This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to:
http://creativecommons.org/licenses/by-nc-nd/2.5/
Cost estimation for rapid manufacturing – laser
sintering production for low to medium volumes
M Ruffo*, C Tuck, and R Hague
Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Leicestershire, UK
The manuscript was received on 25 November 2005 and was accepted after revision for publication on 2 May 2006.
DOI: 10.1243/09544054JEM517
Abstract: Rapid manufacturing (RM) is a modern production method based on layer by layer
manufacturing directly from a three-di mensional computer-aided design model. The lack of
tooling makes RM economically suitable for low and medium production volumes. A
comparison with traditional manufacturing processes is important; in particular, cost
comparison. Cost is usually the key point for decision making, with break-even po ints for
different manufacturing technologies being the dominant information for decision makers.
Cost models used for traditional production methodologies focus on material and labour
costs, while modern automated manufacturing processes need cost models that are able to
consider the hig h impact of investments and overheads. Previous work on laser sintering
costing was developed in 2003. This current work presents advances and discussions on the
limits of the previous work through direct comparison. A new cost model for laser sintering is
then proposed. The model leads to graph profiles that are typical for layer-manufacturing
processes. The evolution of cost models and the indirect cost significance in modern costing
representation is shown finally.
Keywords: rapid prototyping, rapid manufacturing, cost model, low-v olume manufacture,
laser sintering
1 INTRODUCTION
With the arrival of additive manufacturing technolo-
gies, some traditional production methods could be
replaced with technologies that are derived from
existing rapid prototyping (RP) [1]. The main benefit
of implementing these new technologies lies in the
ease of passing from design to production, avoiding
intermediate steps such as tool creation. If tooling
can be removed from manufacturing, several advan-
tages can be gained; namely, enabling the manufac-
ture of low-volume products and increased design
flexibility [2, 3].
The evolution of RP for the production of end-
use parts is termed rapid manufacturing (RM) [4].
The basis of RM lies in the direct production of
components from a three-dimensional computer-
aided design (3D-CAD) model. The model is
‘digitally’ sliced into a distinct number of layers
and these layers are reconstructed into a physical
form by the RM machine [5]. The main feature
of current machines, althoug h their modus
operandi differs greatly, is that they are able to
produce virtually any geometry without the
need for tools. RP processes include stereo-
lithography (SL), laser sintering (LS), fused depos-
ition modelling (FDM) and three-dimensional
printing (3DP) among others [6]. LS and SL systems
are currently the most widely used for RM
applications.
Though materials developments are still necessary
for the more widespread use of RM, other limitations
currently exist [7, 8], namely:
(a) process speed;
(b) dimensional accuracy;
(c) surface finish;
(d) repeatability.
All these problems are the subject of research on a
global scale, although many manufacturers are
today able to cope with the limitations of the current
systems to their advantage.
*Corresponding author: Wolfson School of Mechanical and
Manufacturing Engineering, Loughborough University, Leices-
tershire LE11 3TU, UK. email: M.Ruffo@lboro.ac.uk
JEM517 Ó IMechE 2006 Proc. IMechE Vol. 220 Part B: J. Engineering Manufacture
1417
An important consideration for the uptake of RM
is its cost effectiveness compared to classical pro-
duction methods, among which injection moulding
is significant for plastic products. In fact, if the future
of RM is to be competitive with traditional processes,
economic visibility will play a determining role,
assuming that those technical limitations discussed
earlier are overcome.
This paper aims to define and outline the thinking
behind a cost model for RM and further develops
this with a working model of a typical part being pro-
duced using an LS machine. The costing method
should provide a transparent costing of the part for
comparison with other manufacturing methods.
2 LITERATURE REVIEW
2.1 Costing in modern manufacturing
In a modern manufacturing environment, overhead
costs are growing as manufacturers promote levels
of automation and computerization, and thus, the
cost distortion of traditional cost systems is signifi-
cant [9]. For this reason, a gradual change of cost
models is necessary. From the literature, the possible
reasons for adopting new cost syst ems are [10]:
(a) traditional costing systems do not provide non-
financial information, useful for manager’s
decision making;
(b) product costing is inaccurate;
(c) costing systems should encourage improve-
ments;
(d) overhead costs are higher than labour costs.
The last point, in particular, is interesting for
automated technologies introduced in modern
industrial processes. In fact, the continuous increase
of automation and decrease of manual labour in
manufacturing processes changes the product cost,
thus increasing the importance of overheads.
2.2 Costing for RP and RM
Grimm [11] studied the hourly cost to run different
RP machines and compared the results. He devel-
oped the tests with three ‘typical’ part s, but the defi-
nition of ‘typical’ as representative for production is
arguable. The assumptions made by Grimm were
interesting (such as percentage of the envelope
capacity used, working hours, etc .), but suitable for
an RP environment, while in the case of RM the sce-
nario changes owing to simultaneous multiple parts
building.
An RM cost study was developed in 2003 by
Hopkinson and Dickens [12]. The authors calculated
the cost of a part assumi ng that the machine was
producing only copies of the same part and using a
constant production time. The ir model was used to
calculate a first approximation break-even analysis
with injection moulding (IM). LS manufacture was
compared against IM techniques in order to find
when RM wa s economically convenient. Figure 1 is
a typical example of the results of the study con-
ducted by the two authors.
The IM curve decreases because the initial cost of
the mould is amortized across the production
volume. The RM line is constant, supposing that all
indirect costs are charged on every single part, divid-
ing the total indirect cost for the number of parts
produced (i.e. machine depreciation in 8 years).
This model is a good approximation, but only valid
where the RM method is making (a) copies of the
same part; (b) relatively high production volumes.
Fig. 1 Example of break-even analysis comparing LS with injection
moulding (source: Hopkinson and Dickens [12])
1418 M Ruffo, C Tuck, and R Hague
Proc. IMechE Vol. 220 Part B: J. Engineering Manufacture JEM517 Ó IMechE 2006
The flexibility of additive techniques allows the
production of more than one part at a time. In addi-
tion, the parts in production can be different from
one another. For this reason it is possible to define
RM as a parallel process, where different parts can
be built contemporaneously. Also, if the production
regards only copies of the same part, the graph of
Fig. 1 is incorrect for lower production volumes. In
fact, just as the IM process has to amortize the initial
cost of the tool, the RM process needs to amortize
the investment of buying the machine. Therefore,
the RM production curve must have a deflection for
low-volume production, taking into consideration
the fixed time and cost described above.
It is the object of this study to find a relationship
between a part and its cost in the case of LS manu-
facturing. It follows a production anal ysis of copies
of the same part, which lead s to a model, valid for
both low- and high-volume production, expanding
on the existing model of Hopkinson and Dickens
[12]. A comparison between the old and the new
model is then presented, detailing the main
differences.
2.3 Cost-estimatio n techniques
There are some principal quantitative app roaches to
cost estimation for building the mathematical model
[13, 14].
1. Analogy-based techniques. These are based on the
concept of deriving an estimation from actual
information regarding similar real prod ucts.
2. Parametric models. Here, the cost is expressed
as an analytical function of a set of variables,
usually called cost-estimation relationships
(CERs) [15].
3. Engineering approaches. Here, the estimated cost
is calculated in a very analytical wa y as the sum
of its elementary components, constituted by the
value of the resources used in each step of the
production process. This approach can only be
used when the characteristics of the processes
are well defined.
4. There is also a different approach developed by
Cavalieri et al.[16]. They studied the possibility
of replacing a classic costing model with one
based on an artificial neural network. The results
obtained in a case study confirm the validity of
this innovative method, giving results similar
and sometimes better than classical approaches,
but with the limitation of a reduced possibility of
interpreting and modifying data.
The model presented in this paper is placed
between the parametric model and the engin-
eering approach, as the relationships found are
approximations based on statistics, although most
of the data are defined.
Besides the mathematical model approaches,
there are different methodologies to split costs to
different sections of the model. Durin g the study,
several costing methodologies were approached,
such as activity-based costing [17–20], total lifecycle
costing [21], target costing [16], and full costing [21].
The model presented in this paper has been formu-
lated in order to attribute the full cost of an RM
organization. This includes all costs of plant and
production, costs of administration, and costs of
the necessary overheads.
3 EXPERIMENTAL METHOD – MODEL
FORMULATION
The most common ste ps in cost modelling involve
the deter mination of [22]:
(a) the scope, i.e. costs are subdivided into different
types, which have to be modelled;
(b) the allocation base for (overhead) costs;
(c) the cost functions, i.e. the relationships between
product parameters and costs.
The methodology used in this paper is general and
open to any additive manufacturing technique,
although the particular case studied here regards
an LS machine, the 3D-Systems Vanguard [23].
3.1 Scope: activities involved with RM
Table 1 shows activities involved with RM and their
definitions. These activities were confirmed during
previous work by Wohlers and Grimm [24].
The diffe rent activity costs of Table 1 can be split
into two categories: direct and indirect costs. In the
model presented, only the activity ‘material’ was
considered to be a direct cost. Labour and machine
Table 1 Activities associated with RM
Activity Definition
Material Cost of material purchase
Software Cost of software purchase and
upgrades
Hardware PC purchase and upgrade cost
Capital equipment
depreciation
Depreciation cost of capital equipment
(i.e. LS machine)
Labour Labour cost for machine set-up and
any required post-processing
(introduced in annual salary)
Maintenance Capital equipment maintenance
costs per annum
Production overhead Costs incurred due to production,
energy, and floor space
Administration
overhead
Costs incurred due to running the
enterprise, administrative staff,
office space, and consumables
Cost estimation for rapid manufacturing 1419
JEM517 Ó IMechE 2006 Proc. IMechE Vol. 220 Part B: J. Engineering Manufacture
maintenance, which could be seen as direct costs,
were allocated indirectly as they are annual payees
with regular contracts. Moreover, it was supposed
that the technician is working full-time only on RM,
setting up machines and cleaning parts; this is a con-
servative model because the entire salary is allocated
to RM production instead of supposing the operator
is wor king on different tasks.
3.2 Data included in the model
The cost ing data collected and used in the model
are typical for an RP department. An important
assumption was made about the productivity of the
LS machine, which was est imated to work 100
hours/week for 50 weeks/year (utilization of 57 per-
cent). Contact with industrial partners on the DTI-
funded Foresight Vehicle Project, under which this
work was undertaken [25], confirmed the difficulty
of increasing the utilization over 60 per cent. The
indirect activities of Table 1 can be summarized
into four categories. Table 2 shows the categories
with the associated cost/build hour.
Table 3 includes a detailed breakdown of the indir-
ect costs used in the model.
Costs quoted are in line with those of the project’s
industrial partners [25]. The only direct cost used in
this model was the material purchase; Duraform PA
[26] is the material selected for the case study and
sold at around e58 per kg (UK 2005).
3.3 Allocation base for costs
The machine purchase absorption and other indirect
costs were allocated to each individual product by
the time in which the machine takes to produce
them. Machine set-up and cleaning, warming up,
and cooling down phases imply times in which the
machine is not building layers. However, they must
be considered for cost allocation, as each new build
needs these fixed times (equivalent to a fixed cost).
A scheme of the entire conceptual model is shown
in Figure 2.
3.4 Cost estimation relationships (CERs)
The cost of a build (Cost
B
) is the sum of the indirect
cost associated with the time of building (t
B
) and
the direct cost associated with the material used dur-
ing manufacture (m
B
)
Cost
B
¼ Costðt
B
ÞþCostðm
B
Þð1Þ
where
Cost m
B
ðÞ¼
direct Cost
mass
unit
m
B
ð2Þ
Cost t
B
ðÞ¼
P
indirect
Costs
working
time
t
B
ð3Þ
The time and material used during the build (t
B
and m
B
respectively) are the main variables of the
costing model. Time refers to how long the
machine works for the build; part mass (or volume)
is an index of the raw mat erial used.
3.4.1 Equations for material
The material used in this case was Duraform PA,
which is in a powder form. The material that ha s
not been sintered is, in theory, recyclable. However,
recycled powder has suffered from a thermal treat-
ment and its mechanical properties are modified
from the virgin state. Therefore, recycle is possible
but with limitations, and must never exceed 67 per
cent of the total, as stated in the material manual
[27]. Moreover, after a few recycles it is advisable to
discard the old powder, operating with virgin pow-
der once more.
Table 2 Main indirect, cost activities and hourly rate
Main activities Cost/h (e)
Production labour/machine hour 7.99
Machine costs 14.78
Production overhead 5.90
Administrative overhead 0.41
Table 3 Indirect costs details
Production overhead e Production labour e
Yearly rent rate (per m
2
) 130.5 Technician annual salary þ employer contributions 32 770 ( þ 22%)
Building area (m
2
) 246.5
Energy consumption/h 1.5 Machine costs e
Machine & breakout station purchase 362 500 þ 24 360
Administration overhead e Purchase cost/year
*
45 313 þ 3045
Hardware purchase 2175 Maintenance/year 21 750
Software purchase 2175 Software purchase 7250
Hardware cost/year
*
435 Hardware purchase 4350
Software cost/year
*
435 Software cost/year
*
1450
Consumables per year 1450 Cost of software upgrades/year 1450
Hardware cost/year
*
870
* Depreciation time for computer hardware and software is 5 years, for the RM machine purchase is 8 years
1420 M Ruffo, C Tuck, and R Hague
Proc. IMechE Vol. 220 Part B: J. Engineering Manufacture JEM517 Ó IMechE 2006
![Fig. 1 Example of break-even analysis comparing LS with injection moulding (source: Hopkinson and Dickens [12])](/figures/fig-1-example-of-break-even-analysis-comparing-ls-with-tzukszh1.webp)
