IN
DEGREE PROJECT MECHANICAL ENGINEERING, SECOND CYCLE, 30 CREDITS
STOCKHOLM SWEDEN 2019,
Design benefits with Additive
Manufacturing from a convective heat transfer perspective
JOAKIM STORFELDT
KTH ROYAL INSTITUTE OF TECHNOLOGY SCHOOL OF ENGINEERING SCIENCES
Design benefits with Additive Manufacturing from a convective heat transfer perspective
JOAKIM STORFELDT
Master’s thesis in Aerospace Engineering
Supervisors: Emmelie Simic, Ylva Larberg, Evelyn Otero Examiner: Evelyn Otero
Abstract
Nowadays manufacturing processes are rapidly developing.
Salt-bath dip brazing is a conventional manufacturing method commonly used by Saab AB to fuse aluminium components in a high temperature salt bath. However conventional manufacturing methods have shown some limitations. Ad- ditive Manufacturing, or 3D printing, is a newer technol- ogy which has become very popular in the industry offer- ing competitive advantages regarding production time and size, and structural complexity of the components among other aspects. In this work, Additive Manufacturing is in- vestigated to assess if the performance of heat sinks can be increased compared to the salt-bath dip brazing method.
Geometrical shapes of heat sink-fins were studied by em- pirical research to compare their characteristics in air-flow, convection and pressure drop. Eight different geometrical shapes have been analyzed using Additive Manufacturing, and the control plate fins was used as a reference for com- parison with salt-bath dip brazing. It was found out that the NACA 0010 fins and Square Grid fins geometries gave the best performance with a 63% and 64% decrease in pres- sure drop per diverted energy compared to the control plate fins, respectively.
Referat
Designfördelar med Additiv Tillverkning ur ett perspektiv med konvektion och
värmeöverföring
Numera utvecklas tillverkningsprocesser snabbt. Saltbad- slödning är en tillverkningsmetod som vanligen används av Saab AB för att foga aluminiumkomponenter i ett salt- bad med hög temperatur. Dock har konventionella tillverk- ningsmetoder visat vissa begränsningar. Additiv Tillverk- ning eller 3D printing är andra tekniker som har blivit myc- ket populära i industrin på grund av dess många fördelar med avseende på produktionstid, storlek och bland andra aspekter strukturkomplexitet. I det här arbetet undersöks det om Additiv Tillverkning kan användas för att öka pre- standa på kylflänsar jämfört med Saltbadslödning. Geomet- riska former av kylflänsar studerades genom empirisk forsk- ning för att jämföra deras egenskaper i luftflöde, konvektion och tryckfall. Åtta olika geometriska former skapade uti- från Additiv Tillverkning har studerats och jämförts med en kontroll som representerar saltbadslödning. Det upptäcktes att NACA 0010-fenor och kvadratiska rutnäts-geometrier gav den bästa prestandan med en minskning i tryckfall på 63% respektive 64% per avledd energi jämfört med kon- trollfenorna.
Acknowledgements
With these words, I would like to thank friends, colleagues and supervisors who have helped me make this thesis possible. Saab AB for providing resources, time and room. Supervisors from Saab AB: Emmelie Simic and Ylva Larberg. Advisors from Saab AB: Samuel Gottfarb, Computational Fluid Dynamics and Richard Ing- man, Additive Manufacturing. Johannes Eliasson for support with Computational Aided Design. Examiner from KTH: Evelyn Otero. Christer Fuglesang for recom- mendation to Saab AB and the enabling of the following degree project. Thanks.
Joakim Storfeldt
Contents
1 Introduction 1
1.1 Beneficiaries, Ethics, Environmental impacts . . . 2
2 Background 3
3 Method 5
3.1 Test Blocks . . . 5 3.2 Full Size Heat Sink . . . 9
4 Results and Discussion 10
4.1 Simulation Environment . . . 10 4.2 Test Blocks . . . 13 4.3 Full Size Heat Sink . . . 20
5 Conclusions and Future Work 22
Bibliography 23
List of Figures
3.1 Circular Fins sketch . . . 5
3.2 Elliptic Fins sketch . . . 6
3.3 Droplet Fins sketch . . . 7
3.4 NACA 0010 sketch . . . 7
3.5 Square- and Triangular Grid Fins sketch . . . 8
3.6 Clam Sketch . . . 8
4.1 Test blocks rendered in NX 11 . . . 10
4.2 Test blocks rendered in NX 11 . . . 11
4.3 Square and Triangular Grid Fins . . . 11
4.4 Control Fluid . . . 12
4.5 Control Fins . . . 14
4.6 NACA 0006- and Clam Grid Fins . . . 15
4.7 Circular- and 45 Degree Circular Fins . . . 16
4.8 NACA 0030- and NACA 0020 Fins . . . 16
4.9 NACA 0010- and 45 Degree NACA 0010 Fins . . . 17
4.10 Elliptic- and Droplet Fins . . . 18
4.11 Square- and Triangular Grid Fins . . . 18
4.12 Test Blocks Plot . . . 19
4.13 Full Size Heat Sink Illustration . . . 20
4.14 Full Size Heat Sink Plot . . . 21
List of Tables
4.1 Simulation Environment Settings . . . 124.2 Test Blocks Results . . . 13
4.3 Full Size Heat Sink Results . . . 21
Nomenclature
AM Additive Manufacturing CAD Computer Aided Design
CFD Computational Fluid Dynamics FB Fin to base cross sectional area
KTH Kungliga Tekniska Högskolan (Royal Institute of Technology) SLM Selective Laser Melting
TRL Technology Readiness Level
Chapter 1
Introduction
Saab AB uses several methods to manufacture components in aluminium. Salt-bath dip brazing is one of them and is used to fuse components such as heat sinks. The benefit is that complex surfaces and shapes can be fused together at the same time on an inside where it would be impossible to reach with a common welding tool.
However there is also a need to mill the metal, including the restrictions that go along with that. It is hard to tell if all intended surfaces actually have been fused together and to what extent, something that will have an impact on the conductive heat transfer. There is a limit on the minimum structure width that can be dip- brazed, prices tend to rapidly grow with complexity of milled components and the lead time can be months [1].
Additive Manufacturing (AM) is a method that can compete with Salt-bath dip brazing. AM is another type of manufacturing process which has become popu- lar since Salt-bath dip brazing was developed. AM is currently being exercised at a variety of industries, for example; aerospace, dental, cars, hearing aid, highly customized parts, space, rapid prototyping etc [2]. When mentioning AM in this re- port is it mainly referred to powder bed fusion using Selective Laser Melting (SLM).
SLM is an additive method and can be used to print metal parts in 3D. The biggest concern at the moment is that the initial investment is high and such components may not have the same strength as solid metal cut by deductive methods, and more interesting for this project is, if they have the same heat transfer.
The method of AM is actually important research for Saab AB due to the level of complexity and requirements on their components. The intended system is mounted on a fighter airplane, a radar system that warns the pilot for radar lock on, incoming missiles and applies countermeasures accordingly. This system uses heat sinks to cool its circuit boards. The heat sinks needs increased heat diversion to compensate for updated circuit boards that emit more thermal energy. One solution would be to make the heat sinks larger but the equipment has an allocated space in the airplane which can not be expanded.
Heat sinks are common cooling systems for electronics and mechanical processes that require heat dissipation, and heat sink pin-fin geometries for air cooled electronics systems are studied. Heat sink fins can be compared to the wing of an airplane, for which you want certain flow characteristics over a surface. Fins of different shapes can be placed in an inline or staggered formation to give an effect on the next one, similar to a wind farm. A flow that works to maximize heat transfer to the surrounding air by convection is desired. The problem is that traditional heat sinks have a hard time meeting the requirements of new, high power, heavy dissipating electronics. It has been increasingly common to apply liquid cooling on systems
1
CHAPTER 1. INTRODUCTION
with such requirements, but this technique is not always practical to use for reasons such as they are known to clog and have to be a closed system. Therefore it is of interest to research air cooled heat sinks built by the means of AM processes, that allow for more advanced structure designs and less lead time.
The goal is to find out if geometrical thermal conductivity differs when it comes to Salt-bath dip brazing or AM. The methods will be analyzed by comparison.
Potential alternative AM geometries will be compared to Salt-bath dip brazing which is represented by a Control, plate fins that are based on the heat sink in the radar system from Saab AB which has to be updated.
The report will start with a background on AM, Salt-bath dip brazing and heat sinks in general. Thereafter the method will be described on how alternative geometries were generated and constructed followed by simulation results and discussions. The report will end with some conclusions and future work.
1.1 Beneficiaries, Ethics, Environmental impacts
The beneficiaries of the thesis are Saab AB and the public, since this report is published through KTH. Saab AB is the supplier of both the task and resources involving the project. It is also agreed that Saab AB owns all material and files that are created during the project. “Saab AB serves the global market with world- leading products, services and solutions from military defence to civil security.”[3].
This type of company could be sensitive for some ethical aspects, research for mili- tary purposes can definitely be questioned. The world is a dangerous place and the need for defence and peace keeping products is real. In this case the work is related to a radar system that warns the pilot for radar lock on, incoming missiles and applies countermeasures accordingly. Therefore it can be argued that the research of this thesis would not imply the harm on others but protect the pilot from being attacked or spied upon. On the other hand heats sinks are commonly used in both defence and attack systems, and the results of this research could be applied to an attack system at a later stage. Saab AB is again the owner of the research and decides what to do with it.
From an environmental standpoint AM should lead to less material waste. The SLM process takes place in a sealed chamber, and after a component is built, excess aluminum powder can be brushed off before it is taken outside. The excess powder would then be saved and used to print the next component. The process itself requires a lot of energy, especially lasers and motors, this does not mean that other methods require less energy. AM has a short lead time which means from a sustainable point of view that a broken part can be replaced quickly instead of scrapping the whole machine or having it idle while waiting for a replacement part [4].
2
Chapter 2
Background
In this section relevant background information is presented to provide a deeper understanding of heat sinks manufactured by the means of the manufacturing pro- cesses Salt-bath dip brazing and AM.
In the Salt-bath dip brazing process are aluminum blocks milled to a desired shape.
The metal is cleaned from any grease and oxides. A brazing alloy in the form of paste, foil or wire is applied at the points of union, depending on the location and shape of the joint. The components are assembled and held together by screws, wire or tack-welding, or self-fixing designs that can be rivet together. The components are pre-heated to 560 ℃ and then dipped in a 585-590 ℃ salt bath consisting of sodium, potassium and lithium chlorides and sodium fluoride. The temperature is close to the melting point of aluminum (660 ℃), the brazing alloy melts which fuses the two blocks together. The minimum milled structure width that can be dipped in the salt bath without melting away is 0.05 mm. The salt can be rinsed out with water and you are left with a durable product [1].
In the AM process there is a laser that is directed on a metal powder bed and selectively melts the powder into a solid. A CAD file dictates where the laser should be directed and powder will be melted. New layers of powder are successively added and melted to create a 3D solid [2]. Complex structures inside a product can be manufactured with AM without the requirement of pre- and post processing. The option for post processing is still there, examples of this are: blasting, grinding a surface smooth, laser cleaning, drill holes or fuse together parts [5]. The key here is that low- or highly detailed structures makes a minimal difference for AM. The thickness of each powder layer affects the quality of the end product, thick layers shortens the print time but thin layers provide a higher surface quality. There is room to increase manufacturing rate by balancing the layer thickness to quality and optimizing printing orientation to fit several parts next to each other in one print.
Research on heat sinks was focused on straight-fin heat sinks and pin-fin heat sinks in general. Commonly are straight-fin heat sinks thin long plates working by air blowing along the plate, pin-fin heat sinks are pillars in a formation receiving air from the side, similar to the direction a wind mill receives air. A few studies from both universities and the private sector have been published on heats sinks related to AM.
Different pin-fin geometries affect convection from the hot wall into the fluid by disrupting the flow. In Part I – Optimization of Staggered Plate Fin Heatsink, are Circular-, Squared-, Elliptical- and Parallel Plate Fins in both inline and staggered 3x3 formations investigated. The conclusions were that rounded geometries out-
3
CHAPTER 2. BACKGROUND
perform similar sharp edged shapes at higher values of pressure drop and pumping power. At lower values of pressure drop was the Elliptical Fins most effective. The fin to base cross sectional area (FB) in this test was 20 % for all shapes. Staggered formation worked best in all cases. In Part II – Optimization of Staggered Plate Fin Heatsink, fin placement is investigated. The Ratio of length-wise- to span-wise pitch of the fins were varied from 1 to 4. Results show that optimum performance may occur at ratios higher than 4. This was done with rectangular fins without overlap [6].
In the paper by Feng Zhou and Ivan Catton a similar setup is investigated. Instead of having a fin formation of 3x3 as in the previous paper, they use three fins in a row to see reactions in the flow. The geometries that is tested are Circular-, Square-, Elliptic-, Dropform-, Plate- and NACA 0050 fins. The conclusions here are that the performance of the shapes from best to worst comes in the following order: Elliptic equal to NACA, Dropform, Circular, Thin Plate then Square [7].
In the paper by Robert Smith, P.E. a 3D printed mesh is compared to a 3D printed plate-fin design. The 3D printed mesh is a lattice structure with holes in several directions. The mesh had a 30% increased wet area and the same material volume as the straight fin. The mesh greatly accelerated the air in the system, similar to partly blocking a garden hose with your finger. This is related to the pressure drop which increased by an order of five of the straight-fin heat sink and would preferably be reduced to require less pumping power. This removed any benefit with increased surface area in contact with the air and resulted in a less performing heat sink.
There might be a design that fulfills the balance between increased surface area and minimized pressure drop. The paper suggests more testing with a less dense mesh.
[8]
The Design Guide for Metal Laser Powder Bed Fusion by Johan Sönegård and Maria Warholm, will act as a reference for what fin shapes are possible to manufacture using AM. Restrictions on overhanging structures etc. must be considered.
4
Chapter 3
Method
The tools that will be used are Computer Aided Design (CAD), Siemens NX 11 and Computational Fluid Dynamics (CFD), Ansys 18.2. Through empirical tests and simulations, different design solutions are investigated. The purpose is to prove that the design freedom of AM can further push the performance of heat sinks, minimize component volume and maximize conduction. Aluminum is used as material.
3.1 Test Blocks
From the information presented in the background about fin geometries eight main shapes were chosen: Elliptic, Clam, Circular, Dropform, Square and Triangular long channel grids, NACA and the Salt-bath dip brazed plate fins named Control. The fins will be placed in a staggered formation, the Control heat sink uses this formation already and is suggested By Denpong to be better than an in line placement [6].
The Control Fins are 1 mm wide and 15.3 mm long with rounded edges.
Figure 3.1: Circular Fins sketch
The geometries has been firstly analyzed, Figure 3.1. A decision had to be made whether the outer walls of the test blocks should be straight or have a specific pattern to streamline the flow. Straight were chosen to be the most comparable, since different geometries require individual walls which would have had an impact on the performance. The size of the fins had to be decided, and how close to each other the fins should be. This was easiest with the circular fins, the distance from the center of a circle to the next in the span-wise direction could be set equal to
5
CHAPTER 3. METHOD
another in the length-wise direction, same as the span-wise pitch between the fins in the Control: 3.12 mm. All other shapes except for the NACA 0010- 45 Deg NACA 0010- Square- and Triangular Grid Fins have this span-wise pitch. In Figure 3.1, Lx and Ly are the length-wise- and span-wise pitch, respectively.
The radius of the circles was calculated from the area of a circle
Ac= πr2c (3.1)
Ac is the area of one fin. A = 144 mm2 is the total surface area covered by fin geometry in the Control. The FB of the Control is 30%, and the base area is marked with red lines in Figure 3.1, and the fin area with blue lines. FB can be calculated with the following equation
F B = Ac
LxLy (3.2)
From this, was Equation 3.1 adapted from 1 fin to 3x15 fins, in order to compute each circle radius from the total fin area in the Control A, which has 3x3 fins.
rc=
√ A
45π (3.3)
Circular Fins with radius rc= 1.01 mm was fitted in a staggered pattern along the test block.
For the Elliptic Fin in Figure 3.2, the major-axis b was chosen to be twice the size of the minor axis a, a=b/2.
Figure 3.2: Elliptic Fins sketch
They were fitted in a staggered pattern of 3x15 fins and the area can be calculated by
6
CHAPTER 3. METHOD
Ae= πab = πb2
2 (3.4)
The droplet fins in Figure 3.3 has a left circle which is twice the radius of the circle to the right r1 = r2/2.
Figure 3.3: Droplet Fins sketch
The two circles should have from the centers of each other, the diameter of the left circle between them D1 = 2r2. The area of the droplet was estimated by
AD = πr22 2 + πr21
2 + 2r1D1+ D1(r2− r1) (3.5) The NACA Fins were generated through NACA 4 digit airfoil generator [9]. It generates coordinates for an airfoil with a chord of 1, in this case mm. These points could be scaled up using MATLAB to any length. The scaled coordinates were then imported to NX 11 with a spline tool. This process required a couple of interpolations to match the area of the Control, A.
Figure 3.4: NACA 0010 sketch
The thinner fins such as NACA 0010 seen in Figure 3.4 had to be placed further apart lengthwise and some needed to be removed, but could then fit a fourth wing beside the other three in the spanwise- direction. The number sequence followed by NACA corresponds to the characteristics of a wing. The first digit defines the max camber(arching) in percentage of the chord, in this case the wing is symmetric so the number is set to 0. The second digit defines the max camber position along the chord, since there is no camber it is also set to 0. The last two digits represent
7
CHAPTER 3. METHOD
the max thickness of the wing in percentage of the chord length. A NACA 0010 wing with a chord of 10 mm would have the max thickness of 1 mm, a NACA 0030 wing with equal chord would have 3 mm max thickness. The chords of the profiles NACA 0030, NACA 0020, NACA 0010, 45 Deg NACA 0010 and NACA 0006 are respectively 4.0 mm, 4.8 mm 7.1 mm, 8.1 mm and 19.8 as, see Figure 4.1.
The Square- and Triangular Grid Fins in Figure 3.5 were sketched directly in NX 11, see Figure 4.3.
Figure 3.5: Square- and Triangular Grid Fins sketch The wall thickness is 0.26 mm and 0.15 mm, respectively.
The Clam Fins were placed in a 45 fin staggered formation and is constructed by a circle and four arches, as seen in Figure 3.6, sketched directly in NX 11. The diameter of the circle is 1.5 mm and the center line is 5.8 mm.
Figure 3.6: Clam Sketch
All blocks were ultimately measured with NX 11 internal tools to make sure the material volume is kept constant with the Control. The hand sketches proved to be useful when drawing up and constraining the geometries in NX 11.
After all geometries were completed in NX 11, they were exported to Ansys. The test blocks were meshed and simulated with Ansys CFX (CFD), and the outcome can be found in the Results Section 4. The simulation environment settings can be seen in Section 4.1.
8
CHAPTER 3. METHOD
3.2 Full Size Heat Sink
From the results of the test blocks two geometries stood out as better performing;
NACA and Grid Fins. Specifically the NACA 0006, NACA 0010 and the square grid fins in Figure 4.1 and 4.3 were chosen. The long plate fins from the Control heat sink were removed and replaced by the three new profiles. It was fairly easy to duplicate the patterns from the Test Blocks and apply them on larger scale. The square grid fins were moved into Ansys and extracting the fluid volume and meshing was not a problem. The simulation could compute and generate data. The NACA 0010 wings had encountered some issues, the irregular shape caused memory problems for Ansys when trying to mesh. After consulting Ansys support and trying on a computer with 256 gb ram instead of 32 gb, the same memory problem happened.
It was concluded that there are limitations to the specific geometry, number of geometries and Ansys. The solution came to be that the wings had to be more sparse in order to mesh. The number of fins on NACA 0010 was reduced from 1486 to 722 and the chord increased from 7.1 mm to 10 mm. With this alteration Ansys could mesh and then run the simulation. The NACA 0006 fins did not meet this problem since they had a chord of 19.8 mm and 312 fins to begin with. The simulation environment setting was the same as seen in Section 4.1. A depiction of the full size heat sink can be seen in Figure 4.13.
9
Chapter 4
Results and Discussion
In the results section, the simulation environment, results and simulations from NX 11 and Ansys are presented on both the Test Blocks and the Full Size heat sink. The results will be interpreted and discussed. The results should present if it is possible to increase heat dissipation of a heat sink by altering geometry and formation, taking the design freedom of AM into account.
4.1 Simulation Environment
The intended heat sink has restrictions to its outside parameter. The test blocks will be restricted to 5 mm in height of the fins and a condition that the pressure drop should not be increased, preferably lowered. Other than that it should not be altered from its current internal volume, unless some material could be removed to save mass.
In Figure 4.1 and 4.2 rendered images of the test blocks are shown. A section on top of the blocks in the images is hidden for visibility and in the tests they are closed channels with open ends. Air is blown into the open area where the test block names are written in white.
Figure 4.1: Test blocks rendered in NX 11
The mass, inlet air temperature and wet area temperature is constant throughout all test blocks. It is assumed for simplicity that components built by AM and Salt- bath dip brazing have the same heat transfer characteristics. This was assumed to isolate the variables and focus on the hypothesis: geometry and position can affect the convective heat transfer within an air cooled heat sink.
10
CHAPTER 4. RESULTS AND DISCUSSION
Figure 4.2: Test blocks rendered in NX 11
The 45 degree geometries were intended for printing the heat sink standing vertically instead of horizontally, so that more heat sinks can be printed next to each other at the same time. This resulted in a swept wing configuration.
Figure 4.3: Square and Triangular Grid Fins
A mesh with one directional channels could possibly give uphold to low turbulence and pressure drop. The wet areas of the grid fins is great in relation to the pin-fin geometries. see Table 4.2.
11
CHAPTER 4. RESULTS AND DISCUSSION
The simulation environment settings in Table 4.1 should be somehow similar to the ones applied to the heat sink in use.
Table 4.1: Simulation Environment Settings Inlet Air Velocity 7 [m/s]
Inlet Air Temperature 5 [℃]
Outlet Relative Pressure 0 [Pa]
Temperature of Wet Area 60 [℃]
Material Aluminium 6061-T651
In Figure 4.4 is a rendering of the simulated fluid volume depicted.
Figure 4.4: Control Fluid
The test blocks are used as molds to extract a fluid volume from each one. The blue area represents the cold air entering the system through the inlet. The red area represents the heated wall which is in contact with the air. The violet area represents the heated air exiting the system through the outlet.
12
CHAPTER 4. RESULTS AND DISCUSSION
4.2 Test Blocks
The results presented in Table 4.2 are derived from the Test Block simulations.
Pressure drop between the inlet and outlet is the pressure that is lost when pumping through a specific geometry, the pressure drop should be ideally reduced in order to reduce the required pump pressure. Average heat flux at outlet is the energy passing through the outlet during one second, the opening at the violet area in Figure 4.4.
The 5 ℃ air has some energy entering the system, the air temperature is increased by the heat sink and the heated air exiting the system is the average heat flux at outlet. The wet area is the surface area of the heat sink which is in contact with air.
The first column is an index used to compare the effective pressure drop per average heat flux with the Control. A lower number is related to better performance.
Table 4.2: Test Blocks Results Geometry
(Total number of fins) [PaW]
Pressure drop between inlet and outlet [Pa]
Average Heat Flux at Outlet[W]
Wet Area [m2]
Control Fins(9) 39.8 74.8 1.88 3153
NACA 0006 Fins(9) 22.2 71.7 3.23 3427
Circular Fins(45) 69.0 482.6 6.99 3064
45 Degree Circular Fins(42) 40.9 422.5 10.33 3313
Clam Fins(45) 29.0 246.0 8.47 4360
Droplet Fins(45) 46.3 283.4 6.12 3228
Elliptic Fins(45) 33.4 201.3 6.03 3190
NACA 0030 Fins(45) 26.3 149.5 5.69 3581
NACA 0020 Fins(45) 20.9 119.7 5.72 3919
NACA 0010 Fins(42) 14.7 110.2 7.49 4659
45 Deg NACA 0010 Fins(32) 15.2 127.8 8.42 4306
Square Grid Fins 14.5 134.7 9.32 7004
Triangular Grid Fins 16.3 224.9 13.74 9610
13
CHAPTER 4. RESULTS AND DISCUSSION
Figures 4.5 to 4.11 are images from the simulations on the test blocks described with the plane cut at the center. Each geometry has representations of air velocity, temperature and pressure.
Figure 4.5: Control Fins
The Control Fins in Figure 4.5 show a fairly well spread velocity profile in the yellow areas. The wake shown in blue behind the wings is a close to zero velocity region, a typical phenomenon that increases pressure drop. Some air escapes by accelerating in the red regions on the outer edges. The air temperature is mostly increased close to the fins. The length of the fins is causing the air to somewhat mix at the very back of the test block, this means that the heat transfer has a long ”take-off”
distance and the fins close to the opening is not being utilized to transfer heat. The fins are thin which means that the pressure region in the front is small.
14
CHAPTER 4. RESULTS AND DISCUSSION
Figure 4.6: NACA 0006- and Clam Grid Fins
The thin and long NACA 0006 fins in Figure 4.6 show a fairly well spread velocity profile. The fins had the same length problem as the Control Fins, with the air mixing at the very back. The wake has been removed, which agrees with the results in Table 4.2 that the pressure drop of NACA 0006 is lower than the Control and the pressure regions in the front of the fins is small. On the trailing edge of the fins in the back, it can be noted an anomaly appearing as a zero velocity region, from the fins being too close to the outlet, something that Ansys meshing software did not totally agree with. This should have a minimal impact on the results. It can be seen that the Clam Fins have two wakes on each fin. The heat transfer is good which can be seen in the red area of the free stream near the outlet, and the air starts to become hot early in the test block. The pressure area is spread out and is not focused in one point.
15
CHAPTER 4. RESULTS AND DISCUSSION
Figure 4.7: Circular- and 45 Degree Circular Fins
The Circular Fins in Figure 4.7 have large wakes behind them, a reason for the large pressure drop. The heated air is transferring well on to the next fin. The pressure points are large due to the width of the fins. The tilted Circular Fins have much smaller or non existent wakes behind them. The heated air transfers better than the circular fins onto the next one. The pressure regions are small compared to the circular fins but still are large.
Figure 4.8: NACA 0030- and NACA 0020 Fins
16
CHAPTER 4. RESULTS AND DISCUSSION
The NACA 0030 Fins in Figure 4.8 have almost no wake behind it. The flow follows the surface well. The heat is mixing in to the air gradually. The pressure region in the front are medium sized. The NACA 0020 Fins manage to keep the flow tighter to the surface. The heat transfer is even faster than its wider brother due to the larger wet area. The pressure region is smaller and less concentrated.
Figure 4.9: NACA 0010- and 45 Degree NACA 0010 Fins
The NACA 0010 Fins in Figure 4.9 are so thin that they fit four in a row. The airflow follows the airfoils well and heat transfer is good. The fins could have been placed further apart lengthwise to let the air flow better onto the next fin. The pressure regions are small in comparison to previous geometries. The 45 Deg NACA 0010 Fins have a similar velocity profile to the straight NACA 0010 Fins. The heat is mixing into the air the fastest of the NACA profiles. The pressure region is spread out and not so focused in one area.
17
CHAPTER 4. RESULTS AND DISCUSSION
Figure 4.10: Elliptic- and Droplet Fins
The Elliptic Fins in Figure 4.10 have small wakes behind them which is ultimately good to minimize pressure drop. The heat transfer to the air is similar to the Droplet Fins. The pressure regions are medium sized. The Droplet Fins have wakes behind them, starting after the left circle in the geometry. The heat transfer to the air is decent. The pressure region in the front is large.
Figure 4.11: Square- and Triangular Grid Fins
The Square Grid Fins in Figure 4.11 have a linear velocity profile. The heat transfer 18
CHAPTER 4. RESULTS AND DISCUSSION
is good due to the large surface area in contact with the air, but the air could be mixing better. pin-fin geometries could be placed in the channels. The pressure region is diffused and low which is promising. The Triangular Grid Fins were hard to show in a one plane cut, but it can be seen that the velocity profile act similar to the Square Grid Fins, one directional. The heat transfer is the best so far, and the wet area is the largest of all the test blocks. The pressure region is diffused but the amount of wet area gives rise to a fairly large pressure drop over the whole test block.
Figure 4.12 is a graph that shows heat flux with respect to the pressure drop for each geometry.
Figure 4.12: Test Blocks Plot
The diverted energy should be a high number and the pressure drop should be as low as possible. Thus, the upper left corner corresponds to a better performance.
19
CHAPTER 4. RESULTS AND DISCUSSION
4.3 Full Size Heat Sink
In Table 4.2 there are three geometries with an index number lower than 16; NACA 0010, 45 deg NACA 0010 and Square grid fins. Out of these, were the NACA 0010 and square grid fins chosen to be tested on the large heat sink. Their dissimilarity in geometry and printing orientation led to this decision. The NACA 0006 was also chosen to be tested on large scale as it was interesting to see that only the fin geometry and not the number of fins could affect the heat transfer. The Triangular grid fins performed the highest heat transfer at a cost of pressure drop. The tilted geometries provided an increase in heat transfer and decrease in pressure drop com- pared to their straight finned equivalent, this could be useful when deciding printing orientation.
Since the CAD-model is restricted by being classified, an illustration of the intended heat sink is depicted in Figure 4.13. The picture illustrates the whole heat sink where processing units can be mounted in between the two cooling walls (between the parallel lines of the H shape). A section is exposed where the pin-fins can be seen.
Figure 4.13: Full Size Heat Sink Illustration
20
CHAPTER 4. RESULTS AND DISCUSSION
In Table 4.3 there are four geometries with results derived from the full size heat sink simulations described in the method section. The NACA 0010 fins show a slight decrease in pressure drop and a significant increase in heat transfer. It is also reduced more than 3% in mass, only due to the change in fin geometry. It is certain that weight can be shaved off in other sections of the heat sink when manufacturing it by the means of AM. The NACA 0006 fins were, at a disadvantage, placed very sparse. They could have been fitted tighter and had a shorter chord. The Square Grid Fins shows effective cooling but at a cost of pressure drop, its wet area is 2,5 times larger than the Control. A grid type heat fin should be printed but a NACA can be both milled and printed. The problem for milling comes when the profiles are reduced in size and increased in numbers. Salt-bath dip brazing could also melt off the back end of the NACA wings during the dip brazing step.
Table 4.3: Full Size Heat Sink Results Geometry
(Total number of fins)
[PaW]
Pressure drop between inlet and outlet [Pa]
Average Heat Flux at Outlet [W]
Wet area [m2]
Relative mass
Control(503) 1.1 162.2 149.8 109548 1
Square Grid Fins 1.2 233.2 188.3 244058 0.979
NACA 0006(312) 0.9 125.2 139.8 106510 0.969
NACA 0010(722) 0.9 159.4 173.2 116641 0.967
Figure 4.14 is a graph similar to Figure 4.12 that shows heat flux with respect to the pressure drop for each geometry.
Figure 4.14: Full Size Heat Sink Plot
21
Chapter 5
Conclusions and Future Work
A study has been carried out on eight different geometries using additive manufac- turing. Heat sinks can be altered in FB by size and fin density to fit the required heat flux while the pressure drop is kept to a minimum. Thin NACA wing profiles and gridded channels have shown to be of high interest in heat sink applications.
The NACA 0010 geometry reduced the mass by more than 3% for the Full Size Heat Sink. The NACA 0010 Fins and Square Grid Fins resulted in a 63% and 64%
decrease in pressure drop per diverted energy compared to the Control representing salt-bath dip brazing. The improved performance was confirmed in the Full Size Heat Sink simulations. AM is therefore an effective method that can be used as an alternative manufacturing method. Some of the non-Control geometries could probably be manufactured by Salt-bath dip brazing but it is considered that due to the large number of fins in some other samples, they would be more expensive and difficult to manufacture. One should pay attention to the zero velocity regions behind pin-fins since the temperature results showed that the contribution to the mix of heated air with cold air was at a large cost in pressure drop. On the other hand, tightly packed pin-fins of geometrical shape which does not create zero ve- locity wake is recommended. Furthermore, considering the front of a pin-fin, this should cut the air as thin as possible, and therefore the circular fins had very large pressure regions on the leading edge.
Insights for future work have been gathered and relevant research identified to build a complete picture of the problem. The effects of a rough AM surface should be investigated and compared to a smooth milled surface with relation to convection.
A Full Size Heat Sink will be printed to perform physical tests, verify simulation results, and figure out how to test the quality of a component. A heat sink can have areas that reach higher temperatures than other areas. It will be investigated if the type and position of the fins can be adapted to maximize convection at those hot areas. It will be studied if in the same heat sink, other geometries with less heat transfer and requiring less pressure, can be placed on relatively cool areas.
Moreover AM can print out geometries independently of other fins and patterns, so a combination of geometries from this report and the amount of fins in a specific area could be the optimum solution for a specific system. In narrow channels and places where fins are obstructed, the fins could be reduced in size or even be rotated to fit the flow. Features that are not affected by fin geometry could be adapted to AM with the purpose of reducing weight. Finally alternative metals and alloys should be investigated.
22
Bibliography
[1] Alumbra. The brazing process. http://www.alumbra.se/thebrazingprocess.aspx, 2019.
[2] Sculpteo. Selective laser melting definition.
https://www.sculpteo.com/en/glossary/selective-laser-melting-definition/, 2019.
[3] Saab. https://saab.com/, 2019.
[4] Emmelie Simic. Additive manufacturing for field repair and maintenance of the assault rifle ak5c – a feasibility study. Master’s thesis, Uppsala University, Microsystems Technology, 2018.
[5] Maria Warholm and Johan Sönegård. Industrialization of additive manufac- turing - development of an additive manufacturing design guide for metal laser powder bed fusion. Master’s thesis, Chalmers tekniska högskola, Institutionen för produkt- och produktionsutveckling, Produktutveckling, 2016.
[6] Denpong Soodphakdee, Masud Behnia, and David Watabe Copeland. A com- parison of fin geometries for heat sinks in laminar forced convection: Parts i and ii. Int. J. Microcircuits Electron. Packag., 24:68–76, 2001.
[7] Feng Zhou and Ivan Catton. Numerical evaluation of flow and heat transfer in plate-pin fin heat sinks with various pin cross-sections. Numerical Heat Transfer Applications, Part A: Applications:107–128, 2011.
[8] Robert Smith. A thermal testing of a 3d printed super dense mesh heat sink against state-of the-art finned geometry.
http://www.qualifiedrapidproducts.com/wp-content/uploads/2015/09/Mesh- heatsink-experiment-01.pdf, 2015.
[9] NACA. 4 digit airfoil generator. http://airfoiltools.com/airfoil/naca4digit, 2019.
23
TRITA SCI-GRU 2019:225
www.kth.se
