Production, Manufacturing and Logistics
Traditional or Additive Manufacturing? Assessing Component Design Options through Lifecycle Cost Analysis

https://doi.org/10.1016/j.ejor.2018.04.015Get rights and content

Highlights

  • We compare additive manufacturing to regular production in a component design setting.

  • Break-even characteristics of printed parts are obtained via lifecycle cost analysis.

  • The significant logistical benefits that printing offers are investigated in detail.

  • Printing investments can be offset by reduced lead times and by performance benefits.

Abstract

We consider an original equipment manufacturer that can either design a system component that is produced with traditional technology, or design an alternative component that is produced with additive manufacturing (AM). Designing either component requires a technology specific one-time investment and the components have different characteristics, notably in terms of production leadtime, production costs and component reliability. We support the design decision with a model that is based on evaluating the lifecycle costs of both components, covering design costs, maintenance and downtime costs, and performance benefits. We derive analytic properties of the required reliability and costs of the AM component such that its total lifecycle costs break even with that of its regular counterpart. Through our analysis, a numerical experiment and cases from two different companies, we find that component reliability and production costs are crucial to the success of AM components, while AM component design costs can be overcome to a certain degree by generating performance benefits or by using the short AM production leadtime to lower the after-sales logistics costs.

Introduction

Capital goods are complex technical systems that are essential to their users’ business processes. Examples of such systems are airplanes, trains, military weapon platforms and semiconductor production machines. These systems are characterized by their high lifecycle costs, a large part of which is generated during their exploitation phase. Much of these exploitation costs are predetermined by decisions taken by the original equipment manufacturer (OEM) in the system design phase.

One interesting development, which may help reduce total lifecycle costs, is the development of additive manufacturing (AM) technology. Additive manufacturing, which is also sometimes called 3D printing, can be defined as parts fabrication by creation of successive cross-sectional layers of an object, usually based upon a three-dimensional solid model (Gao et al., 2015). These parts can be plastic parts, but also many types of metal can be used. AM offers engineers greatly improved design freedom compared with traditional manufacturing technologies. This can offer large performance benefits, for instance by reducing fuel consumption for airplanes through reduced component weight or lowering the power consumption of pumps by optimizing fluid flow through cooling channels. Furthermore, AM offers much shorter leadtimes for small production series, which can reduce required spare part investments and increase the responsiveness of after-sales service supply chains.

At the same time, AM components currently require high development costs compared to their regular counterparts. While the regular component is usually an adapted version of a component installed in earlier systems, the AM component often requires new design features, especially if there is a wish to capture potential performance benefits by making use of AM’s design freedom. Further complicating the shift to AM components is the fact that design rules for production engineers are still under development and that preliminary rules are not easily generalized over different products and different AM systems (Yang et al., 2017, p. 83). This complicates the design process and increases development costs. When a design has been decided upon, trial production runs are required to test product reliability and to fine-tune production parameter settings, such as laser intensity or layer thickness. Such fine-tuning must be done in great detail, as each of the different geometric properties of complex products may require individual attention. Laser intensity, for example, may need to change repeatedly when the laser passes over alternating sequences of solid material and cavities of different sizes. This detailed testing can further increase the development costs compared to traditional manufacturing, such as machining, for which material properties are standardized and remain largely constant during the production process (Yang et al., 2017, p. 83). Note that there are exceptions for which the development costs of an AM part may be smaller than those for developing an traditional part, for example in cases where expensive tooling such as casting or injection molds are required. Our model takes this into account by including a development cost difference that is unrestricted in sign.

When evaluating the potential of AM, many product developers currently focus on weighing off the potential performance benefits against the required design investment. This means that potential cost reductions in the after-sales service supply chain, which are due to changes in reliability, production costs and production leadtime, are neglected. In some cases, opting for AM over traditional manufacturing is obvious. GE’s new jet engine fuel injection nozzle, for example, exhibits improved characteristics over the former traditional design in terms of reliability, lead time, performance and production costs (GE Aviation, 2014). However, there are many components for which the characteristics of AM components are not all favorable. In fact, a current limitation of AM technology is that there is often uncertainty concerning the mechanical properties of such parts (Bikas, Stavropoulos, & Chryssolouris, 2016). This may have a large negative effect on the maintenance and repair costs that are incurred over the course of an asset’s lifecycle.

Potential negative characteristics may be offset by a reduction of the average production leadtime, which is expected to be much shorter for small series of spare parts when AM is employed as opposed to traditional manufacturing methods. This is the key assumption that we make for our analytical model. The question remains what properties the AM component must have in terms of reliability and production costs in order to be preferred over its regular counterpart. We introduce a model that compares the total lifecycle costs of the regular part with those of the AM part, taking into account design costs, performance benefits and all spare part related costs, including maintenance and downtime costs. This model is used to evaluate the break-even component production costs and the break-even component reliability, such that the total lifecycle costs of the regular part equal those of its AM counterpart. For these break-even characteristics we derive analytic properties, conduct numerical experiments and we present two case studies to gain insight into the conditions under which an AM component outperforms a regular component.

The design decision that we consider takes place during the system design phase. The OEM can either design a regular component, which is usually based on a component that was used for earlier versions of the system, or he can design a completely new component and make use of the capabilities of AM to capture performance benefits and profit from a much reduced production lead time. Note that not all component types are suitable for AM, so a pre-selection of candidate components can be made based on methods that evaluate AM suitability based on basic component characteristics (e.g., Knofius, van der Heijden, & Zijm, 2016). We support the decision to design a regular component or an AM version by developing a model that considers either reliability or unit production costs as given and provides the values of the other parameters for which one design option dominates the other in terms of lifecycle costs. This is useful in practice if it is hard for design engineers to find good estimates for these parameters. In that case, optimizing a component’s design would be impossible as that involves taking into account the even more difficult to characterize relationships between the design investment, the unit production costs and the component reliability. Each of these relationships, such as the one modeled in Mettas (2000) between design investment and component reliability, is very difficult to parameterize. Doing this properly for all the relationships involved is even more difficult. If engineers estimate that the break-even properties provided by our model will be comfortably met, then AM is preferable. In cases where estimated properties are much worse than the break-even properties, it is better to opt for traditional production technology. If the estimated properties are similar to the break-even properties, more research must be done to more accurately determine the eventual AM component characteristics, or an organization may consider more qualitative reasons to op for the AM version, for example to gain experience with the technology.

Our model can also be applied to redesign decisions that are taken during the exploitation phase. In this case, we require a negligible transition period for replacing the old component with new versions, to avoid a period of the lifecycle where two component types are operating in the field simultaneously. Such fast transitions can occur when there is a large performance benefit to exploit in combination with ample opportunity to upgrade to the AM part, for example in the aviation industry when lighter AM components become available. Another application of our model during the exploitation phase, is when the OEM must redesign a poorly designed component. Earlier studies have shown that in such upgrade situations, it is often advantageous to preventively replace all components directly after redesign, instead of replacing them one-for-one at the time of failure (e.g., Clavareau, Labeau, 2009, Öner, Kiesmüller, van Houtum, 2015).

We determine the lifecycle costs that are generated by all parties in the supply chain, from the OEM to the end user. If the AM component is preferable due to lower lifecycle costs, and there are multiple parties involved in generating the lifecycle costs, including potential benefits, then a method is required to determine in which way each party benefits, for example via game-theoretical methods. Developing such a framework, however, is beyond the scope of this paper. If the level of cooperation in the supply chain is limited, our model can still be used by individual parties, who must then recognize which parts of the lifecycle costs and potential performance benefits apply to their situation. In summary, our contribution is as follows:

  • 1.

    We develop an original model for a component design decision, based on the evaluation of the total lifecycle costs of two competing types of components, one produced with traditional technology and one produced via additive manufacturing. We take into account design costs, performance benefits and after-sales service logistics costs.

  • 2.

    We generate analytic insights into the relationship between design costs, performance benefits and the minimally required AM component characteristics. We conduct a numerical experiment to generate additional insight into situations where AM can likely be successfully applied to component design.

  • 3.

    Two case studies are conducted to test the current applicability of AM in a component design setting.

The remainder of this paper is organized as follows. In Section 2, we survey the literature on related system design problems and on spare parts related to AM. Next, Section 3 contains the model formulation. Section 4 contains the analysis of our model, and Section 5 contains a numerical experiment that is used to generate managerial insight into the potential of AM for spare part supply. In Section 6 we present the two case studies and Section 7 includes some extensions in which we add stochasticity to two of the variables in our model. Section 8 contains our conclusions.

Section snippets

Literature review

We evaluate the design decision to opt for either a regular part or an AM part based on its effect on total lifecycle costs. In this section, we first review literature related to such design decisions in reliability allocation problems and then in warranty problems. Our total lifecycle cost model also includes a spare part inventory system, so we also briefly review literature on spare parts management in relation to additive manufacturing. Finally we review a case study on AM component

Model

In this section we introduce our modeling assumptions and define cost expressions related to the development, production and exploitation of regular and AM parts. A complete overview of all model variables and all model input parameters can be found in Tables 1 and 2, respectively.

An OEM designs a critical component for one of its next generation capital goods, to which we refer as the system. The OEM estimates that it will sell N units of the system and the time until the systems are phased

Analysis

To identify when AM is preferable over traditional production, we require the break-even point where the total lifecycle costs for both components are equal. Such break-even values can be used by decision makers to determine whether or not to opt for AM. In Section 3 we introduced K( · ) as a measure that includes performance benefits and the difference in development costs for AM and regular parts. Given that these play a large role in deciding whether or not to opt for an AM component, we

Numerical experiment

To generate insight into the applicability of AM in practice we conduct a numerical experiment on a range of input parameters. Our goal is to provide managerial insights into situations where AM is most suitable to replace traditional technology, and to provide insights into AM characteristics that require the most attention for the technology to become more widely applicable. We report the following outcome variables:

  • (i)

    K1( · )/cp: AM is not a mature manufacturing technology yet, and high

Case studies

We perform case studies to illustrate the practical applicability of our model, as well as the current performance of AM compared with traditional technology. The first case study is performed at a company that manufactures access equipment and its spare parts. The evaluated component is a stainless steel hydraulic valve block that is used to control the bucket movement of a 60 meter boom lift (see Fig. 5). This component must be able to withstand large hydraulic pressure, but the amount of

Extensions

Some of the variables that we have so far treated as being deterministic may be stochastic in practice. As extensions to our model, we consider the impact of stochasticity in N and in T. We still assume that all systems are both installed and taken out of service at the same time.

We first add uncertainty in T to our model. We assume T to be uniformly distributed on the interval [aT, bT], i.e., its pdf f(t)=1/(bTaT) and its expectation E{T}=0.5·(aT+bT). Our optimized cost function is: C¯x(·)=aT

Conclusions

In this paper, we introduce and develop a model for evaluating two production methods that can be used to produce two differently designed, but functionally the same system components. The practical motivation for this model is the potential that additive manufacturing offers compared to traditional technology, which in our case is increased design freedom and reduced production lead times. The former can create performance benefits, while the latter is beneficial to the after-sales service

Acknowledgment

The research leading to this paper has been supported by NWO under project number 438-13-207. For providing details regarding the two case studies, the authors thank Jorn Jansman, Stijn Verputten, JLG Ground Support, specifically Ton Wolters, and Fokker Aerostructures, specifically Marko Bosman. The authors also gratefully acknowledge two anonymous referees for their role in improving this paper.

References (30)

  • R.H. Teunter et al.

    End-of-life service

    International Journal of Production Economics

    (1999)
  • E. Atzeni et al.

    Economics of additive manufacturing for end-usable metal parts

    The International Journal of Advanced Manufacturing Technology

    (2012)
  • A. Ben-Ner et al.

    Decentralization and localization of production: The organizational and economic consequences of additive manufacturing (3d printing)

    California Management Review

    (2017)
  • H. Bikas et al.

    Additive manufacturing methods and modelling approaches: A critical review

    The International Journal of Advanced Manufacturing Technology

    (2016)
  • GE Aviation (2014). World’s first plant to print jet engine nozzles in mass production. Accessed June 20 2017,...
  • Cited by (89)

    View all citing articles on Scopus
    View full text