On the economics of additive manufacturing: Experimental findings

• Published date: 01 December 2019
• Author: Matthias Holweg,Martin Baumers
• Date: 01 December 2019
• DOI: http://doi.org/10.1002/joom.1053

TECHNICAL NOTE

On the economics of additive manufacturing:

Experimental findings

Martin Baumers

1

| Matthias Holweg

2

1

Centre for Additive Manufacturing, Faculty

of Engineering, University of Nottingham,

Nottingham, UK

2

Saïd Business School, University of

Oxford, Oxford, UK

Correspondence

Matthias Holweg, Saïd Business School,

University of Oxford, Oxford, UK.

Email: matthias.holweg@sbs.ox.ac.uk

Funding information

Engineering and Physical Sciences Research

Council's 3DP-RDM Networking Grant,

Grant/Award Number: EP/M017656/1

Handling Editors: Jan Holmström,

Matthias Holweg, Benn Lawson, Frits Pil,

and Stephan Wagner

Abstract

Additive manufacturing offers great potential for both product and process innovation

in manufacturing across a wide range of industry sectors. To date, most applications

that have been reported use additive manufacturing to produce either customized parts

or produce at small scale, while the volume manufacture of standard parts largely

remains a conjecture. In this article, we report on a series of experiments designed to

elucidate how quantity, quality and cost relate in additive manufacturing processes.

Our findings show that traditional economies of scale only partially apply to additive

manufacturing processes. We also identify four build failure modes and quantify their

combined effect on unit cost, exposing an unusual property whereby the cost-optimal

operating point occurs below maximum machine capacity utilization. Furthermore,

once additive manufacturing technology is used at full capacity utilization, we find no

evidence of a positive effect of increased volume on unit cost. We do, however, iden-

tify learning curve effects related to process repetition and operator experience. Based

on our findings we propose a set of general characteristics of the additive manufactur-

ing process for further testing.

KEYWORDS

additive manufacturing, capacity utilization, economies of scale, organizational learning, throughput

1|INTRODUCTION

Additive manufacturing (AM) technology, also commonly

known as 3D printing, has captured the imagination of many

technology observers and manufacturing professionals. It is

widely perceived as a means to rethink design, digitize

manufacturing, produce to demand and customize products

without cost penalty (Berman, 2012; D'Aveni, 2015; Manyika

et al., 2013; Segars, 2018). Successful applications have been

reported across manufacturing sectors such as hearing aids,

footwear and prosthetics. A few sectors, like hearing aids,

have switched their entire manufacturing process from tradi-

tional to additive manufacturing within a short timeframe

(D'Aveni, 2013), sparking predictions that additive will

replace traditional (tool-based) manufacturing: “(…)within

the next five years we will have fully-automated, high-speed,

large-quantity additive manufacturing systems that are

economical even for standardized parts”(D'Aveni, 2015,

p.48). AM technology is being seen as “(…)readyto

emerge from its niche status andbecomeaviablealternativeto

conventional manufacturing processes in an increasing number

of applications”(Cohen, Sargeant, & Somers, 2014, p. 1).

It is worth noting that most examples of AM applications

leverage the technology's ability to economically produce

items at small scale, that is, to either customize products or

make “one-offs”with little or no cost penalty. Only very

few examples of the application of additive manufacturing

to the manufacture of standard parts have been reported; the

most commonly cited one is the 3D-printed fuel nozzle for

CFM's LEAP engine that powers popular single-aisle air-

liners. (Annual LEAP production in 2018 was 1,118 units,

which each engine containing 19 identical fuel nozzles.)

Received: 30 October 2018 Revised: 21 July 2019 Accepted: 27 July 2019

DOI: 10.1002/joom.1053

794 © 2019 Association for Supply Chain Management, Inc. J Oper Manag. 2019;65:794–809.wileyonlinelibrary.com/journal/joom

CFM's justification for using AM for this application, how-

ever, is not a lower unit manufacturing cost but a weight

reduction for the part that is now made of a single compo-

nent, compared to 18 components previously, as well as a

reduced risk of “coking”(the build-up of fuel residue in the

hot nozzle), as additive manufacturing allows for the provi-

sion of cooling channels that prevent this from happening

(Shields & Carmel, 2013). The AM applications reported in

the literature are generally based on the technology's specific

advantages to operate without costly tooling, to deal with

variety at little or no cost penalty, and to be able to design

part geometries with few restrictions.

The question that has not been answered, and marks the

focus of this note, is to what degree additive manufacturing

is able to also displace traditional tool-based manufacturing

in contexts where it has to compete on a unit cost basis

alone. In the terms of Locke and Golden-Biddle (1997) this

marks a “noncoherent intertextual field”, characterized by a

common sense of importance yet fundamental disagreement

of the economics of the AM process (Ruffo, Tuck, & Hague,

2006; Baumers, Beltrametti, Gasparre, & Hague, 2017 ver-

sus Hopkinson & Dickens, 2003; Atzeni & Salmi, 2012;

Weller, Kleer, & Piller, 2015).

Specifically, while some studies have suggested that unit

cost levels in additive manufacturing are dependent on quan-

tity (Baumers et al., 2017; Ruffo et al., 2006), others have

argued that this relationship is entirely absent (Atzeni &

Salmi, 2012; Hopkinson & Dickens, 2003; Weller et al.,

2015). This question has substantial implications for the

availability of economies of scale that determine the cost of

large-scale manufacturing operations (Schmenner & Swink,

1998). If additive manufacturing is to fulfil predictions of

large-scale adoption (Conner et al., 2014), it too will have to

exhibit a similar volume-cost relationship. This remains an

area that has not yet received much attention within the oper-

ations management literature, as most research addressing

operations management issues focusses on single case appli-

cations or conceptually outlining additive manufacturing's

potential to disrupt existing manufacturing value chains

(e.g., Cotteleer & Joyce, 2014; D'Aveni, 2013, 2015; De

Jong & De Bruijn, 2013; Laplume, Petersen, & Pearce,

2016; Tuck, Hague, & Burns, 2006; Weller et al., 2015).

In this note we thus build on studies in the engineering

literature that have investigated specific cost aspects of AM

technology (Alexander, Allen, & Dutta, 1998; Atzeni,

Iuliano, Minetola, & Salmi, 2010; Atzeni & Salmi, 2012;

Baumers, Dickens, Tuck, & Hague, 2016; Hopkinson &

Dickens, 2003; Rickenbacher, Spierings, & Wegener, 2013;

Ruffo et al., 2006). We report on a series of experiments that

seek to elucidate the economic characteristics of the additive

manufacturing process; Section 2 reviews the theoretical

foundations, before introducing our experimental setup in

Section 3. Section 4 presents our findings, before proposing

a set of general characteristics of AM processes for further

testing in Section 5.

2|THEORETICAL BACKGROUND

2.1 |Sources of economies of scale in

traditional manufacturing

The economics of traditional (or tool-based) manufacturing

have been widely discussed (e.g., Chandler, 1990; Schmenner

& Swink, 1998), and are fundamental to current manufactur-

ing practice. Processes gain “economies of scale”when an

over-proportionate cost saving is achievable by increasing the

level of production. Economies of scale form a key determi-

nant of the concept of returns to scale, which is often defined

by using the standard Cobb–Douglas production function

(Cobb & Douglas, 1928).

Haldi and Whitcomb (1967) systematically classify the

sources of economies of scale in manufacturing, dis-

tinguishing between economies of scale in static cost rela-

tionships due to throughput-related and indivisibility-related

effects, increasing returns from dynamic sources due to

learning curve effects and stochastic effects relating to the

reduction of process variance.

The concept of static economies of scale reflects (a) the

effect of capacity utilization, which forms some share of

the possible level of machine throughput, thus resting on the

relationship between machine size and unit cost, and (b), the

effect of production volume due to indivisibilities resulting

from a key aspect of traditional manufacturing, which is that

machine operation requires a tool of some kind, and hence

involves a setup cost derived from the need of changing over

tools to produce a given good.

Similarly, the concept of dynamic economies of scale

also stems from the indivisibility of equipment and worker,

in as far as they conjointly determine the outcome. Unlike

static economies of scale, however, they essentially lead to

cost reductions as manufacturing activity progresses. One

source of dynamic economies of scale is process repetition

leading to learning curve effects, which draw on a structured

cycle of defining potential problems, measuring the process,

devising improvements, and verifying their effectiveness

(Anand, Ward, Tatikonda, & Schilling, 2009; Upton & Kim,

1998). Repetition of a standard process allows for the identi-

fication and eradication of unnecessary process steps and/or

reduction of undesired variation, often using bundles of

established “best”practices (Schroeder, Linderman,

Liedtke, & Choo, 2008; Shah & Ward, 2003). Having been

described in the digital context as “wetware”(Shapiro &

Varian, 1999), accumulating knowledge in effect builds an

asset stock residing within a manufacturing firm conferring

BAUMERS AND HOLWEG 795