Additive manufacturing of spare parts for the
mining industry
- a pilot study on business impact from an aftermarket perspective
Julia Alfredsson, Kristian Vingerhagen
Master of Science in Industrial Engineering and Management Örebro spring term 2021
Examiner: Christer Korin
Additiv tillverkning av reservdelar till gruvindustrin
- en förstudie om affärspåverkan ur ett eftermarknadsperspektiv
Örebro universitet Örebro University
Institutionen för naturvetenskap och teknik School of Science and Technology
The purpose of this pilot study was to identify and evaluate different business cases for Epiroc’s Parts & Services Division (PSD) regarding the use of additive manufacturing (AM), also known as 3D printing, for their spare parts within the mining industry. This study presents an approach for how spare parts promising for AM can be identified and shows the difficulties with AM. The study follows the design research methodology (DRM) standard for research within product and process development. Through interviews and literature searches, a "top-down" approach was applied. A developed cost-benefit model accompanied this approach and was used to identify and evaluate potential spare parts for AM from Epiroc’s current spare parts portfolio. The results were evaluated as promising for several of the spare parts in terms of reduced manufacturing, procurement, tool cost, and lead time reduction, which results in increased uptime for the customer. With reduced lead times, the availability increases for the customer, who may increase Epiroc’s sales and aftermarket revenues in the long run. There is also great potential for reducing the costs for warehousing, where spare parts of low demand can have their stocks reduced or eliminated by securing supply through on-demand manufacturing. Although many exciting business cases have been identified and evaluated, it has been acknowledged that CNC-machining in many cases is the cheaper alternative. Despite this, it is worth investing in AM from a strategic point of view as it is seen as a tool for the future. Before it can be adopted and implemented, Epiroc should do test trials with companies offering AM services. These can be used to update and tune the cost-benefit model accordingly to increase its reliability and validity. The model could also be developed further to incorporate AM’s additional benefits, such as weight and material reduction through design for additive manufacturing (DfAM).
Key words: additive manufacturing, spare parts, aftermarket, mining industry, design research methodology
Sammanfattning
Syftet med denna förstudie var att identifiera och utvärdera olika affärsmöjligheter för Epirocs Parts & Services-division (PSD) gällande användningen av additiv tillverkning (AM), även känd som 3D-printning, för deras reservdelar inom gruvindustrin. Denna studie presenterar ett tillvägagångssätt för hur reservdelar passande för AM kan identifieras och visar på svårigheterna med AM. Studien följer "design research methodology" (DRM), vilket kan översättas till designforsknings-metodologin, som är vanligt förekommande vid forskning inom produkt- och processutveckling. Genom intervjuer och litteratursökningar tillämpades en "top-down"-metod. Detta åtföljdes av en utvecklad kostnadsnyttomodell som tillsammans användes för att identifiera och utvärdera potentiella reservdelar för AM från Epirocs nuvarande reservdelsportfölj. Resultaten utvärderades som lovande för flertalet av reservdelarna vad gäller reducerad kostnad för tillverkning, inköp, verktyg och minskad ledtid, vilket resulterar i ökad drifttid för kunden. Med minskade ledtider ökar tillgängligheten för kunden, som kan öka Epirocs försäljning och eftermarknadsintäkter på lång sikt. Det finns också en stor potential i att minska kostnaderna för lagerhållning, där reservdelar med låg efterfrågan kan få sina lager att reduceras eller elimineras genom att säkra utbudet genom tillverkning vid behov. Även om många intressanta affärsmöjligheter har identifierats och utvärderats har det uppmärksammats att CNC-bearbetning i många fall kan vara det billigare alternativet. Trots detta är det värt att investera i AM ur en strategisk synvinkel eftersom det ses som ett verktyg för framtiden. Innan AM kan anammas och implementeras bör Epiroc göra testförsök med företag som erbjuder AM-tjänster. Dessa kan användas för att uppdatera och justera kostnadsnyttomodellen i enlighet med detta för att öka dess validitet och reliabilitet. Modellen kan också utvecklas vidare för att införliva AM:s ytterligare fördelar, såsom vikt- och materialreduktion genom design för additiv tillverkning (DfAM).
Nyckelord: additiv tillverkning, reservdelar, eftermarknad, gruvindustri, designforsknings- metodologi
We would like to direct a big "Thank you!" to Anders Johansson, the Global Product Manager and our supervisor at Epiroc Rock Drills for coming up with the idea for this thesis and building up the confidence in us to execute this study with good results. Thank you for supervising us along the way, providing us with the contacts needed to conduct valuable interviews, and giving great input on moving forward and developing our study further.
We are also very grateful for the cheerful guidance from our supervisor at Örebro University, Jens Ekengren, who kept us on track and gave valuable insights whenever we were in doubt.
Lastly, we would like to thank everyone at Epiroc and the interviewed representatives in AM industries, who not only showed keen interest in our project but helped us find the right track and openly contributed from personal lessons with the technology. A special thanks to Amexci for showing their facility and sharing plenty of time and knowledge several times during the course of this thesis.
Table of contents
1 INTRODUCTION ………..………….……... 1
1.1 Epiroc ………... 1
1.2 The research problem ……….………....….. 1
1.3 Outline ………...………....…... 2
2 BACKGROUND TO AM OF SPARE PARTS ………..…………....…... 3
2.1 Why additive manufacturing? ………...………....………....…... 3
2.1.1 Objectives ……….…….…... 4
2.1.2 Research questions ………...………..………... 4
2.1.3 Delimitations ………... 4
2.2 Business context ………...……….…... 4
2.3 General industrial context ……….……….…... 5
3 SCIENTIFIC FRAMEWORK ………..…...……….…... 6
3.1 AM - from model to print ……….……….…... 6
3.2 AM technologies ………...………....…... 8
3.2.1 Vat polymerization ………...……….…... 9
3.2.2 Material extrusion ………..………... 10
3.2.3 Material jetting ………...………... 11
3.2.4 Binder jetting ………..……...………..….... 1 2 3.2.5 Powder bed fusion ……….………... 13
3.2.6 Direct energy deposition ………...….... 1 5 3.2.7 Sheet lamination ………...….... 1 6 3.3 Traditional manufacturing technologies ……….……... 16
3.3.1 CNC machining ………..……….……..………..….... 16
3.3.2 Casting ………..………...……….……... 1 7 3.3.3 Injection moulding ……….…...……...…...…... 17
3.4 AM of spare parts ………...….…….……....…... 17
3.4.1 Considering AM for spare parts ……….…...…….……...……... 1 8 3.4.2 Identification of spare parts promising for AM ………...………....…….………..… 19
3.5 Research methods and approaches ………...………... 2 0 3.5.1 Tools and methods for data processing ………...………...…...………... 21
3.5.2 Tools and methods for data gathering ………...………....………... 2 2 4 METHODOLOGY ………..………... 2 4 4.1 Design research methodology ……….………....…... 2 4 4.1.1 Stage 1 - research clarification ………...…….……...…... 2 5 4.1.2 Stage 2 - descriptive study I ………...………...……….... 2 5 4.1.3 Stage 3 - prescriptive study ……….………...…………....……….… 2 6 4.1.4 Stage 4 - descriptive study II ……….………….……….…….…... 2 8 4.3 Methodological considerations ………...………...……….…... 2 8 5 COST-BENEFIT MODEL ……….………....………….…... 3 0 5.1 The basics ………...………...…... 3 0 5.2 Datasheets ………...………...…….... 3 2 5.3 Support sheets ………...……….………….……... 3 3 5.4 User sheets ………...………..……….……... 3 4 5.5 Validation ………...………....……….……... 39 6 RESULTS ………..………....…………....……... 4 1 6.1 Identified KPIs ………....……….……... 4 1 6.2 Identified spare parts ………...………....…... 4 2 6.3 Cost-Benefit analysis of selected spare parts ……….………....… 4 4 7 DISCUSSION ……….………….……….... 4 7 7.1 Evaluation of model and results ………..………... 4 7
7.1.4 Additional comments on identified and evaluated parts ………..……... 50
7.2 Results in comparison to earlier work ………... 51 7.3 Identified uncertainties with AM ………... ....……….... 5 2 7.4 Suggestions for future work ………....………... 5 3 8 CONCLUSIONS ………..…………...………... 5 5 9 REFERENCES ………..…………...………... 5 6 Appendix
A. The Cost-Benefit model B. Data used in the model LIST OF ABBREVIATIONS
AM Additive Manufacturing CAD Computer-Aided Design CNC Computer Numerical Control DC Distribution Center
DfAM Design for Additive Manufacturing DRM Design Research Methodology KPI Key Performance Indicator ME Material Extrusion
1
Introduction
Chapter 1 provides a brief introduction of the company where this study was executed, followed by a description of the research problem and the outline of this report.
This pilot study took place at Epiroc’s subsidiary Epiroc Rock Drills AB in Örebro, presented in section 1.1 Epiroc . The aim was to research how additive manufacturing (AM) of Epiroc’s spare parts would impact their business and identify spare parts suitable to produce with AM, described further in section 1.2 The Research Problem . The chapter concludes with 1.3 Outline of this report .
1.1 Epiroc
Epiroc is a leading global productivity partner that develops, manufactures, sells, and maintains machinery in the mining and infrastructure industry. The company consists of two segments and seven divisions within the company. This pilot study was carried out in the segment of "Equipment and service" in the "Parts & Services Division" (PSD) at the subsidiary Epiroc Rock Drills AB, which has its head office in Örebro, with over 2 000 employees. Epiroc considers themselves to be in a highly competitive market regarding pricing, product design, service quality, development of new products, customer service, and financing terms. In addition to these factors, competitors are also prone to consolidate and increase competition further [1] . Epiroc also continuously faces legislative risks and other risks in connection with running an international business where various political, economic, and social conditions can significantly impact the company and industry as a whole. Market risks also exist as their products and services are used in industries that are either cyclical or affected by general economic conditions that are also very sensitive to fluctuating mineral raw material prices [1] .
In 2019, Epiroc reached sales of SEK 40.849 million with 24 production facilities in 12 countries and sales in more than 150 countries, where two-thirds of Epiroc's sales came from aftermarket revenues [1] . These revenues are driven by their customers' needs for consumables, maintenance, renovation, and repairs, as well as various types of service, tools and hydraulic tools [1] . Epiroc's customers value the availability of spare parts, consumables, service, support solutions, training and maintenance since their equipment often is used around the clock and downtime costs are high. The consumption of spare parts is also high and some parts even require daily replacements as the equipment often is used in demanding environments [1] . The Manager of Application & Analysis at Epiroc R&D Drilling (Interview 22 Mar 2021) points out that these spare parts, with high consumption, invites competition and creates a need to protect Epiroc’s intellectual property (IP).
1.2 The research problem
As the technology shifts toward Industry 4.0 , which is the ongoing development of intelligent manufacturing systems, additive manufacturing (AM), also known as 3D printing, is assumed to have a significant impact on the future of manufacturing and industrial practices. Therefore, AM is a technology that many companies have shifted their focus towards, as it provides new business opportunities and innovations compared to traditional manufacturing methods [2–6] . Even though AM might seem easy at first, it holds many complexities and challenges, making it difficult for companies to identify viable business cases creating a need for a reliable approach of how parts could be identified that especially might benefit from AM.
representation, commonly known as a computer-aided design (CAD) [5, 7] . With this technology, a completely new mindset is required in order for the technology to reach its full potential. A new mindset means that one should not use the same mindset from traditional manufacturing methods and apply it to AM as this does not translate well (Interview CEO, Amexci 4 Mar 2021). The technology also holds a steep learning curve due to characteristics such as anisotropic strength and uncertain lifetime due to differing microstructure and appearance from what the design engineers are used to achieve with traditional manufacturing [5] . The AM technology also has its complexities and challenges, e.g. in the build process where heat conduction, material fusion, and build orientation need to be considered (Interview CEO, Amexci 4 Mar 2021). With that being said, it is essential that companies subjected to new and disruptive technology, such as AM, should develop a strategy towards the risk and potential of implementing that technology.
This pilot study analyses the business impact of AM from an aftermarket perspective and identifies spare parts suitable for AM from Epiroc’s current spare parts portfolio concerning business and technical requirements.
1.3 Outline
The report is organised as follows: in Chapter 2, a contextual background to AM of spare parts is given. Chapter 3 describes the different manufacturing technologies and identification approaches for spare parts suitable for AM and research methods that can be used for this type of study. In Chapter 4, the methodology of this study is presented. Next, in Chapter 5, the cost-benefit model is presented in-depth on how it was developed and validated. Chapter 6 presents the results of this study which are later discussed and validated in Chapter 7 along with recommendations for future work. Finally, Chapter 8 states the conclusions to be drawn from this study.
The next Chapter 2 Background presents the objectives, delimitations and research questions for this study together with a background of AM for spare parts and what Epiroc and others have done previously within the field.
2
Background to AM of spare parts
Chapter 2 presents a background as to why this study within AM of spare parts was conducted, and the objectives, research questions, and delimitations for this study were. The chapter is concluded with Epiroc’s business context and the general industrial context within the field.
In February 2021, Epiroc announced their new vision "Dare to think new". Representatives within the company believe that investments in AM is in line with this vision as it might offer potential for differentiation and increased availability. Furthermore, AM technology is evolving fast and with this in mind Epiroc wishes to adopt the technology in their processes at the right time and in the most beneficial way. Section 2.1 Why additive manufacturing? will present this further followed by section 2.2 Business context , where Epiroc’s previous studies within the field will be presented and then put in perspective with section 2.3 General industrial context .
2.1 Why additive manufacturing?
One of the hardest things with being best in class within a segment or industry is constantly evolving and predicting what will define the industry in the future [8, 9] . Considering that Epiroc’s aftermarket sales account for more than two thirds of their revenues, it is reasonable to further evolve the aftermarket business in order to benefit both Epiroc and their customers. As many of Epiroc’s customers' mines are located in remote areas, and machines are dating back to the 1970s that Epiroc strives to maintain, it is natural from an aftermarket perspective to strive for improved key performance indicators (KPIs) such as reduced lead times, maintained delivery, reduced warehousing and increased machine uptime.
Epiroc’s commitment to maintaining machines that range back to the 1970s, along with conceptual versions of machines that were only manufactured once, places high demands on their supply chain in order to maintain such a large spread of both old and new spare parts. Hence, the warehousing of spare parts results in large amounts of working capital. At times this results in stocked items being subject to scrapping if not sold within a certain timeframe. If AM could be used to manufacture spare parts, Epiroc could potentially achieve increased profit and significantly reduce their warehousing. AM also enables reduced lead times and thus reduced machine downtime, which can be assumed to increase Epiroc’s customers’ incentive to accept a higher purchase cost to get the spare part in place faster. Another interesting view was highlighted by the Engineering Manager at Epiroc PSD (Interview 3 Feb 2021), who reflected upon the win-win situation that a reduced lead time could uphold. As the customer achieves increased uptime and profit, this also invokes an increased need for maintenance and service as the machines are continuously worn.
When applied correctly, AM can offer improvements to form, function, costs and lead time. With advancements in AM technologies’ production speeds and reduced costs, companies can find new opportunities and solutions. The same AM equipment used for prototyping can also be used for end-use production without expensive tooling and setup costs which injection moulding, casting and machining are known for [10] . AM is becoming more capable of producing end-use parts and can be used to complement or replace traditional manufacturing processes for a growing range of applications in low- to mid-volumes [10] . The difficulty within this field is to find viable business cases and spare parts for AM that a company’s management can approve in order for the company to get started with the technology and learn from the opportunities that it brings in order to be resilient to the future market changes.
2.1.1 Objectives
The mission of this study was to identify and evaluate different business cases for AM within Epiroc’s current spare parts portfolio in the mining industry. The results can be used to improve the company’s aftermarket business.
2.1.2 Research questions
The research questions for this study were:
● How would introducing additively manufactured spare parts affect Epiroc’s key performance indicators from an aftermarket perspective?
● For what business cases can additive manufacturing of spare parts be beneficial and add value?
2.1.3 Delimitations
This study examines a segment of Epiroc's current spare parts portfolio for their customers’ machines and whether certain spare parts could benefit from being manufactured with AM using the current technology as of 2021. The spare parts are assumed to retain their original design, which is why limited accounts will be given to redesigning and optimisation for AM.
Since Epiroc was primarily interested in what business cases and for what spare parts AM can be viable, this study will mainly focus on the economic aspects and Epiroc’s key performance indicators (KPIs) from a supply chain and aftermarket perspective. The study is therefore limited with regards to the technical and mechanical aspects of the spare parts. The analysis will compare traditional parts to AM parts, which means that unknown variables are assumed to have a similar effect between the two and will not be analysed further. Unknown variables include, but are not limited to, administrative costs, quality-related costs and exchange related costs. Due to a limitation in time and resources, the support tool for analysis will not be implemented and tested in real-life situations.
2.2 Business context
Epiroc started investigating the possibilities of implementing AM in 2016, more particularly, for their spare parts at the beginning of 2019 [11] . Some test parts have been printed without any redesign for AM, from which good results were recorded (Interview Global Product Manager, Epiroc PSD 21 Jan 2021). An example of this was the printing of a critical Product X tested in different materials. This product was considered challenging to manufacture with traditional manufacturing methods and proved to perform better and with unexpectedly good durability if manufactured with AM, with only a 10 % decreased lifetime (Interview Global Product Manager, Epiroc 2 Mar 2021). Another example was a mixer initially consisting of 12 parts that Epiroc redesigned and printed as one component. Even though this mixer has not been tested inside a machine yet, it has shown great potential in reducing the need for warehousing, the number of articles, leakage points and increased serviceability (Interview Engineering Manager PSD 3 Feb 2021). Although these examples have shown promising results, AM is still not adopted as a tool for manufacturing within Epiroc since many divisions find it hard to motivate AM form an economic perspective. This goes hand in hand with the fact that it seems like they currently have no particular strategy as to how spare parts are identified for AM.
2.3 General industrial context
Today AM is used for a variety of purposes. Shahrubudin, Lee, and Ramlan [7] mention artificial heart pumps, jewellery, artificial corneas, rocket engines, cars, steel bridges, and other aviation and food industry products. AM can also be used to manufacture hearing aids [12] , shoes [13] , bone tissue engineering scaffolds [14] , and even to build houses [15] . Compared to traditional manufacturing methods, AM allows for much less material being used, which is a prominent factor for future sustainability. Traditional methods either remove material to create features, e.g. machining, turning, woodworking, or fill pockets and gaps to create geometries, e.g. casting, injection moulding.
Generally speaking, Sweden is falling behind in AM development compared to other countries [16, 17] . In Sweden, a handful of companies are spearheading AM in an industrial context. Siemens Energy AB is one of the companies heavily invested in AM and currently uses it to produce and repair products with the aim to achieve "AM on demand" services [18] . Another company working with AM is Company X, which is a large worldwide company. They started an AM initiative at the beginning of 2018 that resulted in an AM lab consisting of four material extrusion (ME) printers, three for polymers and one for metal (see section 3.2.2 material
extrusion ) (Department Manager Company X Interview 17 Feb 2021). This AM lab is mainly used for R&D as a first step to integrate the technology as an everyday tool in the employees' operations before manufacturing end-use products. According to the Manager at Company X, Design for Additive Manufacturing (DfAM) plays a crucial role in finding viable business cases and will naturally impact all key performance indicators. At the same time, the ownership of Company X’s parts and control of the brand increases as they do not have to hand over their design to an external party for manufacture. The third and final company active in Sweden to be mentioned in this study is Amexci AB, a shared venture first proposed to Swedish industrial giants by Marcus Wallenberg in 2017. The funding shareholders are Atlas Copco AB, Electrolux AB, FAM AB, ABB, Husqvarna group AB, Höganäs AB, Saab AB, Scania AB, SKF AB, Stora Enso AB and Wärtsilä AB [19] . Amexci was founded to grow metal AM in Sweden [20] . Their primary focus is to investigate, analyze, find applications and support companies within the field of AM. Amexci is already contributing to publicly funded research, where they investigate the health risks and environmental impact that nanoparticles might pose from metal PBF processes [21] . Currently, they own several metal Powder Bed Fusion (PBF) printers, a few Material Extrusion (ME) and Vat Polymerization (VP) printers and a polymer PBF printer. The different AM technologies are further explained in Chapter 3 Scientific framework. In addition to this, they have a well-equipped lab, experienced personnel and everything needed for printing except the machines needed for post-processing. For post-processing, they have cooperated with HITAB, which specializes in machining applications.
In the following Chapter 3, the scientific framework and theories within the field of this study are presented.
3
Scientific framework
Chapter 3 outlines the scientific framework for this study and describes different additive and traditional manufacturing technologies and processes, and a section for different methods on how spare parts can be identified as promising with AM. Figure 2-10 illustrates the different additive manufacturing technologies and are solely made for demonstrational purposes. The chapter is concluded with different research methods and approaches that can be used to research the nature previously described in Chapter 2.
In order to understand where and when AM of spare parts can be considered beneficial it is first and foremost essential to be aware of how AM works and what different AM and traditional manufacturing technologies are available today. Therefore, the following sections will introduce how AM works and what technologies are most commonly available and their optimal applications. Furthermore, topical traditional manufacturing technologies are briefly described and different methods for identifying spare parts promising with AM.
3.1 AM - from model to print
Each AM technology is slightly different, but the workflow can be considered quite similar even though each printer and method may vary in how they must be prepared. Figure 1 shows a simplified workflow for 3D printing, where the first step is to create a model in CAD software, then export it to a slicer software which then creates a code used in the printer software to build the object [5] .
Figure 1. Workflow from CAD model to print
CAD is an acronym for Computer Aided Design and describes a 2- or 3-dimensional drawing created with a computer [5, 22] . CAD designs can be created with a wide range of software. What software is used depends on many variables and preferences; applications and end-use, knowledge and familiarity. From a CAD model, the next step is to translate it into a toolpath for the specific printer [5] . In order to do so, slicer software can be used. The file generated from the CAD software is often exclusive to the same program and can therefore not be opened or edited with other software. For this reason, CAD models must be described in a way that is compatible with the slicer software. A generally accepted format is the Standard Triangle Language (STL), which describes CAD models in the forms of triangles [5] . Considering a cylinder of 1 mm in diameter and 0.1 mm in height, this may be described in a CAD model with one text line as simple as " cylinder(h = 0.1, d = 1, center = true) " [23] . However, when exported to an STL file, the designer, or the software, must select an acceptable level of detail since the same circular shape cannot be described as a circle in the STL format [24] . This issue is visualized in Figure 2 , where the model on the right has a high resolution, described with 1432 triangles which altogether describes the object quite well. The model on the left is described in a low resolution with 370 triangles. Whilst both models describe an object of the same size and orientation, the amount of information in each model varies depending on the resolution (i.e. number of triangles). Hence, a lower resolution results in circular features being more angular.
Figure 2. A low amount of triangles in a) described with 370 triangles. A high amount of triangles in b) described with 1432 triangles. The planar surfaces are still planar; rounded edges lose information and become more angular.
When the CAD model has been saved into an STL or similar format that the slicer software can open, the object can be transformed and sliced as a readable code for the printer. Slicers, just as CAD software comes in many forms and slicer choice differ depending on chosen AM-technology and printer. For the slicer to create a suitable toolpath, it must know the printer specifications and limits, e.g. build size, movement speeds, nozzle size, laser spot size, laser intensity, and acceptable code language [5] . Today the slicers usually create a two-dimensional toolpath that the printer executes before moving up one layer height and prints the next layer. This system is then repeated for n number of layers until the model is complete. See Figure 3 , showing the toolpath for the first and second layer.
Figure 3. a) first layer, and b) second layer. The grey area in both a) & b) visualizes what is left to print.
This restriction of working layer by layer is sometimes referred to as 2.5D and not true 3D [5, 6, 25] . This layer by layer method is the reason behind the name slicer, as it slices each object into layers. When slicing an object, the user can make choices within the printer's capabilities. In order to do this correctly, skill and knowledge are required. The choices made in this step may be the difference between a successful or unsuccessful print and dictating, e.g. print time, quality,
strength, accuracy and layer bonding. See Figure 4 for reference. The toolpath to the left is optimized for a small nozzle diameter and will result in a fine surface finish in x, y and z-direction using low layer heights and thin line width. The small nozzle to the left will increase the build time significantly, whereas the toolpath to the right is optimized for a large nozzle diameter and much shorter build time at the expense of visible layers and distinct lines. The same principle also applies to laser systems, where the laser spot size corresponds to the nozzle diameter, and layer height is the same.
Figure 4. a) 0.08 mm layer height with a 0.2 mm nozzle versus b) 0.7 mm layer height with a 1 mm nozzle on the right.
When the CAD model has gone through all steps above, the instructional code can be saved for the specific printer. AM printers are just as machining or turning machines, a computer numerically controlled machine (CNC) [5, 26] . Therefore, this instructional code is saved as a numerical control code (NC-code) which may also be known as G-code. The NC code can be read and interpreted by the software installed in the printer [5, 26] . This code has everything the printer needs in order to perform the correct movement and instructions. One difference between these types of code is that machining equipment usually relies on linear and curve commands, whereas slicer software and AM machines usually rely on linear commands due to the STL step.
3.2 AM technologies
When identifying spare parts promising for AM, Frandsen et al. [27] points out that it is vital to understand the advantages and disadvantages of different AM technologies, before beginning this process. Understanding the different pros and cons facilitates choosing the appropriate technology and equipment for the individual company’s needs and desired objectives. The industry for AM is continuously developing, and naturally, there are many different processes available, most of which are mentioned in this report. These processes can be divided into seven technology groups classified by the American Society for Testing and Materials (ASTM); as seen in Figure 5 , where processes that share a common type of machine architecture and similar material transformation physics are grouped [5, 28] .
Figure 5. Additive manufacturing technologies and some of their corresponding processes mentioned in this framework.
Each AM technologies and processes use various 3D printing materials, ranging from plastic filaments to photosensitive resin to powdered material. Since a review of all the processes is beyond this paper’s scope, the following sections will only describe the AM technologies in brief and the most common processes used therein.
3.2.1 Vat polymerization
Vat polymerization (VP), displayed in Figure 6 , is a technology that uses a vat filled with a liquid, radiation-curable resin called photopolymer that is cured by selectively delivering energy layer by layer, using a laser or UV-light [5, 10] . The curing source is not shown in the illustration but is located under the vat. Flat objects are often printed at an angle as this reduces the contact time between the cured and uncured fluid and changes the roughness of the surface [29, 30] . There are three main processes used for VP printing: Stereolithography (SLA), Digital Light Processing (DLP) and Liquid Crystal Display (LCD) [5] . What sets them apart is mainly the different light sources used for curing. The SLA uses a laser to cure the object point by point and is, therefore, a slower yet more precise process compared to DLP and LCD that uses a UV light source to cure the object one layer at a time [5] . The VP technology is considered an affordable, flexible and high-resolution printing process that easily fits the workspace [5] . However, the process is limited to only photopolymers that tend to age which causes the mechanical properties to degrade and therefore makes the technology difficult to use for many applications [5] . This technology is mainly used for objects that require a high level of detail and a smooth surface finish, such as moulds for jewellery, dental applications, rapid prototyping, functional prototyping, concept modelling and short-run productions [5] .
Figure 6. Illustration of the VP process, where the build platform moves up and down into the vat filled resin tank (curing source excluded).
3.2.2 Material extrusion
Material extrusion (ME) is a technology where the material is extruded through a nozzle and melted in a chamber onto a build plate [5] . For each layer, the part either needs to move down one layer height or the extruding system needs to move upwards [5] . A pattern represents each two-dimensional layer. The pattern created will be controlled by the chosen nozzle size, where bigger nozzles result in wider lines and fewer passes, see Figure 4 . The minimum feature size is determined by the nozzle size [5] . Because the objects are created layer by layer, parts tend to be stronger in the X-Y direction than Z [5] . Most ME systems rely on using temperature or chemical approaches to control melting and solidifying. The chemical approach relies on air, curing agents, solvents, or drying to cure the material [5] , which can be used for processes such as concrete printing of structures.
Figure 7 illustrates a simplified ME system. The material filament is stored on a spool and is either pushed through a tube by the extruder and into the heating chamber (Bowden extruder) or pulled directly into the heating chamber (direct drive). The extruder is a gear system that grips the filament, pulls it from the spool and pushes it into the heating chamber. Extrusion of material through the nozzle is controlled by pressure built in the heating chamber by the extruder. There are many different designs for controlling the XYZ motion. The model in Figure 7 can control the X motion by moving the heating chamber left to right on the grey gantry, Y motion by moving the build plate back and forth, and Z by moving the gantry up or down in relation to the build plate.
Figure 7. Material extrusion illustration - the printer head is mounted and can move in the X-direction on a gantry. The gantry can move in the ZZ-direction, and the build plate supporting the printed part can move in the YY-direction.
ME is more commonly known as fused filament fabrication (FFF), fused filament deposition (FFD) and fused deposition modelling (FDM). The original term, FDM, was trademarked by Stratasys upon the invention of the process, which led to several other abbreviations [31] . ME is the most commonly available AM process due to the wide range of printers and materials available at all price ranges. The principle for each machine is basically the same; however, the more expensive the printer, the printing environment becomes increasingly controlled. A desktop printer for home use allows for control of the hot end and print bed temperature, whereas a high-end machine allows for control of, e.g. chamber temperature, humidity, automatic filament switching and dual extrusion print heads [32] . The availability and range of materials to choose from are increasing every day. Many or all companies that sell filaments are experimenting with different additives to achieve desired characteristics of the material. These new products are primarily based on the materials PLA (polylactic acid), ABS (acrylonitrile butadiene styrene), TPE (thermoplastic elastomers), PETg (polyethene terephthalate + glycol), PVA (polyvinyl alcohol) or Nylon (polyamide). Many companies produce filaments, printers or both. Some companies experiment and sell their own filament brands, consisting of common filament material mixed with materials, e.g. carbon or glass fibre [33, 34] . The latest news within ME are filaments consisting of metals combined with a polymer, which allows for low cost-metal AM. However, this process requires a debinding process to remove the polymer followed by a sintering process to achieve the necessary binding and mechanical properties [35] . The removal of the polymer and sintering results in shrinkage, which may or may not be controllable to an acceptable accuracy. Markforged is a company that offers additional possibilities with their continuous carbon fibre solution, which enables the designer to add a continuous string of carbon fibre embedded in the part in order to increase and control the strength of a part. This solution led to the first 3D printed lifting tool with a CE certification [36] .
3.2.3 Material jetting
The material jetting (MJ) technology is similar to that of conventional inkjet printers in that it distributes all of its print material from one or several printheads [5] . The technology is considered a low cost, high-speed process which has the opportunity of building parts in multiple materials and colours [5] . It is also easily scalable by adding extra printheads, which, compared to other AM
technologies that use a laser or electron beam, makes MJ a lot cheaper [5] . The biggest issue with MJ is its limitation of available materials, limited to photopolymers and waxes. Any changes in the material composition or the physical setup may also dramatically influence the results due to the sensitivity of the process in converting a continuous volume of liquid into a number of small discrete droplets [5] . The parts produced with MJ are not considered suitable for industrial products, according to Frandsen et al. [27] , as they tend to be brittle and show poor mechanical properties. MJ is therefore mostly used for prototypes and low-run injection moulds [5, 27] . Due to its complexity and technical shortcomings, Gibson et al. [5] believe that further growth in the MJ industry is limited.
3.2.4 Binder jetting
The binder jetting (BJ) technology is similar to both the material jetting (MJ) and powder bed fusion (PBF) technology. Compared to MJ, the BJ printhead only distributes a binder selectively onto a powder bed that covers the build chamber, similar to PBF, where the binder and powder forms the cross-sections of the part [5] . BJ is therefore one of the fastest technologies on the market and enables printing of multiple parts with an easy conversion from printing prototypes up to serial production by simply adding extra affordable printheads, similar to that of MJ. See Figure 8 for an illustration of how a BJ printer may look. There is no need for support structures, tooling, or build plates, and it can fabricate assemblies of parts and kinematic joints since the loose powder can be easily removed and completely reused for future prints since no heat is used in the process [5] .
Figure 8. Binder jet illustration - The platform (1) can move down for each new layer, the powder distribution (2) disperse a thin layer of powder across the platform, the gantry (3) carries the print head (4) across the build area so the binder can be distributed and bind with the powder.
BJ is a fast technology, but it requires extensive post-processing in the form of removing the part from the powder bed after the binder has cured completely, which may take up to several hours before removing unbound powder with pressurized air. The finished part results in a green part with a 50-60% density, which can be infiltrated to make it stronger and improve its mechanical
properties. Metal BJ parts are also subjected to three furnace cycles; 1) burning off polymer binder at low temp, 2) light sintering in high temperature, and 3) infiltration of bronze to make the part 90-95% dense [5] . Ceramic parts follow similar processes as metal parts, with thermal decomposition before sintering and sometimes an infiltrant is used to form a ceramic binder or ceramic-metal composite. Machine finishing is also required for high tolerance and mating surfaces [5] .
The challenge with BJ for metal is that to print a high-density part, very fine metal powders (<5 microns similar to the size of baking powder) must be used. These powders are called MIM powders and are considered the most challenging materials to process [37] . It is difficult to distribute it uniformly onto the powder bed since the particles are prone to clumping, caking and powder clouds. Once a layer is in place, one droplet of binder can cause the particles to ripple or displace [37] . Furthermore, parts fabricated with BJ are fairly weak and are mostly used as visual or light-duty functional prototypes but can be strengthened with infiltrants for more functional purposes. BJ parts can also be used as patterns for investment casting and as moulds and cores for sand casting [5] . A wide range of polymers, composites, metals, sand and ceramic materials can be used in this process, but only a subset of these are commercially available [5] . Materials can be combined into cermets (composites of metal and ceramic), which is not possible or easily achieved using the direct processes of AM, e.g. ME, sheet lamination [38, 39] .
3.2.5 Powder bed fusion
The powder bed fusion (PBF) technology was invented at the University of Austin, where they used a selective laser sintering (SLS) process [5] . Laser sintering (LS) is still the most commonly used thermal process within PBF for achieving powder fusion [5] . LS was initially used for fusing polymer and later adopted to fuse metal and ceramics [5] . A PBF machine function according to a few simple principles, see the illustration in Figure 9 and consists of a closed chamber with a build bed, chamber heating, powder dispensing system, a heat source for the fusion process and scanning mirrors. The build platform is lowered for every layer, and a new layer of unfused powder is deposited across the entire chamber surface. When the new layer of powder is deposited, the fusion heat source (often a laser) is used together with the scanning mirrors to trace the cross-section of each layer. The chamber heating source secures an evenly distributed temperature slightly below the powder’s powder’s melting point to eliminate warping problems as the fused powder cools down. This process is repeated until the part is finished. After the production and cooldown period, the part and excess powder can be removed from the part and build chamber.
Figure 9. Powder bed fusion illustration - The platform can move down for each new layer. The “ powder distribution” dispenses a thin layer of powder across the platform. The mirror (1) traces the cross-section of the layer, the laser (2) emits the energy needed to fuse the powder, the chamber heating (4) maintains an evenly heated chamber.
The PBF processes have gained popularity worldwide due to their broad range of materials that can be processed and used as end-use products instead of just as prototypes. As long as the material can be made into powder, melted and resolidified, it is possible to use it in a polymer or metal PBF process [5] . For metal PBF, weldable materials can be used in most cases [5] . For polymer PBF, traditionally, polymers with semi-crystalline structures such as polyamide (nylon) are preferred and almost exclusively used. This is because they have a definite melting point, which results in better accuracy as the exact amount of energy needed to melt a certain amount of material is known, which in turn enables quick solidification after printing [5] . More amorphous materials with a glass transition temperature instead of a melting point often results in poorer quality and reduced accuracy [5] . For metal PBF there are primarily four different processing methods; full melting, liquid phase sintering (LPS), pattern and indirect processes . For full melting and LPS (a partial melting process), the metal part can usually be used right of the build plate after removing it. Pattern processes use wax that creates a pattern that either melts away when casting or is used as a pattern when sand-casting [5] . The indirect process uses a polymer either coated or mixed with metal particles which results in the ability to fuse the polymer and metal with high precision into a green body without using temperatures deteriorating the mechanical properties of the metal. After printing a metal part with the indirect process, the part must be debinded, similar to that of BJ, through sintering which results in a brown part that is fairly porous. The brown part can be processed either by continuing the sintering process to create a denser part or by infiltrating a second metal with a lower density to create a fully dense component. The difference between all of these PBF processes, such as directed metal laser sintering (DMLS), electron beam melting (EBM), etc., are slight modifications to the SLS process. This is to achieve better productivity, enable the processing of different materials, or to avoid specific patented features [5] . Compared to AM technologies that do not use powder, PBF has longer lead times due to certain requirements such as post-processing, skilled operators, and advanced material handling systems. Usually, the chamber also needs to be filled with several parts to be economically viable [40] . PBF is best suited for producing strong functional parts with complex geometries and a consistent surface
finish. Hence, its primary applications are in functional parts, low volume part production and complex ducting. Thus, PBF can be considered as a potential technology for spare parts [27] .
3.2.6 Direct energy deposition
Directed energy deposition (DED) is a technology reminiscent of plasma welding and laser cladding. The feedstock material, either in wire or powder form, melts during or after its deposition into the melt pool from a previously laid layer [5] . The DED technology is used for many different processes, such as Laser Engineering Net Shape (LENS), Direct Metal Deposition (DMD), Electron Beam Additive Manufacturing (EBAM), and dozens more. What differentiates them are the laser power, laser spot size, laser type, powder delivery process, inert gas delivery process, feedback control scheme, and motion control. See Figure 10 for an illustration of DED. The technology is typically used for repairs, adding of features onto existing components, or manufacturing parts in near net shape that typically demands a lot of post-processing and material removal [5] . DED can also improve components’ performance and lifetime by depositing dense, corrosion and wear-resistant metals in thin layers [5] . The technology is relatively slow compared to other AM technologies and creates parts with a high density and a highly controllable microstructure, similar to PBF parts. Its poor resolution and surface finish are the main limitations due to the relatively large melt pools that make it difficult to produce small scale functions, thin walls, and complex parts.
Figure 10. Illustration of direct energy deposition technology. Inert gas and metal powder (1) is fed through tubes to the nozzle fused by a fusion source (2). The entire assembly can move in all three dimensions in relation to the part.
What makes this technology unique compared to other AM technologies is that DED machines can be of either a 3-, 4- or 5-axis system with rotary tables or robotic arms that can access all paths of a part and enable parts to be manufactured without the need of support structures [5] . The machines can also handle very large parts, such as the Sciaky EBAM 300 series, which has a build volume of 8605 litres [41] . Some companies provide DED printers combined with CNC (computer numerical control) machines that can accomplish both additive and subtractive manufacturing in one apparatus. This enables users to manufacture parts with geometries inaccessible in the finished condition and, therefore, not possible to manufacture using any traditional or other AM technique. DED therefore, gives the designer a lot of freedom [5] . The main challenge with this technology is to find the optimal deposition conditions that demand tradeoffs between build speed, accuracy, and preferable microstructure.
3.2.7 Sheet lamination
Sheet lamination (SL) is a technology that stacks and cuts sheets of material on top of each other to form an object. The material can be of either paper, polymer, metal, ceramic or composite and the cutting is usually made with a laser or mechanical cutter. The bonding is made with either glue, adhesive bonding, thermal bonding, clamping, or ultrasonic welding. The sheet lamination technology’s most common processes are: laminated object manufacturing (LOM) and ultrasonic consolidation (UC), also known as ultrasonic additive manufacturing (UAM) [5, 28] .
This technology offers parts of little shrinkage, low residual stresses and distortion problems where large parts can be fabricated rapidly with a variety of build materials and with the possibility to integrate wiring, fibre optics, sensors, and instruments. SL is therefore considered a robust, flexible and valuable process for many applications and materials that can create complex multifunctional AM parts. Bhatt et al. [28] present in their study that a developed robotic sheet lamination-based AM process has several advantages. Some of them are increased build speeds, lowered costs, multi-material capabilities, the inclusion of prefabricated components, and the capability to easily build large parts compared to other AM processes (e.g., stereolithography, direct energy deposition, etc.) [28] . As this is a process that stacks sheets of material, it evidently causes the parts to have anisotropic strength where many types of complex overhanging geometries cannot be built due to the lack of an automated support material [5] .
3.3 Traditional manufacturing technologies
In contrast to additive manufacturing technologies, traditional and subtractive manufacturing technologies, such as CNC machining, casting, and injection moulding, are described below.
3.3.1 CNC machining
Computer numerically controlled (CNC) machining is the automated control of machining tools, such as drills, lathes (also known as turning), and mills that process a block of material from a predetermined program code, commonly known as an NC- or G-code [42, 43] . It is one of the most common subtractive manufacturing technologies [44] and remains a cost-effective way to produce metal parts [45] . The CNC process is mainly used for hard and brittle materials, such as steels and metal alloys, but can also be used for softer materials, such as machinable foams, waxes, and even some polymers [5, 45] . The G-code can be written either by hand or obtained using CAD, computer-aided design, or CAM, computer-aided manufacturing [42] . Therefore, it is a technology that, similar to AM, requires a specific mindset of the designer to optimize the machining strategy, i.e., optimize the tool path pattern, tool shape, size, as well as the cutting parameters for feed, speed, etc. [45, 46] .
To compete with the freedom of design that AM provides, a 5-axis CNC machine is required [45, 46] . However, even with a 5-axis machine, some internal geometries are still impossible to achieve, and the process is both loud and messy compared to AM processes [45] . It is also a manufacturing technology that results in wasted material, making the process less sustainable [47] . As a rule of thumb, it is said that geometries that can be machined using a single set-up orientation are often considered faster and more cost-effective with CNC machining [5] . More complex geometries with multiple set-up orientations are thus limited to the tool shape and tool access, making them more cost-effectively manufactured with AM [5] . Even so, Bo et al. [44] and Gibson et al. [5] mention that CNC machining remains an important manufacturing process, especially for parts that require high stiffness, homogeneity, and predictable quality.
3.3.2 Casting
Casting is one of the oldest shaping processes, where molten metal flows by gravity or other force into a mould and solidifies to the shape of the mould cavity [43] . The mould can be either an expendable mould or a permanent mould. The expendable mould, usually made of sand, plaster or similar materials maintained by binders, enables more intricate geometries. However, the mould must be destroyed to remove the casting [43] . The permanent mould enables higher production rates and can be used repeatedly and is usually made in metal or other material that can withstand the high temperatures of casting [43] . Although casting is considered a versatile manufacturing process where almost all types of materials can be processed and where parts weighing over 100 tons can be manufactured, it also comes with safety hazards for humans processing hot molten metals and environmental issues from the pollution caused in the process [43] . Some casting processes also have poor dimensional accuracy, surface finish, problems with porosity, and limitations in mechanical properties. An example of this is the sand casting process, one of the most common and important casting processes. Castings are generally inaccurate with surface finishes of 6 µm and tolerances of 10 to 20 times higher than those for machined parts [43] . When it comes to manufacturing spare parts of low-demand with casting, there has now been a shift of focus in the industry towards the potential of using AM to print disposable moulds [48, 49] . This is already commonly used for smaller objects in dental and jewellery applications [7] . The tooling cost may be reduced for heavier parts and eliminate the need for maintenance of older moulds and tools. It may also reduce the warehousing costs for low volume or final batches of spare parts often kept with a safety stock (Interview Global Product Manager, Epiroc PSD 4 apr 2021).
3.3.3 Injection moulding
Injection moulding, introduced in 1921, is the most widely used moulding process for thermoplastics similar to casting [43] . It is a process that enables the manufacture of complex and intricate shapes of parts up to 25 kg that are almost always in net shape [43] . The moulds are often very expensive as they are custom designed [43] and typically manufactured in steel or aluminum [50] . For large, complex parts, the mould can cost hundreds of thousands of dollars. If it is a mould for smaller parts, it usually contains multiple cavities, making these moulds more expensive and therefore mainly considered economical for large production quantities [43] .
Since AM technologies can be used to produce tools and moulds, additional opportunities arise if AM would be combined with traditional manufacturing methods, such as injection moulding for spare parts production [27] . Simpson et al. [50] have shown that AM injection moulds can create parts with comparable properties to those made from a standard metal tool, yet at a dramatically lower cost and lead time, which is highly beneficial when developing moulds for R&D, but also short-run production.
3.4 AM of spare parts
When it comes to AM spare parts, not only is it important to understand the advantages and disadvantages of every AM technology, it is equally important to understand that AM has not been developed to replace traditional manufacturing methods [2, 51] . Instead, it has been developed to widen the selection range of processes for manufacturers and customers, where the choice of the manufacturing method is application-dependent [2, 51] . Therefore, the following sections will describe when and why AM of spare parts might be beneficial and how to identify and analyze how a company's key performance indicators (KPI’s) can be affected.
