the heat equation:

∂T (x, t)

∂t = k

ρc

∂ ² T (x, t)

∂x ²

It allows the use of different values of temperature T , time t and distance x of the 316L samples to calculate the thermal conductivity K:

k = T (x, t + ∆t) − T (x, t)

T (x + ∆x, t) − 2T (x, t) + T (x − ∆x, t) (∆x) ²

∆t cρ considering the following assumptions:

• One-dimensional heat flow and temperature gradient.

• Heat flux is constant within the direction of the thermal energy transfer.

• The thermal conductivity is constant along the sample.

• The specific heat capacity is constant for all samples since they all have the same chemical composition and no phase transformation.

• The length of the samples is much higher than their thickness, thus, the distribution of heat is only a function of distance x and time t.

• The sample gains and loses heat only through its ends.

• The heat dissipates at the end of the samples.

• No other losses to the environment.

Figure 3.1: Thermal conductivity experiment set-up.

Chapter 4 Results

4.1 Relative and Energy Density Results

The density of each sample is measured using the Archimedes method. The values are then used to calculate the relative density (table 4.1).

4.2 Optical Micrographs of SLM samples and Grain Size Calculation

Optical macrographs were taken of the SLM samples along the horizontal sec-

tion normal to build direction, with the exception of sample number 9 due to

an error during sample preparation (Figures 4.1- 4.3). It was not possible to

take a new section of sample number 9 due to a lack of material. The images

in this report are representative areas of the SLM samples. To analyze the mi-

crostructure, the samples are separated into three main groups. Group A are

the samples fabricated with an energy density varying from 40J mm ^-3 to 60J

mm ^-3 . Group B and C are manufactured with an energy density ranging from

60J mm ^-3 to 80J mm ^-3 and 80 to 100J mm ^-3 respectively. The samples in group

A show wider and larger pores than groups B and C. In addition, it is observed

that small pores are the most abundant for all samples except sample number

10. The pores are randomly distributed along all samples and not clustered

in one particular region. Small equiaxed are present at the boundaries of all

samples. Columnar grains are observed in the micrographs growing in the

scanning direction. Cellular structure and balling effects are mostly present in

sample number 1 (Figure 4.3). In addition, sample number 9 (section parallel

to the building direction), exhibits the overlapping meltpools and multiple gas

Table 4.1: Relative and energy density of SLM samples.

Samples Energy

Density

Measured Density

Theoretical Density

Relative Density (J mm ^-3 ) (g cm ^-3 ) (g cm ^-3 ) (%)

1 90 6.80 7.9 86

2 72 6.98 7.9 88

3 60 6.98 7.9 88

4 72 6.69 7.9 84

5 58 6.98 7.9 88

6 48 6.75 7.9 85

7 60 6.75 7.9 85

8 48 6.98 7.9 88

9 58 6.87 7.9 87

10 40 6.47 7.9 81

pores caused by a lack of fusion between layers. In terms of grain size, SLM

stainless steel samples in Group A exhibits the smallest grain size and most

refined grains ranging from 10 to 13µm. Whereas Group B and C exhibits

larger grain size ranging from 10 to 16µm and 20 µm respectively (Table 4.2).

5 6

8 9

10

Figure 4.1: Micrographs of SLM samples with energy density ranging from 40 J mm ^-3

to 60 J mm ^-3 . The images are labelled with the sample number.

2 3

4 7

Figure 4.2: Micrographs of SLM samples with energy density ranging from 60 J mm ^-3

to 80 J mm ^-3 .The images are labelled with the sample number.

Figure 4.3: Micrographs of sample number 1 with the highest energy density:90J

mm ^-3 .

4.3 Thermal Conductivity Results

The thermal conductivity of each sample was calculated from the temperature, position and time data collected in experiments (table 4.2).

Table 4.2: Energy density, grain size and thermal conductivity of SLM samples.

Samples Energy Density Average Grain Size Thermal Conductivity

(J mm ^-3 ) (µm) W m ^-1 k ^-1

1 90 17 14.1

2 72 15 9.8

3 60 13 25.6

4 72 11 6.2

5 58 10 12.2

6 48 13 12.5

7 60 15 8.9

8 48 9 12.2

9 58 12 13.9

10 40 10 8.5

Chapter 5 Discussion

5.1 Relative and Energy Density

The decrease in density of SLM samples is usually caused by the formation of pores, balling or cracking defects. In fact, gas inside the pores is consid- ered the second largest phase by volume in 316L SLM material [14]. Such defects can be generated through various mechanisms, such as, lack of fusion between layers or gas absorption during the SLM process. The determina- tion coefficient was found to be 0.57 proving a moderate association between energy density and relative density of the samples (Figure 5.1). A quadratic polynomial regression line was selected for this graph. From previous research paper [26, 27], it was found that both high and low energy density lead to low density with some maximum in the middle. Therefore, a quadratic equation was chosen. Similar results were found in previous research papers [28, 29].

A low energy density produced meltpools too small to sufficiently fuse the build layers during SLM printing [14]. The same is common for very high en- ergy density in which keyhole shape meltpools are formed, leaving space for gas pores to form inside the sample [14] (Figure 5.2). Good fusion of melt- pools are achieved through a medium energy density ranging from 60 to 75 J mm ^-3 (table 4.1). However, two samples stand out from this theory: sample number 8 (high relative density - 88 % and low energy density - 48%) and sample number 7 (60 J mm ^-3 energy density and low relative density 85 %).

Those anomalies and the poor coorelation between energy density and relative

density can be the result of errors during density measurement, non-accurate

balance, the possibility of air stuck in sample’s pores or random chance. The

Archimedes method can be used as an initial experiment to analyse the relative

density of the samples. However, it is not the most accurate technique to an-

10/20/20 1

75 80 85 90 95 100

35 45 55 65 75 85 95

Re la %v e D en si ty (%)

Energy Density (J/mm

)

Determina%on Coefficient = R

= 0.57

Figure 5.1: Dependence of relative density on energy density.

alyze the porosity. The Archimedes method does not describe the size, shape nor the distribution of the pores. It is then recommended to use micrographs analysis or X-ray computed tomography techniques for more accurate porosity analysis.

5.2 Microstructure and Grain Size

The equiaxed grains and the elongated grains at the borders seem to be at-

tributed to the thermal gradients between layers and fast solidification induced

by the SLM process [30]. Gas pores are the most common defects in all sam-

ples. The balling effect is only present in sample 1 due to insufficient melting

of some powder particles. A cellular structure is formed around large pores in

the sample (Figure 4.3). This structure can be attributed to high cooling rate

that reduces the grain growth after nucleation or residual stress distribution

due to temperature gradient between unmelted 316L particles and the melt-

pool. It is then concluded that grain size of the samples can be managed by

Normal Melt Pool

Keyhole Melt Pool

Pores Small Melt Pool

Figure 5.2: Type of SLM meltpools.

6 8 10 12 14 16 18 20

30 40 50 60 70 80 90 100

Av er ag e G ra in S iz e (μ m )

Energy Density (J/mm

)

90

48

Correlation Coefficient = 70.71%

Figure 5.3: Dependence of average grain on energy density.

density and grain size with a correlation coefficient of 71%. Random resid- uals are observed in the graph, meaning the model could be right but more evidence is needed. The increase in energy density leads to an increase in the average grain size due to the low cooling rate, meaning the material has high residual heat accumulation that results in the materials spending more time at high temperature. This leads to grain coarsening. Therefore, an energy density higher than 60 J mm ^-3 is deemed the most suitable to get high average grain size. In addition, it is observed that increasing scanning speed, decreases the average grain size of the samples (table 4.2). This seems to be the result of the high cooling rate which decreases the time available for grain coarsening.

5.3 Thermal Conductivity

The correlation coefficient between thermal conductivity and grain size is 15%

with random residuals (Figure 5.4). Thermal conductivity increases with in- creasing average grain size [31]. That is attributed to grain boundaries that restrict phonon travelling through the lattice. Therefore, increasing the fre- quency of the grain boundaries by decreasing the grain size leads to a decrease in thermal conductivity. In contrast, samples number 3 and 7 show different results probably due to errors during thermal conductivity or grain size mea- surement. Furthermore, the correlation coefficient between relative density

2.0 7.0 12.0 17.0 22.0 27.0 32.0 37.0

7 9 11 13 15 17 19

Th ermal Co nd uc/vi ty (W.m

K

)

Average Grain Size (μm)

Correlation Coefficient = 15%

Figure 5.4: Dependence of thermal conductivity on grain size values.

and thermal condutivity is 36% with random residuals (Figure 5.5). The in- crease in relative density increases thermal conductivity due to the lack of de- fects and pores that hinder the heat flow within SLM samples. No correlation

2 7 12 17 22 27 32 37

5.00 5.50 6.00 6.50 7.00 7.50 8.00 8.50 9.00

Th ermal Co nd uc/vi ty (Wm

K

)

Density (g/mm

)

Correla/on Coefficient = 35.70%

Figure 5.5: Dependence of thermal conductivity on density.

between energy density and thermal conductivity is observed (Figure 5.6).

Samples number 9 and 5 have the same energy density, yet different thermal conductivity. This is attributed to the printing strategy of the samples (Figure A.1). During the thermal conductivity experiment, both samples had the same dimensions (50 × 10 × 10mm) and the same experimental set up but different printing layer direction. The latter is parallel to the heat flow for sample num- ber 9 and normal to heat flow for sample number 5. According to the result from the thermal conductivity, layers direction seems to affect the thermal con- ductivity of the materials. Sample number 5 had lower thermal conductivity value than sample number 9. The high number of grain boundaries formed be- tween the print layers acts as a barrier for heat flow and decreases the thermal conductivity of the material (Figure 5.7). However, the results from the ex- periments may be inaccurate because of the large uncertainties in the thermal conductivity experiments. More accurate technologies are needed to increase the certainty of this conclusion.

There are several limitations for the thermal conductivity experiment, namely,

the low temperature reached at the end of the sample. Using the experimental

set-up only 100°C could reached the samples instead of 500°C , the tempera-

Correla'on Coefficient = 3%

0 5 10 15 20 25 30 35 40 45

30 40 50 60 70 80 90 100

Th ermal Co nd uc'vi ty (W m

K

)

Energy Density (J/mm

)

60

72

Figure 5.6: Dependence of thermal conductivity on energy density.

ture of the exhaust gas in a truck. However, thermal conductivity can increase linearly with temperature [32, 33]. Thus, the ranking of samples’ thermal conductivity is not likely to change when the temperature increases. Addi- tionally, austenitic stainless steel 316L is not subject to phase transformation up to 500°C. However, prolonged exposure above 1000°C can affect the mi- crostructure of the material and its thermal conductivity. Another limitation of the thermal conductivity experiment is the isolation of the sample during measurements. Although an insulation panel is placed above the sample in the experiment, heat loss is unavoidable for this experimental set-up which can affect the accuracy of the thermal conductivity values and consequently the ranking of the samples.

5.4 Propagation of Error

Uncertainties in experiments come from variety of sources, for instance, sys- tematic errors attributed to faulty equipment, incorrect calibration or usage of instruments [34]. Random errors results from the inability to control the envi- ronment of the experiments leading to inability to take precise measurement.

These errors can propagate during the calculation of the thermal conductivity

Figure 5.7: Effect of SLM layer boundaries on heat flow from the exhaust gas.

For sake of simplicity, all errors in this research are considered random and calculated following the error propagation formula by combining different errors of the equations used in this research: thermal conductivity, density and average grain size equations. As shown in Figure 5.8 and 5.9, it is found that thermal conductivity and density’s error bars overlap for all the samples.

Therefore, the experimental setting used in this research cannot scientifically

prove that thermal conductivity or density values change due to the printing

strategies and settings used in this project. After calculating the uncertainty

values, it is concluded that there is a significant possibility that the values can

be the same which affects the accuracy and validity of the samples’ ranking. It

is advise to use more accurate machine to measure thermal conductivity such

as the Transient Plane Source or Modified Transient Plane Source which can

be used for wide range of materials including metals.

0 5 10 15 20 25 30 35 40 45

30 40 50 60 70 80 90 100

Th ermal Co nd uc/vi ty (W m

K

)

Energy Density (J/mm

)

Figure 5.8: Error bars overlap for thermal conductivity values.

70 75 80 85 90 95 100

35 45 55 65 75 85 95

Re la %v e D en si ty (%)

Energy Density (J/mm

)

Figure 5.9: Error bars overlap for relative density values.

Chapter 6 Conclusion

The effect of energy density on the thermal conductivity, density and mi- crostructure of 316L stainless steel has been investigated. Due to high error propagation during experiments set-up, accurate ranking of these properties is not reached. It is recommended to use high precision technologies to measure the heat conductivity and density in the future. Correlations between different parameters is investigated, table 6.1:

Table 6.1: Correlation summary.

Parameter A Parameter B Correlation

Energy Density Relative Density R ² = 0.57%

Energy Density Grain Size 70%

Energy Density Thermal Conductivity 3%

Thermal Conductivity Relative Density 35%

Thermal Conductivity Grain Size 15%

• Thermal Conductivity seems to be more influenced by relative density and the direction of printing layers than energy density and grain size.

• Energy density between 60-80 J mm ^-3 demonstrates the least porosity according to the microstructure analysis.

• It is then recommended to use a scan speed of 742 mm s ^-1 and a hatch

distance of 0.12 mm to print prototypes with the least porosity and thus

more similar properties to the 309 hydroformed stainless steel.

Chapter 7 Future Work

Urea deposition on exhaust pipes is very likely to be affected by thermal con- ductivity, energy density and microstructure of the stainless steel. It is then important to investigate the effect of these properties on the urea deposition.

Additionally, more advanced technologies should be used to measure the ther-

mal conductivity and density of the samples. This will allow more accurate

ranking of the SLM parameters and the possibility to decrease the build time

of exhaust pipes while maintaining good mechanical properties. The reduc-

tion of build time will simultaneously decrease the costs of printing and avoid

unnecessary expenses for Scania. Last but not least, different section of 316L

samples micrographs should be analysed, for example, the section parallel to

the build direction. It allows the analysis of the size and shape of meltpools.

Chapter 8 References

[1] Maroa Semakula. “The Formation, Effects and Control of Oxides of Nitrogen in Diesel Engines”. In: International Journal of Applied En- gineering Research ISSN 0973-4562 Volume 13, Number 6 (2018) pp.

3200-3209 (2018).

[2] Colin R. Ferguson. Internal combustion engines: applied thermo- sciences. J. Wiley and Sons, 1986.

[3] Felipe Augusto Ferreira Gomes et al. “A Simplified Method For The Computation Of The Premixed Mode Heat Release In Di Diesel En- gines”. In: 17th Brazilian Congress of Thermal Sciences and Engineer- ing (2018). doi: 10.26678/abcm.encit2018.cit18-0665.

[4] Gokhan Dursun et al. “Influence of laser processing parameters on the surface characteristics of 316L stainless steel manufactured by selective laser melting”. In: Materials Today: Proceedings (2020). doi: 10.1016/

j.matpr.2019.12.061.

[5] William E. Frazier. “Metal Additive Manufacturing: A Review”. In:

Journal of Materials Engineering and Performance 23.6 (Aug. 2014), pp. 1917–1928. doi: 10.1007/s11665-014-0958-z.

[6] Manickavasagam Krishnan et al. “On the effect of process parameters on properties of AlSi10Mg parts produced by DMLS”. In: Rapid Pro- totyping Journal 20 (Oct. 2014). doi: 10.1108/RPJ-03-2013-0028.

[7] Imade Koutiri et al. “Influence of SLM process parameters on the surface finish, porosity rate and fatigue behavior of as-built Inconel 625 parts”. In: Journal of Materials Processing Technology (Dec.

2017). url: https : / / www . sciencedirect . com / science / article / pii /

S0924013617306349.

[8] Yubao Huang et al. “Microstructure and wear properties of selec- tive laser melting 316L”. In: Materials Chemistry and Physics (July 2020). url: https : / / www . sciencedirect . com / science / article / pii / S025405842030852X.

[9] Ya’nan Song et al. “Effect of scanning strategies on the microstructure and mechanical behavior of 316L stainless steel fabricated by selec- tive laser melting”. In: Materials Science and Engineering: A (July 2020). url: https : / / www . sciencedirect . com / science / article / pii / S0921509320309515.

[10] Punit Kumar et al. “Fatigue strength of additively manufactured 316L austenitic stainless steel”. In: Acta Materialia (Aug. 2020). url: https:

//www.sciencedirect.com/science/article/pii/S1359645420306303.

[11] J. A. Cherry et al. “Investigation into the effect of process parameters on microstructural and physical properties of 316L stainless steel parts by selective laser melting”. In: The International Journal of Advanced Manufacturing Technology 76.5-8 (June 2014), pp. 869–879. doi: 10.

1007/s00170-014-6297-2.

[12] Naci Kurgan. “Effect of porosity and density on the mechanical and microstructural properties of sintered 316L stainless steel implant ma- terials”. In: Materials Design 55 (2014), pp. 235–241. doi: 10.1016/j.

matdes.2013.09.058.

[13] O.o. Salman et al. “Effect of heat treatment on microstructure and me- chanical properties of 316L steel synthesized by selective laser melt- ing”. In: Materials Science and Engineering: A 748 (2019), pp. 205–

212. doi: 10.1016/j.msea.2019.01.110.

[14] Tobias Ronneberg, Catrin M. Davies, and Paul A. Hooper. “Revealing relationships between porosity, microstructure and mechanical proper- ties of laser powder bed fusion 316L stainless steel through heat treat- ment”. In: Materials Design 189 (2020), p. 108481. doi: 10.1016/ j.

matdes.2020.108481.

[15] T. L. Bergman, Adrienne Lavine, and Frank P. Incropera. Fundamentals of heat and mass transfer. John Wiley Sons, Inc., 2006.

[16] Anabel Palacios et al. “Thermal conductivity measurement techniques for characterizing thermal energy storage materials – A review”. In: Re- newable and Sustainable Energy Reviews 108 (2019), pp. 32–52. doi:

10.1016/j.rser.2019.03.020.

[17] H. Ezbakhe et al. “Method of Testing Thermal Conductivity at Steady State for Insulating Materials”. In: Dec. 2000, pp. 1788–1791. isbn:

9780080438658. doi: 10.1016/B978-008043865-8/50373-1.

[18] Marijana Terzić et al. “Development of a single-sided guarded hot plate apparatus for thermal conductivity measurements”. In: Thermal Sci- ence 20 (Jan. 2015), pp. 226–226. doi: 10.2298/TSCI151009226T.

[19] Silas Gustafsson. “A Non-Steady-State Method of Measuring the Thermal Conductivity of Transparent Liquids”. In: Zeitschrift Natur- forschung Teil A 22 (June 1967), p. 1005. doi: 10 . 1515 / zna - 1967 - 0704.

[20] I-Shih Liu. “On Fourier’s law of heat conduction”. In: Continuum Me- chanics and Thermodynamics 2 (Dec. 1990), pp. 301–305. doi: 10 . 1007/BF01129123.

[21] Tec-Science - et al. Heat equation (diffusion equation) - tec-science.

Apr. 2020. url: https://www.tec-science.com/thermodynamics/heat/

heat-equation-diffusion-equation/#mjx-eqn-Q.

[22] url: https://www.eos.info/en.

[23] George F. Vander Voort and Gabriel M. Lucas. “Metallography and Mi- crostructures of Stainless Steels and Maraging Steels”. In: Metallogra- phy and Microstructures (2004). url: https://www.georgevandervoort.

com/images/Metallography-General/Met-of-SS-Vol-9.pdf.

[24] url: https://imagej.net/Welcome.

[25] Tao Peng and Chao Chen. “Influence of energy density on energy demand and porosity of 316L stainless steel fabricated by selective laser melting”. In: International Journal of Precision Engineering and Manufacturing-Green Technology 5 (Jan. 2018), pp. 55–62. doi: 10 . 1007/s40684-018-0006-9.

[26] Tao Peng and Chao Chen. “Influence of energy density on energy demand and porosity of 316L stainless steel fabricated by selective laser melting”. In: International Journal of Precision Engineering and Manufacturing-Green Technology 5 (Jan. 2018), pp. 55–62. doi: 10 . 1007/s40684-018-0006-9.

[27] Hengfeng Gu et al. “Influences of Energy Density on Porosity and Mi- crostructure of Selective Laser Melted 17-4PH Stainless Steel”. In: Aug.

2013.

[28] Ruidi Li et al. “Selective laser melting of a novel Sc and Zr modified Al- 6.2 Mg alloy: Processing, microstructure, and properties”. In: Powder Technology 319 (June 2017). doi: 10.1016/j.powtec.2017.06.050.

[29] Kun Sun et al. “Effect of SLM Processing Parameters on Microstruc- tures and Mechanical Properties of Al0.5CoCrFeNi High Entropy Al- loys”. In: Metals 10 (Feb. 2020), p. 292. doi: 10.3390/met10020292.

[30] Shubo Gao et al. “Recrystallization-based grain boundary engineering of 316L stainless steel produced via selective laser melting”. In: Acta Materialia (Sept. 2020). url: https://www.sciencedirect.com/science/

article/pii/S1359645420307096.

[31] Kirsti Midttømme and Elen Roaldset. “The effect of grain size on ther- mal conductivity of quartz sands and silts”. In: Petroleum Geoscience 4 (May 1998), pp. 165–172. doi: 10.1144/petgeo.4.2.165.

[32] Colin James. Smithells et al. ; Smithells metals reference book.

Butterworth-Heinemann, 1992.

[33] M. F. Rothman. High-temperature property data: ferrous alloys. ASM International, 2008.

[34] Stephanie Glen. “Systematic Error / Random Error: Definition and

Examples”. In: Statistics How To (July 2020). url: https : / / www .

statisticshowto.com/systematic-error-random-error/.

Appendix A Appendix

A.1 Build Layout of SLM samples

A.2 CAD File Screenshot of Build Layout

A.3 Picture of SLM Samples Build Platform

TRITA ITM-EX 2020: