Potential in additive manufacturing of a shaft

56  Download (0)

Full text





Potential in Additive

Manufacturing of a Shaft



Potential in Additive Manufacturing of a Shaft

Royal Institute of Technology

Degree Project in Mechanical Engineering, Second Cycle, 30 Credits

Department of Solid Mechanics

Author Daniel Damerji

Supervisor Examiner

Tobias Bergdal Prof. Per-Lennart Larsson

Stockholm, Sweden

June 5, 2019


Potential i Additiv Tillverkning av en Axel

Kungliga Tekniska H¨ ogskolan

Examensarbete inom Maskinteknik, Avancerad Niv˚ a, 30 HP

Institutionen f¨ or h˚ allfasthetsl¨ ara

F¨ orfattare Daniel Damerji

Handledare Examinator

Tobias Bergdal Prof. Per-Lennart Larsson

Stockholm, Sverige

Juni 5, 2019



Additive Manufacturing (AM), also known as 3D-printing, is the process of joining materials layer by layer from a 3D-model data and has several advantages over traditional manufacturing techniques. AM is destined to change the way products are designed and manufactured in the future. In recent years, the process has rapidly gained interest in all industry segments due to its ability to create customized and complex geometries for no added costs. This study focuses on a rather unexplored area of application of AM, namely of a vehicle component that traditionally is manufactured with conventional manufacturing methods. The purpose of this study is to investigate the potential of AM of a PTO-shaft used in Scania’s trucks. With the help of topology optimization and a developed cost estimation model, different design cases are compared to each other. Three areas are investigated: design, mechanical performance, and cost. The study found that a design with roughly 25 % weight reduction is realizable, and would today cost about 15 times the cost for series production using traditional manufacturing methods. This have clearly suggested that the PTO-shaft is not suitable for AM. However, by forecasting the cost into the future, the study found that printing the PTO-shaft are likely to be cost effective in terms of prototype production in the future, with up to about 200 e in cost savings per part.

Keywords: Additive manufacturing, Topology optimization, Truck, PTO-shaft, Cost estimation



Additiv tillverkning (AM eng. Additive Manufacturing), ¨aven k¨and som 3D-printning, ¨ar en tillverkningspro- cess d¨ar f¨orem˚al byggs upp skikt f¨or skikt fr˚an en 3D-modell och har flera f¨ordelar j¨amf¨ort med traditionella tillverkningsmetoder. AM kommer att f¨or¨andra s¨attet produkter designas och tillverkas i framtiden. Un- der senare ˚ar har processen snabbt ¨okat i intresse inom alla industrisegment p˚a grund av dess f¨orm˚aga att skapa skr¨addarsydda och komplexa geometrier utan extra kostnader. Denna studie fokuserar p˚a ett snarare outforskad till¨ampningsomr˚ade f¨or AM, n¨amligen p˚a en fordonskomponent som traditionellt tillverkas med konventionella tillverkningsmetoder. Syftet med denna studie ¨ar att unders¨oka potentialen av AM hos en kraftuttagsaxel (PTO), som anv¨ands i Scanias lastbilar. Med hj¨alp av topologioptimering och en estimerad kostnadsmodell som tagits fram, j¨amf¨ors olika design alternativ med varandra. Tre omr˚aden unders¨oks:

design, h˚allfasthet och kostnad. Studien visade att en realiserbar design med 25 % viktminskning kostar idag ungef¨ar 15 g˚anger kostnaden f¨or serieproduktion med traditionella tillverkningsmetoder. Genom att prognostisera kostnaden i framtiden fann emellertid studien att axeln sannolikt kommer att vara kostnadsef- fektiv att 3D printa f¨or prototyptillverkning. I ett s˚adant scenario kan det inneb¨ara upp till cirka 200 e i kostnadsbesparingar per axel.

Nyckelord: Additiv tillverkning, Topologioptimering, Lastbil, PTO-axel, Kostnadsestimering.



First and foremost, I would like to place on record my grateful appreciation and gratitude towards my supervi- sor Tobias Bergdal for the endless support throughout the project. I would also like to send my special thanks to Prof. Per-Lennart Larsson at KTH for the continues feedback over the last 5 months. Last but not least, I would like to thank everyone in NTEF and NTEE for their enjoyable company over the course of the project.

Daniel Damerji June 2019



1 Introduction 1

1.1 Aim, Delimitation and Purpose . . . 1

2 Additive Manufacturing 3 2.1 Types of AM . . . 3

2.1.1 Binder Jetting, BJ . . . 3

2.1.2 Direct Energy Deposition, DED . . . 3

2.1.3 Powder Bed Fusion, PBF . . . 3

2.2 Hybrid AM . . . 4

2.3 Workflow . . . 4

2.4 Powder metallurgy, production and quality . . . 5

2.5 Materials and machine suppliers . . . 6

2.6 Why is AM booming today? . . . 7

2.7 How is AM used in the industry today? . . . 7

2.7.1 Automotive applications . . . 7

2.8 Cost and cost structure . . . 8

2.8.1 Cost comparison - Traditional vs. AM . . . 8

2.8.2 Cost structures . . . 9

2.9 Drivers and challenges of AM, summary . . . 10

3 Power Take-Off Shaft, PTO 12 3.1 Original design and manufacturing process . . . 12

3.2 Geometrical and stiffness constraint . . . 12

3.3 Cost to manufacture . . . 12

3.4 FEM - model . . . 13

3.4.1 Boundary conditions . . . 13

3.4.2 Linear elastic material behaviour . . . 14

3.4.3 Mesh . . . 14

3.5 Topology optimization . . . 14

3.5.1 Inspire topology optimization settings . . . 16

4 AM systems and materials 17 4.1 Binder Jetting . . . 17

4.2 Direct Energy Deposition . . . 17

4.3 Powder Bed Fusion . . . 17

4.4 Compatible materials . . . 18

5 Methodology, Part Screening and Selection 19 5.1 Technical fit . . . 19

5.2 Economical fit . . . 20

5.3 Classification and part evaluation . . . 20

5.4 Design cases . . . 20

5.5 Cost estimation and cost model . . . 21

5.6 Added values . . . 23

5.7 Transparency complication . . . 23

5.8 Cost model data . . . 24

5.9 Hypothesis . . . 24


6 Result 25

6.1 Design Case 1 . . . 25

6.1.1 Mesh convergence . . . 25

6.2 Design Case 2 . . . 26

6.2.1 Geometry . . . 26

6.2.2 FEM . . . 27

6.3 Design Case 3 and 4 . . . 27

6.3.1 Geometry . . . 27

6.3.2 FEM . . . 28

6.4 Design Case 5 . . . 28

6.4.1 Geometry . . . 28

6.5 Cost estimation . . . 29

6.5.1 All possible costs . . . 29

6.5.2 Today’s costs . . . 30

6.5.3 Future Costs . . . 33

6.5.4 Annual volume production . . . 35

7 Discussion 37 7.1 Solid mechanics . . . 37

7.1.1 Internal structure - Lattice structure . . . 37

7.2 Technical and economical fit . . . 38

7.3 Source of error . . . 39

7.4 Future work . . . 40

8 Conclusion 41

Appendices 44

A Design sketch: PTO-shaft 44

B Design rules and AM process capabilities 44


1 Introduction

Reducing the weight of a component without jeopardizing its performance is both cost effective and envi- ronmentally beneficiary. Topology optimization is a method that optimizes an object’s material distribution within a given design space. Given a set of loads, boundary conditions and geometrical constraints, it op- timizes the weight and design while it enhances or keeps the performance of the component unchanged [1].

Because of the design complexity of the solutions obtained from topology optimization, it is often constrained due to the limitations of designs in traditional manufacturing techniques. However, the introduction of addi- tive manufacturing (AM) spans up new dimensions of design opportunities and feasibility, allowing the gap between topology optimization and industry application to be filled [1] [2].

During 2018 the global market size of the AM industry exceeded 7, 3 billion US dollars, a 21 % increase since the year before [3]. AM is believed to be one of the foundations of Industry 4.0 and with large investment from both the public and private sector, the market size is expected to reach 50 and 100 billion US dollars between the years 2029 to 2031 and 2031 to 2044 respectively [4] [5].

Scania CV AB, a worldwide leading provider of sustainable transport solutions has already some in-house experience of AM using polymers, and a number of metal prototypes have also been manufactured through external suppliers. However, the internal knowledge in Scania of how AM can be used on a large scale for end products and how they can be optimized are still limited. An effort to change that was made recently, when Scania together with others key Swedish industrial shareholders invested in AM through the Swedish start-up company Amexci, located in Karlskoga, Sweden. Amecxi started in 2017 with the objective to act as a resource for its shareholders to accelerate the adoption of AM in their productions [6].

The load capacity of a truck is limited by the gross weight of the vehicle, which is controlled by, among other things, legal requirements. One kilogram of less curb weight (gross weight or total weight minus cargo weight) can mean one kilogram more in cargo capacity. For customers, requirements for load capacity are often a decisive factor in the purchase of trucks. If, on the other hand, the space limits the load capacity, a reduced gross weight can instead lead to reduced fuel consumption. In both cases, the environment and the costumers are the beneficiaries. New heavy-duty vehicles are targeted to have 15 % and 30 % lower CO2

emissions in 2025 and 2030 respectively than in 2019. These are the proposed strategies from EU in order to contribute to the commitments to achieve fossil-free heavy transport by 2050 in accordance with the Paris Agreement [7]. For Scania, AM can be one of many ways to contribute to that promise and at the same time keep a leading competitive edge in transportation solutions around the world.

1.1 Aim, Delimitation and Purpose

One rather unexplored area of application for AM is traditionally conventional vehicle components such as a shaft. The purpose of this study is to investigate the potential of AM of a shaft within a gearbox. In particular a Power Take-Off (PTO) shaft will be investigated. The study is conducted at the department of development of electrical transmission at Scania (NTEF). Aside from understanding how the shaft can be designed for AM purposes and the potential weight reduction gain one can achieve by the use of AM, the department is also interested in gaining more knowledge about AM. The study also aim to help the department to gain a more general understanding on how far AM has come and what the future holds for this innovative technology.

In details the aim of this project can be summarized into these seven points.

• With respect to given design constraints find a design of a shaft within a gearbox suitable for additive manufacturing.

• Investigate what kind of internal structure the shaft should have with respect to strength and stiffness.

• How to model the internal structure in CAD.


• Investigate what kind of tolerances are achievable with the additive manufacturing.

• Compare the weight potential of such a shaft to a shaft made with traditional manufacturing techniques.

• Investigate which type of metallic materials are available for additive manufacturing and how well these are suitable for surface hardening.

• Investigate the cost of such a shaft.

The study will be limited to metal AM and the literature study will mostly cover metal AM. It will further be limited to only include static strength criteria because of two reasons; cyclic load sequences are not available and material parameters for both the current material and potential AM materials are limited which complicates the conduction of other performance analysis such as fatigue.


2 Additive Manufacturing

Additive manufacturing (AM) is an umbrella term that encompasses any process in which materials are joined together to make objects from 3D model data. In contrary to traditional manufacturing techniques, AM works by adding materials, usually layer by layer, rather than subtracting materials or forming materials [8].

2.1 Types of AM

AM can be divided into seven process categories: Binder Jetting (BJ); Directed Energy Deposition (DED);

Material Extrusion (ME); Material Jetting (MJ); Powder Bed Fusion (PBF); Sheet Lamination (SL) and Vat Photopolymerization (VP). These processes differ with regards to their fusion mechanism, material compatibility, speed, resolution quality and other features [9]. The most interesting processes with regards to metal AM for this study are Binder Jetting (BJ), Direct Energy Decomposition (DED) and Powder Bed Fusion (PBF). An introduction for these three metal AM processes will be given below. Description of the other AM processes, and the differences between these will be left out in this paper. The readers are however encouraged to read about the other AM processes, which also include polymer AM.

2.1.1 Binder Jetting, BJ

The Binder Jetting (BJ) process uses two materials, a building and a binder material. The printing process can be divided into two main steps. The building material is first spread on top of the building surface and then a layer of binder material is deposited where it is required in order to glue the build material together.

This is then repeated, layer by layer until the part is finished. The finished printed part is then sintered in an oven to increase the components mechanical properties from its green state body [10].

One of the biggest advantages of BJ is that support structures are not needed and that the parts are not required to be attached to the build plate. BJ also has a fast build rate, large build volume, high resolution, and is at the same time cost-effective compared to other metal AM methods. On the other hand, BJ parts have lower mechanical properties than other metal AM methods, due to their high porosity. The printed part have 20 − 30 % shrinkage after sintering. The material selection is also more limited compared to other processes [10].

2.1.2 Direct Energy Deposition, DED

Direct Energy Deposition (DED), is a process in which focused thermal energy (usually from a laser or an electron beam) is used to fuse materials as they are being deposited from a nozzle. This allows one not only to build new parts from scratch but also adding materials to existing parts in order to add new functions or repair damaged part [10].

Advantages of DED include high control of micro-structure, high productivity, flexibility and scalability.

Parts are fully dense an ready to use in production application. Complex geometries are easily achievable when 5-axis robotic arms are used which also do not require support structure. The main disadvantage of DED is the low surface quality and dimensional accuracy, which in most cases will require significant post-processing, usually in the form of machining [10].

2.1.3 Powder Bed Fusion, PBF

The Powder Bed Fusion (PBF) process is the most frequently used technique among industries for printing metal objects. The process can further be segmented into several similar techniques; Selective Laser Sinter- ing (SLS) (used for polymers), Direct Metal Laser Melting (DMLM), Electron Beam Melting (EBM) and Selective Laser Melting (SLM). PBF uses a laser or an electron beam as an energy source to fuse, sinter or melt (depending on the specific technique used) the applied layer of 3D printable metal powder on locations


specified by the desired geometry. The object is printed layer by layer and when one layer is completed a new layer of metal powder is applied and the process is repeated until the the full 3D object is finished [11] [10].

The main advantages of metal PBF is the excellent physical properties and the availability of many different materials. However, the cost, the speed, the limited build volume, and the requirement of post processing makes it difficult to be cost effective for parts that can be easily manufactured with traditional manufacturing methods [10].

2.2 Hybrid AM

Hybrid AM is a process in which the disadvantages of the AM methods mentioned above are compensated by integrating a subtractive method in the same machine system. Both AM by a deposition process such as DED and a subtractive process like milling requires a 5-axis robot machine, which makes it the most obvious combination for a hybrid AM process. However, there are also achievement made with the combi- nation of PBF process and milling. The material properties of a part manufactured by hybrid AM is shown to have improved fatigue resistance and ductility. The technology is still faced with challenges like varying mechanical properties, especially at the interface between the substrate and the additive material [12] [13] [14].

The hybrid AM process can work by alternating additive and subtractive processes in order to reduce cost of production, form complex internal structures and correct the dimensional and geometrical inaccuracies resulted from the additive process [12] [13] [14]. Figure 1 shows a schematic overview of a hybrid AM process.

Figure 1: Schematic overview of the hybrid AM process

2.3 Workflow

In general, all AM systems have a similar description of process chain. The general process chain is shown in figure 2, and can be divided into five steps:

Step 1: 3D CAD modelling

Step 2: Data conversion and transmission Step 3: Checking and preparing

Step 4: Building


Step 5: Post-processing

Figure 2: General overview of AM process chain

The finished 3D CAD model is for most AM systems converted into a stereolithography (STL) file format.

The STL format uses triangles to approximate the surface of the model. Before the fourth step, which is the actual construction of the part, the input files are checked and the AM system is prepared. The final step is the post-processing step, which involves removing the part from the machine, detaching any kind of supports, cleansing, painting and polishing. For metallic parts, post-processing also includes any necessary post-processing treatments to obtain the desired mechanical properties [15] [16].

2.4 Powder metallurgy, production and quality

Although the feed-stock for metal AM can be metal powders, wires and even sheets, the study will solely focus on powder technology because of the advancement already made and the great potential in the future compared to other types of feed-stocks [17].

There exists a variety of powder metallurgy techniques such as gas atomiztion, water atomization, plasma atomization, centrifugal atomization, plasma rotating electrode process, and others. The particles are charac- terized by many different features such as: particle size, shape, surface morphology, internal porosity, flowa- bility, particle size distribution, packing density etc. The quality of the powder is directly related but not limited to the choice of powder production technique. The quality of the powders are also directly translated to the properties of the final product and different AM techniques requires different powder characteristics to optimize the performance of the final produced product. Consequently it is important to understand the quality of the powders in order to determine for a given AM process, what is required to achieve a cost


competitive performance [17] [18].

There is a trade-off in the selection of powder characteristics and the main trade-off is cost vs. surface quality. Smaller particles tend to increase the surface quality and also the mechanical properties, but at a cost, both in terms of price and in terms of decreased flowability. While a finer particle size increases the mechanical properties, it decreases the flowability due to increasing forces of attraction (van der waart forces) between particles. The requirement of high flowability, which is the ability of the particles to flow evenly and uniformly, increases as the processing speed is increased. For the BJ, high flowability is essential to ensure good quality of the green body state of the component. This is one of the reasons why the costs of powders suitable for BJ is less expensive than for other metal AM techniques. However this also explain the reason why the mechanical properties are lower. This also holds true for DED compared to PBF, but not to the same magnitude. PBF, such as Selective Laser Melting, SLM and Electro Beam Melting, EBM requires a nominal particle size distribution of 10 − 45 µm and 45 − 106 µm respectively. The nominal particle size distribution for DED is 20 − 200 µm and for BJ even higher [17] [18].

2.5 Materials and machine suppliers

Today companies can offer a wide range of 3D printable materials with different properties from polymers, metals, resins, ceramics and others. In Sweden, there are five main metal powder producers for AM. These are; Carpenter Powder Products, Erasteel, H¨ogan¨as, Sandvik and Uddeholm. As of 2018 polymers are still the most used 3D printing materials even though it decreased from 88 % to 65 % from last year. Metal is the only material that has gained popularity, which increased to 32 % and obtained second place from resins, see figure 3.


Figure 3: Most used materials in AM

A wide range of metal powder can be used with different material properties, including metal powder ma- terials based of steel & stainless steel, titanium alloys, cobalt chrome alloys, aluminium alloys and Gold &

Silver. Many of them are also suitable for post-processing, such as hardening [5].

In terms of machine manufactures, there are two companies in Sweden; Acram and Digital Metal, where


Acram have a 10 % global market share as of year 2017. From an international perspective, the German company EOS is the leading AM solution provider with over 30 % global market take of metal powder bed fusion AM machines/systems as of year 2017 [5]. Both Amexci and Scania’s own in-house capacity of AM systems are from EOS.

2.6 Why is AM booming today?

The history of AM can be traced back to the Japanese researcher Hideo Kodama, who in year 1981 came up with the idea of 3D printing and developed methods for 3D printing plastic models. But it was not until 1993, when the first 3D printing machine was designed at the Massachusetts Institute of Technology (MIT), 12 years after the birth of 3D printing. Although the history of AM had its birth almost 40 years ago, it is not until recent years the market for AM has exploded. Experts believe the leading reason for the delayed market growth of AM is the strong investments from the industries enabled by the freedom to operate due to patent expiration. The advancement of Computer Aided Design (CAD), improved automation and access to more printable materials are also listed among the top reasons [4] [9] [19].

2.7 How is AM used in the industry today?

AM received its first application within the medical and dental sector. Dental products and medical implants exploited the advantage of customization that comes with AM. The aerospace industry followed thereafter, and has for some components reached full scale serial production. The manufacturing of complex and high demand parts at low volumes in the aerospace industry is preferentially aligned to AM. The aerospace industry is today the second largest industry segmented using AM with 18.9 %, just 1, 1 % behind the largest segment, industrial. Figure 4 shows the size of each industry segment currently served by AM. Overall, metal AM has in the past few years taken a huge step towards industrialization in other industry branches as well such as automotive, consumer products, energy and electronics [5]. Below some of the current AM applications within the automotive industry will be introduced.

Figure 4: Industries served by AM today

2.7.1 Automotive applications

Rapid Prototyping (RP) using AM technology was already adapted in the automotive industry as early as in the 90’ and is still the main focus today. However, the interest to use AM for final automotive parts produc- tion is increasing with the increasing requirements and demand of low weight high performing customized parts. As of 2018, the automotive sector corresponds to 16 % of AM market, making it the third largest


industry segment of AM [3] [5].

BMW has already for several years produced water pump wheels for their DTM (German Touring Car Masters, Ger. Deutsche Tourenwagen Masters) race car using AM. Koenigsegg has produced parts in the exhaust system in metal for their new model One:1. Mercedes Benz uses AM to print spare parts for trucks, which will reduce their inventory costs and start manufacturing on demand. Renualt has successfully 3D- printed a 4-cylinder engine prototype with 25 % less parts, which corresponds to 200 components with a total weight of 120 kg. Ford is currently using metal AM to produce pumps, valves and cooling vents. They also use 3D-printed molds for casting cylinder heads, gearboxes and frame parts [5].

2.8 Cost and cost structure

According to a Roland Berger study, the cost of AM would need to be decreased by a factor 10 in order to be cost effective across all mass produced components within all industries. However, there are still major benefits with AM compared to conventional manufacturing [20]. The list of benefits is dominated by the the consequences of new design possibilities. In this chapter, the cost of AM will be discussed.

2.8.1 Cost comparison - Traditional vs. AM

AM redefines cost versus complexity/customization scale by enabling both increased complexity and cus- tomization with little to no added cost. A qualitative example is shown in figure 5, where the break-even point establishes a threshold for when AM is more cost effective for certain parts with certain complexity degree [21] [6].

Figure 5: A qualitative example of the cost per part with increasing complexity or customization, AM versus traditional manufacturing

The issue with the vast majority of conventionally manufactured components, is that they have purposely been designed with simplicity to satisfy conventional manufacturing. Therefore, most conventional compo- nents today have to be redesigned in order to be cost effective and add value from the advantages of AM.

There are basically two possibilities. One possibility is to start manufacturing parts which are consolidated of many separate sub-parts directly into one component. This will result in cut costs and decrease in lead


time. The other possibility is to redesign the component for weight reductions and/or performance improve- ments using topology optimization or lattice structure. The two methods can also be used simultaneously. [5].

On the other hand, one of the biggest downsides of AM is that the cost per part does not decrease substan- tially with increasing volume, as opposed to conventional manufacturing methods. Instead the cost per part stays notably flat. If complexity and or customization stays limited and unchanged, conventional manufac- turing will be most cost effective as production volume increases. A qualitative example from is shown in figure 6, where the break-even point establishes a threshold for when conventional manufacturing is more cost effective for a certain size of production volume [6] [22].

Figure 6: A qualitative example of the cost per part with increasing volume, AM versus traditional manufacturing

Combining figure 5 and 6, one should hypothetically be able to plot a break-even point in a three dimensional space, where the cost, complexity/customization and production volume each spans up a dimension. This will allow one to objectively decide which manufacturing technique is most cost effective. However, as the demand of AM increases, the cost will decrease and the number of components which would be more cost effective using AM will increase in future. More freedom in design, topology optimization and advancement in 3D-printing technique and material knowledge will shift the break-even point in favor of AM [9].

2.8.2 Cost structures

It is important to note that the above cost figures are only a qualitative representation and should be used as a guidance and not as a distinct rule. In fact, AM has a more complex cost structure compared to tradi- tional manufacturing and requires new cost calculations and cost models. According to [23] the costs can be segmented into two groups, ”well-structured” and ”ill-structured” costs. The well-structured cost category include cost of labor, material, and machine costs, whereas the ill-structured costs takes into account costs such as build failure, inventory, transportation, machine setup and other costs which are hidden in the supply chain [23] [24] [25].

However, there are still no accurate cost calculation models of AM because, in the literature, there tends to be more focus on the well-structured costs than on the ill-structured costs. The problem with this approach is that some of the most significant benefits and cost savings in AM are hidden in the ill-structured costs.


This becomes more clearly when you consider the benefits of AM in the context of the key concept of lean manufacturing ”muda” and the identification of waste. AM is most certain to have a positive impact on most of the seven waste categories (Defects, Inventory, Motion, Over Processing, Overproduction, Transportation, Waiting) significantly. AM enables manufacturing on demand, which in turn means that AM may reduce the need of large inventory. These inventories are tied up capital for products that are unused. Inventories also occupy physical spaces, buildings, and land, which requires rent, utility cost, insurance, and taxes.

By producing products on demand, the need of inventories is reduced and the associated costs eliminated.

Moreover, it is common that different parts of a product are manufactured in different locations and later transported to the same location where the parts are assembled into one product, see figure 7 from [24].

As discussed above, AM allows manufacturing of multiple components simultaneously in the same assembly.

This may mean that AM can replace some of the manufacturing flow steps by integrating more parts into one process, build in one location. This in turn mean cost savings associated with a production facility as well as less transportation costs. In brief AM will impact the supply chain by reducing the number of links in the supply chain and bring the production closer to customers by decentralizing production [24] [25].

Figure 7: Example of traditional manufacturing flow. Parts within the same product are manufactured on different location before they are shipped to a location where they are assembled

2.9 Drivers and challenges of AM, summary

The overall drivers and challenges for the full potential use of industrial AM across all industries can be summarized in the two lists below [5] [10]. The list is not complete but includes the most important issues. A bold text specify the driver or the challenge that progress or aggravate the adoption of AM in the automotive industry. Some of the mentions goes hand in hand but are listed separately for elucidation. The drivers are:

• Available materials - more materials suitable for AM with improved properties and manufactured product quality.

• Customization - components can be customized with little to no added costs.

• Eco-efficiency - less material usage, lightweight possibilities, especially important for expensive mate- rials.

• Faster product development - prototypes can be manufactured faster for a lower cost. Mainly because of the fewer tool investment.

• Increased design freedom - no geometry restriction, possibility to integrate numerous parts that formerly required assembling.

• Improved performance - internal channels and structures to improve performance.

• Low volumes production - AM already suitable and cost-effective for low volume production or with high design complexity.


• Reduction in lead time - redesigning for AM to consolidate multi-component parts reduces lead times.

• Spare part production - AM enables manufacturing on demand which reduces the need of large inventory and which in turn reduces the associated costs.

• Supply chain efficiency - cuts costs through the whole supply chain by decentralizing production.

and the challenges are:

• Availability and reliable material data - poor understanding of material properties. More Nonde- structive Testing (NDT) methods has to be developed.

• IPR concerns - new types of Intellectual Property Rights (IPR) questions are raised because of the use of AM.

• Lack of knowledge - engineers in the industry lack general knowledge about AM and expertise in the field are limited.

• Lack of standards and qualification procedures - slows down the industrialization. Its in need to en- courage broader and faster adoption.

• Limited choice of materials - although the number of available materials is increasing, they are still limited and the process to develop new materials are expensive and tedious.

• Low throughput - machines must become faster to meet the throughput needs.

• Manufacturing of large parts - Current AM systems have a limited build volume despite the the need to produce larger components.

• Build rate and robust process - AM machines must be faster to increase throughput of parts and they also need to increase their reproducibility rate to guarantee same quality in series production.

• Material costs - material costs are too high to be economically beneficially.


3 Power Take-Off Shaft, PTO

A Power Take-Off (PTO) shaft is a shaft used to transfer power from a power source (usually an engine) to auxiliary components. In most cases, the PTO-shaft is connected to a hydraulic pump, which in turn has a variety of applications. Scania provides two types of PTO-units, clutch dependence and clutch independence.

A clutch dependent shaft can only be used when the truck is running, while a clutch independent can be used anytime. This shaft is used in a clutch independent operation.

3.1 Original design and manufacturing process

The PTO-shaft is first shaped by a warm forging process. The shaft has a total length of 356, 5 mm and a maximum outer diameter by the gear location of 80 mm. The weight of the shaft is 4, 35 kg. The shaft is then subjected to roughing cuts in order to achieve a near net shape close to the desired form. At the regions where there are critical tolerances, margins are left for the later finishing cuts. From here, the central hole and the two square holes at each end are created. The shaft is then subjected to case hardening to satisfy the hardness requirement of 59 - 63 HRC. Lastly, the shaft undergoes finishing cuts at the bearing positions, gear position, the two square holes, and other critical tolerances in order to achieve the required surface finish. Thereafter the two tapered roller bearings and the gear are pressed onto the shaft in the assembly process. A full design sketch of the PTO-shaft with detailed design restriction and surface quality requirement is given in Appendix A.

3.2 Geometrical and stiffness constraint

When designing a shaft, there are two objectives to keep in mind in the analysis. The stress levels have to be sufficiently low and the deflections have to be controlled to ensure that the gear mounted on the shaft can function as expected.

There is a maximum displacement constraint of 20 µm between two points placed on a horizontal line at the centre line that goes through the gear. If the difference in displacement between those two points exceeds 20 µm, the gear will be exposed to higher contact stresses which in turn will cause a lower lifetime. Furthermore, the unavailability of material property data such as the yield strength of the used steel in the PTO-shaft, makes it difficult to dimension the PTO-shaft to ensure the stress levels are within the elastic region. To overcome this problem, it is assumed that the current design of the shaft is designed to tolerate the maximum allowed torque with a safety factor included. Therefore, the maximum stress level from the FEM simulation of the original design will be used as an upper stress level limit. Nonetheless it is still important to have a reference to ensure that the stress levels are not way off. For example, a case hardened steel 16MnCr5 from Livallco St˚al AB have a minimum yield strength of 440 M P a. The value is based on a 50 mm bar in diameter, much like the PTO-shaft [26].

There are also a couple of critical geometrical constraints which needs to be met in order for the shaft to function as expected. The finest surface quality is placed at the bearing positions. All other critical and geometrical constraints are present in the full design sketch of the PTO-shaft, see Appendix A.

3.3 Cost to manufacture

This particular PTO-shaft is intended to be purchased by an external supplier. There already exists a price proposal for prototype production of 7400 SEK per unit. The annual production volume is estimated to be 1500 units. This is still considered to be low volume production which tend to be unattractive in the market and is therefore usually overpriced by the suppliers. There is no official price proposal for series production, but the cost is estimated to go down to 1000 SEK per unit once series production is reached.

Cost segmentation for this article is not available.


3.4 FEM - model

The shaft together with the associated driven gear are modeled in Altair solidThinking - Inspire [27]. The gear forces are modeled as surface forces applied on one of the teeth, where each force corresponds to the axial, tangential and radial force respectively, here termed as: Fa, Ftand Fr, see figure 8 and 9. The presence of axial gear forces is because the gear is a helical gear. The forces are calculated as:

Ft= 11, 18 · M52 Fa= 4, 49 · M52 Fr= 4, 28 · M52, (1) where M52 is the torque from the driving gear, and 52 corresponds to the number of teeth. The driven gear is dimensioned to withstand a maximum torque of 2000 N m with 41 teeth. The corresponding maximum torque on the driving gear than thus be calculated by:

M52= M41·Z52


(2) where Z52and Z41is the number of gear teeth respectively.

Figure 8: Shaft and gear, isomeric view Figure 9: Shaft and gear, side view

3.4.1 Boundary conditions

Two boundary condition are applied on the shaft, one at each end where the tapered roller bearings are located. The supports are both located at the respective bearings load of contact point with connectors connecting the point to the inner and outer surface of the shaft. The dimensions were collected from the suppliers catalog, SKF. For the left bearing, where the maximum torque will be transmitted to the auxiliary shaft, a fix constrained is applied, see figure 10. On the other side, the bearing is only locked in all translation degrees except in axial direction so that the shaft is not fully stiff. Free rotation in axial direction is also activated to allow rotation, see figure 11.


Figure 10: B.C. left bearing Figure 11: B.C. right bearing

3.4.2 Linear elastic material behaviour

Linear elastic behaviour is assumed. The governing equations for the linear elastic behaviour of the material are described by:

∇S + FV = ~0 , S = C~el , ~el = ~ − ~inel , ~ = 1 2

∇~u + (∇~u)T

(3) where S denotes the stress tensor, FV the force matrix over volume, C the stiffness matrix, and ~, ~el, ~inel

the total, elastic, and inelastic strain vectors respectively.

3.4.3 Mesh

Automated meshing was used on the model. Altair Inspire runs a combination of HyperMesh and Simlab in the background for meshing. For this problem the minimum element size is 0.5mm and the average is 2, 5mm. Figure 12 shows the mesh of the PTO-shaft.

Figure 12: Mesh

3.5 Topology optimization

Topology optimization is a form of a structural optimization problem. A structural optimization problem consists of an objective function, design variables and state variables. The objective function, represents


what could be minimized or maximized. The design variables describes the design of the structure and the state variables represents the structural response (such as stress, strain or displacement) and depends on the design variables.

The goal of topology optimization is to maximize the stiffness of a structure by finding the most optimal distribution of material within a design space, subjected to conditions such as applied loads, boundary con- ditions, design restrictions, and other limitations. The design variables in topology optimization are element densities ρe. Each finite element within the design space is parametrized with the density design variable, ρe, which is assigned the value 1 where material is required and 0 where material is removed.

However, there are no available algorithms that can handle large amount of discrete variables, so the problem is converted into a continuous problem. The most popular mathematical method used to solve topology optimization by considering continuous variables is the SIMP (Solid Isotropic Material with Penalization) method. By introducing a continuous relative density distribution where for each element, the assigned relative density can vary between a minimum value ρminand 1, a discrete problem is avoided. Here, ρmin is the minimum allowed relative density value for empty elements that are greater than zero and ensures the stability of the finite element analysis. The introduction of a continuous relative density distribution results that the material Young modulus at each element also vary continuously. This is modeled as a power law that reads:

E(ρe) = ρpeE0 (4)

were p is the penalty factor and E0 is the inital Young modulus. The penalty factor p is used to penalitize intermediate values to an upper or lower limit. It reduces the contribution of element with intermediate values and steers the solution to elements that are either solid (ρe = 1) or void (ρe = ρmin). Numerical experiments has indicated that a penalty factor of 3 is the most optimal choice. The SIMP method is vulnerable to the checkboarder pattern which in most commercial software is solved with a sensitivity filter.

The checkboarder pattern and sensitivity filter will not be further discussed in this study. The topology optimization formulation problem can finally be formulated as:


ρ C(ρ) = UTKU =





subject to : V (ρ) V0 = Vf

: KU = F

: 0 < ρmin≤ ρe≤ 1

where C is the compliance, U is the displacement vector, K is the stiffness matrix, F is the force vector, V is the volume of the optimized design shape, V0 is the volume of the original design space, and Vf is the desired volume fraction. Instead of minimizing the compliance as the objective function, one can minimize the volume V (ρ) subjected to stress constraints, as:


ρ V (ρ) =




e)pV0,e subject to : σs


≤ SF

: KU = F

: 0 < ρmin≤ ρe≤ 1

where V (ρ) is the volume, V0,e is the initial element volume, σs is the yield stress, σe is the stress level at element e, and SF is the desired safety factor.


3.5.1 Inspire topology optimization settings

The topology optimization settings in Inspire follows initially the same procedure as the FEM - model. The following step is to divide the part to a design- and non-design space. The design space is the part in which the program is allowed to remove material until an optimized shape is reached. The design space for the PTO-shaft is marked brown and the non-design space with grey in figure 13 and 14.

Figure 13: Design space marked in brown Figure 14: Design space marked in brown The total weight of the design space is 2, 2 kg which is roughly 50 % of the total weight of the shaft. Cyclic symmetric shape control with 41 sectors is used on the design space to ensure that the final shape is 41 times cyclic symmetric. The number 41 corresponds to the number of teeth of the driven gear. Maximum Stiffness is selected as the objective and 50 % of Total Design Space Volume is selected as the Mass Target.

A thickness constraint of 5 mm is used. The gravity option is neglected since the contribution from the gravitation can in this case be neglected. The optimized shape from the optimization run is only used as an insight. The PTO-shaft is redesigned based on the result and the final design is checked if the stress and deformation levels are sufficiently low.


4 AM systems and materials

In this chapter, description of some available AM systems for the three different processes and compatible materials will be presented. The majority of the data are based from Senvol, an information service company that provides data to help companies implement AM [28].

The most fundamental criteria for a suitable AM system is the size of the build volume. The minimum volume restriction is simply the size of the PTO-shaft (356,5 mm x 80 mm x 80 mm). A search on Senvol with a dimensional restriction as above gives 3, 50, and 33 search hits for BJ, DED, and PBF machines respectively. This gives a ratio of about 1:17:11 and thus gives an insight to which AM process is the most suitable for a component of a size such as the PTO-shaft.

For each AM process an overview of the available systems and materials will be listed. A search on available metal materials for BJ, DED and PBF gives around 50, 200, and 900 search hits respectively. Although metal powders also include titanium, nickel based alloys, cobalt based alloys, and others, it is only steel and iron based alloys that are of interest for this project.

4.1 Binder Jetting

Binder jetting companies of interest that offers system that prints in metal are Desktop Metal, Digital Metal, ExOne, and HP. However, ExOne is the only company that offers metal BJ machines that satisfy the dimension of the PTO-shaft. In table 1 some of the available systems by ExOne and their technical specification are listed.

Company name ExOne Exone

Machine name M-Print M-Flex

Price, €K 100 - 250 250 - 500

Build volume, mm 800 x 500 x 400 400 x 250 x 250 Min layer thickness 0,15 mm 0,15 mm

Speed 30-60 (s/layer) 30-60 (s/layer)

Table 1: Technical specifications of ExOne machines, BJ system

4.2 Direct Energy Deposition

DED companies of interest are BeAM, DMG MORI, GEFERTEC, InssTek, Optomec, Skiaky Inc, and Trumpf. The number of suitable machines are far more than for BJ (around 50) and only a couple of them, each from a different company will be presented here. Table 2 shows the technical specification of three different machines.

Company name BeAM InssTek Optomec

Machine name Modulo 400, Magic 800 MX-1000 LENS CS 800

Price, €K >1000 500-999 500-999

Build volume, mm min 650 x 400 x 400 1000 x 800 x 650 800 x 600 x 600 Deposition width, mm min 0,8-1,2 Not reported 0,3-5mm

Speed max 90 − 130cm3/h Not reported max 0,5kg/h

Table 2: Technical specifications of some DED machines from different companies

4.3 Powder Bed Fusion

PBF companies of interest are 3D Systems, Aconity3D, Arcam EBM, Conceptlaster, DMG MORI, EOS, Farsoon, SLM Solutions, and Trumpf. PBF is the most widely used AM process for metal application with


about 33 different suitable systems available on Senvol [28]. Table 3 shows three of these systems.

Company name EOS EOS SLM Solution

Machine name EOS M 400 EOS M400-4 SLM 500

Price, €K 1250 >1500 700

Build volume, mm 400 x 400 x 400 400 x 400 x 400 400 x 280 x 365 Min layer thickness 40-50 µ m 40-50 µ m 20-70 µ m Speed max 20 cm3/h max 100 cm3/h max 170 cm3/h

Laser power, Watt 1 x 1000 4 x 400 max 4 x 700

Table 3: Technical specifications of PBF systems

4.4 Compatible materials

As stated before there are around 50, 200, 900 search results on metal powder materials for BJ, DED, and PBF fusion process respectively, which gives a 1 : 4 : 18 ratio. As discussed in previous chapters both the price and mechanical properties of the same metal powder material can vary with the used AM process, but also with the source of supplier. The technical data of the material in table 4 are mostly collected directly from the supplier. They should not be assumed to be true because of the possible variation in mechanical properties due to a range of parameters, such as heat treatment, particle size distribution, shape and others.

Name Stainless Steel PH1 Stainless Steel 316L 17-4PH Maraging Steel MS1

Yield strength, MPa 1200 500 1000 2000

Hardness, HRC 40 - 40 50-57

Density, g/cm3 7,7 7,9 7,78 8

Price,€/kg 70 120 90 110

Supplier/Source EOS, ExOne EOS, ExOne, Optomec EOS,3D Systems EOS

Other noteworthy (without price) 440C Stainless Steel 4340 Alloy 610 Tool Steel

Yield strength, MPa 1896 1500 2758

Hardness, HRC 60 50-60 60

Density, g/cm3 7,6 7,84 7,83

Supplier/Source Carpenter Carpenter Carpenter

Table 4: Compatible materials and their technical information

Most of the materials that have been found to be suitable for this particular shaft have according to the suppliers own data sheet been subjected to an induction or a precipitation hardening heat treatment to improve mechanical performance. It has not been explicitly found that other heat-treatments (such as case hardening) are not suitable for metal powder materials. The applicability to conduct case hardening or any other heat treatment is influenced by the carbon content in its base state, before any heat treatment have been applied. Therefore case hardening can neither be ruled out or accepted as a possible heat treatment for metal powder materials.

The achievable surface quality and tolerances also varies with both supplier and process technology. For this reason they will not be explicitly written here but an overview of design rules and process capabilities will be given in the appendix. For more general information on design rules and process capabilities for different AM process, see the summary overview in Appendix B.


5 Methodology, Part Screening and Selection

In this section a methodology is developed to help one in a systematic way screen and classify parts on how well a part is suitable for AM. Since AM becomes more cost effective the more complex the part is, one fast pre-screening method is to calculate the geometric complexity ratio of the part, which is defined as:

AM complexity ratio = Surf ace area(cm2)

V olume(cm3) (5)

where the volume is simply the mass divided by the density. Keep in mind that although the ratio above has the dimension m−1, it is still treated as a unitless measure. A rule of thumb is that all parts with a complexity ratio over 50 is more suitable for AM than traditional manufacturing [6]. This is however not a complete methodology, mostly because it does not take into account the possibility to redesign the part and the different added values that comes with AM. A more complex and much longer screening and selection process is needed, which include looking into the suitability, cost comparison, re-designing possibilities, functional integration, part integration, develop a business case and prototype production. Figure 15 summarize the part and selection screening methodology for AM.

Figure 15: Process of screening and selecting parts for AM

In this study, only Step 1 (Technical and Economic fit) from figure 15 together with the possibility to redesign, to integrate a function, and part integration will be examined. Different design cases will be developed and analysed with regards to geometry and integration functionality. The results from the different design cases will be presented and compared in order to see what added value AM gives as a result of the changes. The different design cases will be presented later on in this chapter.

5.1 Technical fit

When evaluating the technical fit, there are three components to look into: size/building chamber, material restriction and quality requirements. Question to be answered are such as:

• Does the part fit into current AM systems?

• Are their suitable materials available for AM?

• Does the part fulfill the quality requirements, such as tolerances, surface finish, and other material properties?

• Does the part fulfill AM design rules, such as minimal wall thickness?


5.2 Economical fit

When evaluating the economical fit, there are two components to look into: Costs per part and the potential value add. Cost estimation model of a part printed in metal will be given below. Questions to be answered are such as:

• What are the cost per part and what is the difference in costs compared to conventional manufacturing?

• What are the potential savings in the supply chain costs?

• Is there any possible added value of a redesigned part, which ultimately justify the higher manufacturing costs?

5.3 Classification and part evaluation

When evaluating the technical fit one is essentially asking ”Can we do it?” and when evaluating the econom- ical fit one is essentially asking ”Should we do it?”. However in both cases it is important to investigate from the costumers point of view, in order to understand which advantages of AM bring the strongest added value to the costumer. By rating the technical and economical fit from low to high, one can map each potential component or a set of integrated components in a classification and evaluation matrix, see figure 16. Doing so will help one prioritize the switch of most suitable parts from traditional manufacturing to AM with regards to both the economical and added value benefits.

Figure 16: Classification and part evaluation matrix

5.4 Design cases

Five different design cases will be analysed, where each case corresponds to one change in comparison to the previous design case. The design cases of the PTO-shaft that will be evaluated according to the steps


indicated in this chapter are:

1. Current model, no changes.

2. Topology optimization on current model.

3. Topology optimization on current model with added changes, e.g. holes.

4. Topology optimization on current model with added changes and optimized feature, e.g. holes and spline shaft.

5. Topology optimization on current model with added changes, optimized feature, and integrated with a second shaft, e.g. holes, spline shaft, and oil pump shaft.

From here on, the different design cases will be referred as case 1, case 2, case 3, and so on, in the same order they are presented. For each design case, these questions will be answered:

• What is achievable today with regards to costs and quality requirements.

• What needs to be changed or achieved in order for the product to meet the quality requirement and/or be cost effective in the future?

• How realistic is that within the next 10 to 20 years?

5.5 Cost estimation and cost model

The cost of printing a part in metal is volume dependent and grows almost linearly with the job volume. This is different from printing in polymer, where the cost of printing is dependent on the height of the part [10].

The developed cost model is mostly based from [29] but also generally from different articles. In this cost estimation model, it is considered that the cost of printing in AM can be subdivided into five cost elements expressed on a per build basis, C, as:

C = M + P + L + O + P P (6)

where, M is the cost due to Machine purchase, P is the Powder material cost, L is the Labor cost, O is the machine Operation cost, and P P is the Post-Processing costs. Going forward, cost segmentation of each of the different cost elements will be explained and the associated assumptions discussed.

The cost associated with the machine purchase price for one build is a function of machine cost per year Cm times the the time for the build Tb, calculated as:

M = Cm· Tb (7)

where Cm can be calculated as:

Cm= Mpp

Urate· 24 · 365 · Yd

(8) Mpp is the machine purchase price, 24 x 365 represents the number of hours in a year, Urateis the machine utilization rate, and Yd is the depreciation time of the machine in years. It will be assumed that the depre- ciation time Yd = 6 years. The utilisation can differ drastically. Assuming maximum utilization, meaning the machine is active 24 hours a day throughout the year gives an utilization rate of 8760 hours on 365 days. A more conservative approach, assuming 8 hours work day, 5 times a week, gives an utilization rate of approximately 2000 hours. However, the machines can most of the time run under little to no supervision, meaning a higher utilization rate is more realistic. A study from [30] compares the cost distribution of AM of metal parts by varying factors using a machine utilization rate of 4500, 6000, and 7800 hours/year. This indicate that a utilization rate close to the maximum is in fact possible. Therefore, for the purpose of this


study an utilization rate of 6000 hours/year (68 %) will be used.

It is also shown in [30] and [31] that the envelope utilization, meaning the utilization of the total amount of build capacity, have an impact on the cost. A lower envelope utilization causes a higher cost per part.This is not accounted for in the cost model presented above. In this study however it will be assumed that the effect on cost due to envelope utilization can be neglected.

The build time Tb is the sum of of three separate times: the deposition time, transition time between layers, and delay time [29]. However, for the purpose of this study, the later two will be neglected and the total build time is thus the volume multiplied by the build rate speed.

Finally, combining equation 7 and 8 cost per build related to machine purchase price can be expressed as:

M = Mpp

Urate· 24 · 365 · Yd

· Tb (9)

Powder material cost is ideally the volume V , of the part, multiplied by the price of the material per unit mass, Cp, times the density ρ, of the powder material. However to accurately capture the true cost of material consumed, one has to take into account the extra material volume needed for support structures and the fact that more powder than what is build is needed. This will be modelled as a factor, here denoted as Ks and given the value 1, 2. Thus the true cost of material is 20 % higher than the ideal case and can be expressed as:

P = Ks· V · Cp· ρ (10)

Labor cost include the cost for workers to set up the build, remove fabricated parts, clean the parts, clean the machine, supervision, get the part ready for post-processing and get the machine ready for the next round.

This will be modeled as the labor rate cost Cl in €/h multiplied by workers spent time associated to the construction of the part, Tl. The labor cost can therefore be expressed as:

L = Cl· Tl (11)

The cost associated to labor will be based on an entry-level salary for an engineer with masters degree in Sweden. The recommended entry wage according to Sveriges ingenj¨orer is 34500 SEK per month [32]. Adding taxes due to social security, employers are charged with approximately 45000 SEK per month for hiring an entry-level engineering, or 540000 SEK per year. A year includes roughly 2000 hours of normal work-hours, which means that the labor rate Clis 270 SEK/h. Since the AM process is almost fully automatized, workers do not need to spend a lot of hours concerning the printed part. Therefore, the labor time Tlis assumed to be 10 % of the build time Tbfor all build time greater than 10 hours, and 1 hour otherwise. Equation 11 can ultimately be expressed as:

L = Cl· Tl, Tl=

(0.1 · Td, if Tb > 10

1, otherwise (12)

The operation cost includes costs associated with maintenance, rent, utility, administration, and others. Here it will be modeled as the operation cost factor, Cotimes the build time Tb, expressed as:

O = Co· Tb (13)

Estimating the operation cost factor is difficult. For simplicity the operation cost factor is chosen to be equal to 10 % of the cost associated with the machine purchase cost. This means that equation 13 can be expressed as:

O = 1

10· Mpp

Urate· 24 · 365 · Yd

· Tb (14)


Lastly, most metal AM processes is followed by a post-processing step such as a heat treatment for stress relief and improved mechanical properties, and possible machining depending on the surface quality requirement.

The post-processing cost is a function of both machine purchase price and labor cost. When the part is build with a faster build speed, the surface quality is usually poorer, meaning the need of post-processing is higher, and therefore more expensive. Thus cost for post-processing increases with decreasing build time, Tb. This will be modeled as:

P P = Cpp· Cm+ L, (15)

Cpp =

(1 if Tb > 10 1, 5, otherwise , Tl=

(0.1 · Td, if Tb> 10 1, otherwise

5.6 Added values

In the many cases today, where the cost of manufacturing using AM is more expensive, the switch from TM to AM can be justified with added values. The added value can be related to the product life cycle such as:

faster product development, simplified manufacturing, production on-demand, decentralized production, and reduced downtime. Or the value add can be related to the product itself, such as: customization, improved appearance, functional integration, weight reduction, and performance improvement.

The added values can be translated into cost-savings in order to compare if in the many cases the higher cost of manufacturing is justified. The possible added values for this PTO-shaft, depending on the design case are:

weight reduction, performance improvement (splines), and part integration.The corresponding cost-savings for each type of added value will be discussed below.

Weight reduction will result in cost-savings in terms of less material and fuel consumption. At Scania, the guideline for one kilogram reduced weight corresponds to 300 SEK in cost-savings during the products whole life-time.

Spline shafts have a higher maximum torque capacity than square hole shafts. Hence, the added value can be written as a function of added torque capacity. At Scania, engineers uses a set of references from different available spline shafts when designing new splines. The maximum torque capacity for these splines range from 11000 − 45000 N m, which is about 5, 5 − 22, 5 times as high torque capacity as the current PTO-shaft.

For this particular shaft, a higher torque capacity is not essential since the PTO-shaft will in most time not run on maximum torque capacity of 2000 N m. Nevertheless, the performance of the shaft is still improved, and the added value for this particular PTO-shaft will be modeled the same way as a weight reduction of one kilogram. Thus, the cost-savings due to the presence of spline shaft is 300 SEK.

The cost-savings due to part integration are cost related to cost of producing the second integrated part and cost related to the assembly of the two components. The second part is difficult to estimate, and for the sake of this study it will be neglected. The cost-savings will only account for the production of the integrated part. The chosen oil pump shaft will add an extra 35 cm to the total length, meaning the new total length is about 400 mm. The price for the oil-pump shaft is 100 SEK in series and around 900 SEK for prototype.

The added value will only be compared to series production. Thus, the cost-savings will be estimated to be 100 SEK.

5.7 Transparency complication

The transparency complication refers to the issue that machine and powder material suppliers are not fully transparent with the technical information on their product, including prices. Many powder material com- panies such as H¨ogan¨as and Sandvik disclose all relevant information of their metal powders to the public,


including price rate and material properties. Furthermore, companies that offers a full AM solution (machines and materials), do not explicitly mention that their systems are also compatible with metal powders supplied by a third party, nor do they oppose it. However, going forward, it will be assumed in this study that any machine is compatible with any metal powder material from any supplier. In the case of metal powders that lack information on relevant material properties, they will not be considered as an alternative in the analysis since no accurate conclusions can be drawn.

5.8 Cost model data

Because of the difficulty to validate the findings of the price for a given material and for a given AM system, the cost estimation will not be based on a specific material and a specific machine. Instead, an estimated price of material and machine will be used based on the findings above and from other literature. Since Amexci use PBF technology and their in-house capacity consists of systems from EOS GmbH, the chosen technical information such as deposition rate, machine purchase price and others will be chosen to be more closer to PBF than BJ and DED. Furthermore, three cost factors will be varied together with the different design cases to investigated their influence on the total cost, cost distribution, and cost over time. The chosen cost factors are deposition rate, machine purchase, and powder price. Table 5 shows the necessary information to conduct a cost estimation analysis, where the bold data is considered to be the base value. The cost related to the base values are considered to be the cost if the part is produced today with AM and will be referred as today’s cost.

Cost estimation data

Machine and material specification

Build size 400 x 400 x 400 mm

Deposition rate 15, 50, 100 cm3/h

Annual machine operating hours 6000 h

Utilization rate, Urate 68,5 %

Machine depreciation time, Td 6 years

Material density 7,8 g/cm3


Machine purchase 1 000, 750, 1500 K€

Powder price rate, Cp 80, 60, 100 €/kg

Labor cost rate, Cl 270 SEK/h

Currency exchange rate e 1 10 SEK

Table 5: Cost estimation data

5.9 Hypothesis

Based on the initial background research, it is believed that it will be possible to manufacture the PTO-Shaft using AM with necessary mechanical properties. However, such a shaft will today not be cost effective nor fast enough to meet the demand.


6 Result

Result concerning the stresses, FEM-simulation, and topology optimization will first be presented in this chapter. It is then followed by the economics for different design cases and future possibilities.

6.1 Design Case 1

Original model, no changes

Figure 17 shows the von Mises stresses of the whole shaft. The maximum stress level were around 285 M P a and occurs at the neck area, see the white color area in figure 17. The maximum deformation on the shaft was measured to be 100 µm. The overall maximum stress occurs at the the gear (> 500 M P a), but is not of interest and therefore left out from this report. Stress singularities are present at sharp corners at the inside of the shaft. The stress levels around the design space are kept relatively low and reaches a maximum of 70 M P a. The total volume is 544 cm3.

Figure 17: Stress surface plot - Original model

6.1.1 Mesh convergence

A mesh convergence was made to ensure if the obtained solution is mathematically accurate. Stress concen- tration singularities are present at the sharp edges of the model and are ignored based on two reasons. The stress singularities increases with deceasing mesh size and rounding the edges decreases the stress levels but does not eliminate the singularities completely. Table 6 shows the mesh convergence of the maximum stress, ignoring the singularities. Figure 18 shows the discussed stress concentration singularities.

Mesh size [mm] Maximum stress [MPa]

2,5 285

1,5 284

1,12 286

Table 6: Mesh convergence, ignoring stress levels at singularities


Figure 18: Singularities

6.2 Design Case 2

Topology optimization on original model 6.2.1 Geometry

Figure 19 shows the topology optimized model. The total material savings reached roughly 1.09 Kg, which corresponds to a 25, 23 % weight reduction. Figure 20 shows the redesigned shaft based on the topology optimized result. Here the corners are rounded to avoid sharp corners. The total material savings reached 0, 93 Kg and corresponds to a 21, 32 % weight reduction. The difference in results are due to the rounding of the corners. The total volume is reduced to 428 cm3. The geometry is not possible or at least very tedious and cost ineffective to manufacturing with traditional methods, mostly because of the bigger inner diameter in the center of the shaft. It is not possible to reach in with a tool and remove material with a diameter larger than the square hole. This is the reason why AM is a technical suitable method for this design case.

Figure 19: Result of topology optimization




Related subjects :