Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.

1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12

Digital Mock-up to Optimize the

Assembly of a Ship Fuel System
Angelo Corallo, Alessandro Margherita, Giampaolo Pascali

S.S. ISUFI, University of Salento, Via Monteroni s.n., Lecce, Italy

tel/fax: +39 0832 297922 +39 0832 297927
e-mail: (angelo.corallo, alessandro.margherita, giampaolo.pascali)@unisalento.it

Submitted: 23/12/2009
Accepted: 12/01/2010
Appeared: 16/01/2010
HyperSciences.Publisher

Abstract – Assembly activities represent a crucial phase of new product development processes. The
identification of potential assembly issues early in the design stage can thus allow companies to save time
and cost of successive manufacturing phases. In such context, this article analyses the potential of digital
mock-up (DMU) to enhance assembly feasibility and efficiency through specific design analysis. In
particular, it is here showed the development of a DMU process and its integration with design for
assembly (DFA) procedures at a large Italian company. The article illustrates the application to a ship
fuel system and discusses the main benefits achieved by the company in terms of reduced assembly time,
re-works, number of special tools used and incorrect assemblies.
Keywords: Assembly simulation, Computer-aided design, Design for assembly, Digital mock-up, Digital
pre-assembly, Fuel system.
1. INTRODUCTION structure, to reduce manufacture and assembly costs and
quantifying the improvements.
Companies are increasingly challenged to produce in shorter
time, with higher quality, lower costs, and enhanced capacity In the ‘design for x’ family, design for assembly (DFA) is a
to satisfy customer requirements. In such endeavour, particularly relevant process/approach by which product
concurrent engineering methods have been introduced as a assembly issues (mostly number of parts, their insertion and
more integrated approach to develop high quality products orientation) are addressed in the early design phase, with the
and bringing them to competitive global markets at lower goal to reduce overall assembly cost, time and complexity.
price and in significantly less time (Hyeon, Parsaei and Among company processes, the activities for assembling
Wong, 1991). individual parts to obtain the final product are crucial as they
use over 50% of total production time, with costs varying
In the concurrent engineering perspective, the design is made from 20% to 40% for a single unit. The use of DFA can
taking into consideration all the ‘disciplines’ involved in the improve significantly such performance and several methods
product lifecycle. Companies and researchers focus today have been developed at this purpose like the Hitachi
their attention on fields like ‘design for manufacturing’ (i.e. Assembly Evaluation Method (Miyakawa and Ohashi, 1986),
designing products in a way to reduce production costs), the Boothroyd-Dewhurst method (Boothroyd and Dewhurst,
‘design for assembly’ (simplify product structure to reduce 1986) and the Lucas method (Holbrook and Sackett, 1988).
assembly costs), ‘design for maintainability’ (product designs
which reduce maintenance costs and efforts) but also ‘design These techniques are evaluative methods of assembly
for recyclability’ and ‘design for life-cycle’. These feasibility early in the design process. They use their own
approaches are sometimes referred to as ‘design for x’ (Kuo, synthetic data to provide guidelines and metrics to improve
Huang and Zhang, 2001). the design in its ability to be assembled. All three can analyse
a product for manual assembly, and Lucas and Boothroyd-
A key concern is to achieve integration among single Dewhurst can also analyse automated assembly. Boothroyd-
disciplines and concurrent engineering issues to avoid the Dewhurst goes one step further by considering different types
case where a product (re)design is good from the perspective of automated assembly such as robotic and high speed
of assembly but difficult, time consuming or expensive to dedicated.
manufacture or test (Wu and O’Grady, 1999). At this
purpose, the approach to ‘design for manufacture and The research on DFA is based on the premise that the lowest
assembly’ (Boothroyd, Knight and Dewhurst, 2001) has been assembly cost can be achieved by designing a product in such
introduced as basis for concurrent engineering studies to a way that it can be economically assembled by the most
provide guidance for design teams in simplifying the product appropriate assembly system. A handbook has been proposed

Copyright © 2010 HyperSciences_Publisher. All rights reserved 4 www.hypersciences.org

 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12
to define constraints imposed by design features of product
over assembly processes (Boothroyd and Dewhurst, 1986).
Besides reducing costs, DFA allows additional benefits in
terms of increased quality and reliability, and shorter
manufacturing time. The approach shortens the product cycle
and ensures a smoother transition from prototype to
production. DFA applications have been made in different
sectors, from large-scale industrial systems such as
automotive to compact medical electronics products (Khan,
2008).
The developments in computer technology offer significant
potentialities to dramatically improve assembly performance
and new contributions are needed to identify proper
methodologies, process models and software tools which
allow to integrate digital technologies within product design
and development practices. In such context, this paper aims
to show the definition of a digital pre-assembly process and a
pilot application for the assembly of a ship fuel system at a
large Italian company. The virtual verifications made and a
high-level integration of the DFA procedure with the virtual
verification is showed.
The rest of the work is organized as follows. Section 2
reports the relevant background of the study; section 3
describes the company context and the process model defined
whereas section 4 shows the supporting technology
application and the test-case performed. Section 5 discusses
the main results achieved and section 6 draws some
conclusions and opportunities for next research.
2. DIGITAL PRE-ASSEMBLY
In the last years, there has been a great development of digital
technologies to verify product lifecycle processes and reduce
the need (and thus the costs) for physical prototypes.
Information technologies can contribute not only to
efficiency improvements but also to improved hypothesis
creation capabilities in engineers and organizations through
technical features such as full visualization, digital pre-
assembly and simulation (Baba and Nobeoka, 1998).
In particular, digital mock-up (DMU) allows designers to
investigate the assembly feasibility of a product and the
constraints imposed by manufacturing processes. DMU
allows the user to represent the structure of a product and the
accurate position of its geometry, and enables a
multidisciplinary presentation of assembly processes and
analysis (design ‘in context’) such as insertion, view and
collision.
Through DMU, it is possible to obtain a virtual representation
of a product and simulate the shape, function and spatial
positioning of its components or subsystems, as well as of the
necessary production tools. By providing the basic
representation of a product, DMU permits to share the core
product data that the different company areas and disciplines
use to collaborate.
5
With a larger meaning, DMU is also referred to tools,
applications and methods for the creation, visualization,
manipulation and verification of a virtual prototype. DMU
allows to realize virtual verification of the interaction
between the product and the external environment, using
techniques of sectioning and interference analysis, both
‘hard’ (e.g. crash) and ‘soft’ (e.g. short distance passages).
Verification can be made statically or with movement
simulation (e.g. opening of landing gear of an aircraft).
DMU has been extensively applied in different industries
such as automotive (Rooks, 1998) and aerospace. Ford Motor
Company is today strongly involved in the adoption of virtual
manufacturing tools and processes designed to catch possible
manufacture concerns by simulating automotive
performance. During the prototypal phase of the 2009
vehicles Ford Flex, Lincoln MKS and Ford F-150, the
program teams were able to reduce more than 80 percent of
potential manufacturing concerns by running more than
10,000 advanced digital pre-assembly engineering checks. In
addition, engineers were able to reduce design and production
tooling issues by 50 percent (www.ford.com). An optimized
product development process based on the use of DMU and
rapid prototyping has been defined in the automotive with the
aim to fulfil goals of time and cost reduction, and quality
improvement (Döllner, Kellner and Tegel, 2000).
In the aerospace industry, the experience made with the
Boeing 777 showed how the company had far fewer
assembly and systems problems compared to previous
airplane programs (Abarbanel, 1996) thanks to digital pre-
assembly processes defined for checking designs. Boeing
engineers were able to see up to 1,500 models (15,000 solids)
in 3D view.
Shipbuilding is another interesting field for applying virtual
assembly features as it requires a sophisticated product
information model to achieve the seamless flow of product
information. A DMU system that builds a prototype in a
computer has been proposed for consistent quality control.
The system can simulate models and assemblies on-the-fly as
well as project real-world manufacturability without the
expense and time required to make a physical mock-up (Won
Don, Jong-Ho and Ju Yong, 2007).
Virtual product creation and product design analysis are
essential in the investigation of assembly feasibility and help
reveal manufacturing constraints. An application of a virtual
assembly for a grill plate has examined the factors that
influence the efficiency of assembly, i.e. assembly method,
assembly sequence, and assembly time (Choi, Chan and
Yuen, 2002).
To achieve high concurrence for product design at the
conceptual stage, core concepts for the modelling of the
products in 3-D CAD system have been introduced such as
the solid model involving various kinds of design
information, the functional block with the design information
for conceptual design, the evaluation engine for the
concurrence, and the interpretation of functional block as an
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12
object/agent in knowledge processing (Ishikawa, Yuki and
Miyazaki, 2000). Other contributions have introduced a
graph-type representation of the product structure for use in
an integrated Computer Aided Assembly Process Planning
(CAAPP)/DFA system (Molloy, Yang, Browne and Davies,
1991) and a prototype technology system to provide a
collaborative environment for distributed experts to analyze
and optimize the assembly design (Nan et al. 2008).
Digital pre-assembly (DPA) is a product design verification
process which evolves in parallel to product development,
from concept definition through physical prototyping. DPA
verifies product geometry (e.g. interfaces, interferences, and
collisions) and assesses its capacity to allow the execution of
operations required by the product lifecycle (e.g.
maintenance). The availability of a digital prototype offers
the opportunity to analyse the product by the perspective of
the operator involved in maintenance and assembly, in order
to identify and solve problems related to the accessibility to
parts, use of tools and movement of parts during insertion
and removal (e.g. collision analysis).
Being a parallel process respect to product development, the
DPA process supervises the verification activities on product
definitions which are characterized by an increasing level of
maturity. Such verifications are fundamental when product
definition is realized by distributed work teams. In fact,
without the ‘check-points’ enabled by the DPA, the only
possible verification would be based on the realization of a
physical prototype, with potential issues related to the
geometric definition used during the realization, the revision
activities and related re-works and loops, and the resulting
delays and increased costs. A further contribution of DPA
can be found in the definition of tools needed to produce and
manage the product as a consequence of the availability of a
virtual prototype.
A relevant consideration has to be made related to the
integration of assembly data and design data. In fact, keeping
a DPA repository separate from a design repository can
introduce issues in terms of synchronization among data. As
a consequence, DPA operators cannot be sure they are
looking at the most updated design. Other issues may be
related to notification problems from the DPA environment
(DMU) to the design environment (CAD). For those reasons,
DPA and CAD environments need to be integrated within a
unique Product Lifecycle Management (PLM) system in
which design data updates are automatically notified to the
DPA operators who can proceed to further DPA analysis. The
results of DPA analysis (e.g. clearance analyses) are also
notified to designers in order to inform them about what
should be changed and why.
3. DIGITAL ASSEMBLY PROCESS AT AVIO
3.1 Context and requirements
The research described here is framed within the ‘Design for
Six Sigma (DFSS)’ project activities carried on by a big
Italian company, i.e. AVIO s.p.a.. AVIO is a world leader in
6
the design, production and maintenance of aerospace
propulsion subsystems and components. The company
participates in military programs such as the F-35 JSF and
civil partnerships with companies like General Electric, Pratt
& Whitney, Rolls Royce, and Volvo. Besides, AVIO is
operating in the marine industry for the provision of the
package for the propulsion system, which includes the gas
turbine module and subsidiary bodies of the engine such as
the system of lubrication oil and the fuel feed system.
Among DFSS projects in the marine field, a central activity is
related to FREMM (Fregata Europea Multi-Missione), a
military ship designed by Fincantieri and DCN to operate in
anti-submarine, anti-air, and anti-ship settings. In such
activity, the company looks for improving assembly
feasibility and optimization of the (transducer plate of the)
fuel system, with the aim to minimize overall assembly time
and costs while enhancing compliance and operators’ safety.
At this purpose, four metrics have been identified as target
for optimization: 1) overall assembly time (OAT); 2) number
of reworks due to non-feasible assembly (NoR); 3) number of
Special Tools used for Assembly (NoST); and 4) number of
non-correct assemblies (NCA).
Minimizing assembly time allows to reduce time-to-market
and this requires a strong integration between design and
technology components. Also the reduction of re-works is
based on this synergy since the choice of unsuitable tools or
assembly steps can determine relevant mistakes. Another
requirement is to minimize the costs deriving from the use of
special tools. The definition of the tool depends both on the
geometry of components and the operations to realize. Also
in this case, a strong integration of design and technology
departments is critical. Ensuring the safety of the operator
requires addressing assembly steps and procedures which can
impact on people safety (e.g. narrow space or risky position
to complete an operation). The choice of assembly operations
is determined by the geometry of the product. Finally, in
order to render the assembly process ‘unique’, parts should
be designed in a way to allow only one way of assembly. In
this manner, the mistake-proof requirement can be met.
These requirements have represented the core input for
defining the process model and the application which are
described hereafter.
3.2 The digital pre-assembly process
DPA is an iterative process which generates models whose
detail and complexity increase with the evolution of product
development. Each iteration starts with the import of the
geometric function of parts within the virtual prototyping
environment, i.e. the DMU. These geometric functions are
imported within the environment both for the static analysis
of interferences among parts and for the analysis of assembly
and maintenance feasibility. The DPA process, and thus the
representation of product, progresses through a set of steps
which are illustrated in Fig. 1.
The process starts with a ‘Project Definition’ activity in
which a simplified representation of the product is obtained
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12

to explicit the rough shape and dimension of each part. This
rough model defined is then shared by company teams in the
‘Initial Design Share’. The only verification possible on the
simplified model is ‘Systems Separation’, which consists in
the analysis of the spatial separation among the different
systems that build up the product, which are usually
developed by different design teams.
Fig. 1. Digital pre-assembly process in AVIO
The process starts with a ‘Project Definition’ activity in
which a simplified representation of the product is obtained
to explicit the rough shape and dimension of each part. This
rough model defined is then shared by company teams in the
‘Initial Design Share’. The only verification possible on the
simplified model is ‘Systems Separation’, which consists in
the analysis of the spatial separation among the different
systems that build up the product, which are usually
developed by different design teams.
volume needed for using a tool in assembly or de-assembly
operations); d) ‘visual path envelop’ (model of volume
needed to execute visual checks).
3.3 Integration DPA/ DFA and impact on NPD
The process defined to integrate the DPA process with the
overall DFA approach of the company is represented in Fig.
2. At the beginning, Design goes in parallel with DPA Static
Verifications (i.e. system separation and static interference
check). In fact, DPA makes the verifications required
according to the level of maturity of product definition. When
the Design delivers a complete model, the DFA phase can be
started and the efficiency of assembly e determined (the
Boothroyd-Dewhurst method is applied at this purpose). If
assembly efficiency is not acceptable, a simplification
proposal is made to the project structure in order to optimize
the design process. On the contrary, if efficiency is
acceptable, an optimized design for assembly is obtained.
The design of assembly procedures follows and a simulation
is run to validate or redefine them. In some cases, the
redefinition of assembly procedures is not sufficient and the
design team has to reconsider the whole design.
The whole process is carried on by a multi-disciplinary team
of techniciansand designers. The design defines the product
geometry suitable to optimize the achievement of customer
requirements. This geometry is defined based on assembly
requirements and methodology being adopted.

During the ‘Design Evolution’ (e.g. we move from model’s

shape to shape and interfaces, to shape and accurate
interfaces, to complete solid models) that goes from ‘Initial
Design Share’ to the ‘Complete Spatial Integration’ of
models, the number of verifications to be made grows. In
addition to the ‘System Separation’ check, it can be also
performed a ‘Static Interference Check’ to verify that product
components do not interfere with each other, and thus to
assess the risk of volumes’ overlap. The following step, i.e.
‘Assembly Check’, is aimed to verify the satisfaction of
volume requirements for assembly and de-assembly (i.e.
dynamic interferences) and the physical and visual
accessibility of parts. Finally, the ‘Maintenance Feasibility
Check’ verifies the accuracy of maintenance operations. In
this case, it is also possible to perform a dynamic interference
Fig. 2. Integrated DPA/DFA process in AVIO
analysis.

All the DPA checks described can be realized in the DMU
environment. DMU for DPA checks is obtained by coupling
the definition of single parts which made up the product
through their respective spatial relationships. The analysis of
overlaps among components allows to identify mistakes in
terms of geometric fit, positioning and orientation which can
prevent a correct coupling of parts. It is possible to define
some volumes for DPA checks in order to increase their
accuracy: a) ‘cinematic envelop’ (model of volume needed
for the movement of parts in the assembly); b) ‘removal
envelop’ (model of volume needed for the movement of parts
in the de-assembly); c) ‘tool operating envelop’ (model of
The introduction of DPA has impacted the new product
development (NPD) activities in AVIO. The NPD process
model adopted by the company is known as ‘phase & gate’, a
method (diffused worldwide) which divides the NPD process
in different sequential phases (i.e. concept, development,
launch, etc.) containing several ‘gates’ or check-points. At
each gate, only if expected results are obtained the process
can move to the next step (next gate of the same phase or
next phase). The differences between the approach to NPD
before and after the DPA can be summarized as follows.
Before DPA, company departments communicated only

7
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12

during official design reviews. At this purpose, each
department provided in advance the design review materials
to all other departments. During the design review, assembly
technologies, manufacturing technologies and product
support maintainability departments gave a feedback to the
design team in order to modify and evolve the design.
With the introduction of DPA, each department involved in
product development has access to a common set of product
model designs and DPA data and it can thus verify the design
from its own point of view. For example, product support and
maintenance departments can evaluate the maintenance
issues related to the product whereas the assembly
department evaluates the assembly feasibility, and so forth.
The process proposed can thus enhance product data share
and cross-disciplinary communication thanks to a fully
digital definition of the product in its lifecycle, along with the
use of common tools and data exchange methods. Ultimately,
the concurrent engineering approach implemented in AVIO is
strengthened and the number of design verifications increases
as it’s not limited to official design reviews. Product
verifications can be started once designers put their models in
a ‘shared status’ and hence make them available to the rest of
the organization.
Simplify option has been set to 0.4). Finally we have used a
lossless compression for the first LOD and a lossy
compression in the second LOD.
VisMockUp alternate hierarchies are generally used to
reorganize product assembly models according to the specific
design needs. We used alternate hierarchies in order to
represent the assembly steps. In particular, each component
of this alternate hierarchy corresponds to an assembly step.
An assembly step contains the component to be mounted at
this step, the fasteners used and a reference to the previous
assembly step. In Fig. 3, it is showed a representation of an
assembly step in which the transducer is being installed upon
the plate. The elements highlighted outside the plate are the
transducer to be mounted and the fasteners used (four screws
are visible). This approach was chosen to take into
consideration both the current element to be mounted and the
target assembly on which the element is mounted. Each step
defined is re-mapped in the VisMockUp animation
environment that is used to implement the path defined by the
DFA procedure.

4. APPLICATION OF DMU TO THE DPA PROCESS

The definition of the DPA process was accompanied by a

test-application performed with a purposeful tool, i.e. the
Teamcenter Visualization Mockup (Version 6.0.1.4)
produced by Siemens/UGS. Teamcenter Visualization
Mockup (from now on VisMockUp) is a real-time digital
prototyping solution that includes interactive 3D viewing and
robust advanced analysis of large product assemblies.
Fig. 3. An assembly step
Three-dimensional solid models of the components of the
FREMM fuel system transducer plate were created in UGS
The animation system within VisMockUp is organized by
NX 2.0. However, VisMockUp cannot access directly the
events which are generally organized in a sequence, unless
solid model data saved in the NX file format. For this reason,
there are changes in the interdependencies among the events.
the solid models were exported in JT file format using the
Each event contains a set of actions which are executed in a
translator provided by NX. We made modifications on the
concurrent or sequential way according to what has been
configuration of the NX translator (called PVTRANS) in
specified in the event timeline. In particular, VisMockUp
order to increase reliability in assembly and clearance
supports different types of actions which can be visualized to
simulation. We have used a PER_PART structure option to
make an animation. For instance, the liner path action is used
obtain a JT file for the assembly and a directory containing
to define a path that translates and/or rotates the selected
the JT parts.
parts along a path that is created in the 3D view. The camera
action is used to rotate, pan, and zoom the view of the model
JT file format is capable of storing an arbitrary number of whereas the snapshot action is used to rotate, pan, and zoom
faceted representations with varying levels of detail (LODs) the view and also to set the visibility and transformation of
We have defined two LODs and set the chordalOption to the parts. Finally, the text action is used to add text mark-up to
ABSOLUTE value in order to obtain a constant chordal value the view, either anchored to a part or unanchored.
in the two LODs defined. In this way, the maximum absolute
distance that a line segment may deviate from the smooth
In our application, the assembly simulation at each event
curve is approximately constant for each part and
coincides with an assembly step as defined in the assembly
independent from the part size. This distance is expressed in
hierarchy. Each event should thus contain at least linear path
the same unit of measure as the part. For the two LOD
actions which implement a real assembly path and a snapshot
defined we have set a chordal value of 0.2 and 0.5
action to establish the initial layout of the assembly step. In
millimetres. Furthermore the second LOD defined has forty
this way, it is possible to execute each assembly step
percent less or fewer polygons respect the first (because its

8
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12
separately from others. It is important to remark that the
layout of a given step is obtained from the end of the
execution of the previous step. As an option, an event can
contain text and camera actions to ease up the understanding
and visualization of the assembly simulation.
Another important action for motion simulation is the action
addressed to execute VFM files. These are files containing a
set of discrete positions of the part to be moved, without
interpolation among the points representing the position of
the part. There are several ways to obtain a VFM file, for
instance by applying transformation on the part and capturing
its position or by using an automated functionality provided
by VisMockUp (called “path planning”). This functionality
identifies extraction paths for parts or assemblies that need to
be removed from models for maintenance reasons. In
particular, this functionality generates a collision-free
extraction path in the form of a VFM file by specifying the
parts to be extracted and the parts to be avoided, and a set of
“key positions” of extracted parts. The key positions must be
at least two: a “start” position and an “end” position.
Intermediate positions can also be defined to constrain the
path direction. However, there are several limitations in the
use of VFM files. In particular, they cannot be edited once
defined and it’s not possible to simplify them through
interpolation. We have also used the “path planning”
functionality to discover if a de-assembly path exists for a
particular component and then to establish manually the
related assembly path though the use of the linear path action.
Once the initial layout of an assembly step has been defined
(snapshot action), it is possible to define the assembly paths
through the use of a linear path action. A path consists of a
series of control points (nodes) connected by segments. There
are several ways to add a control point to a path: a) by
picking points directly on the 3D model (a feature called pick
mode); b) by using a dedicated node creation panel; or c) by
selecting the component to be moved along the path and
applying directly on transformations.
Fig. 4. Manipulator and transformation window
The control of one point’s coordinates is obtained through the
use of a manipulator or transformation window (figure 4) in
which it is possible to precisely characterize the current
manipulator‘s position and orientation. Since the manipulator
and the part coincide, this also determines the part’s position
(x, y, z) and its spin angles (θx, θx, θz) which refer to a
reference coordinates’ system defined in the NX CAD
9
environment. VisMockUp doesn’t allow to estimate the path
duration time, which has been therefore deducted based on
the results of the DFA analysis.
Once an assembly path for the components is established,
they undergo a test for clearance analysis. Clearance analysis
identifies points of contact and/or penetration among parts or
groups of parts. We can define clearance as the distance
between a pair of items in which there is no physical
interference. VisMockUp allows to specify a clearance
distance. All the items below this distance (including contacts
and penetrating items) are identified and signalled to the user.
VisMockUp allows both static and dynamic clearance
detection. By setting the dynamic clearance analysis option,
when components are selected and the path associated with
them is executed, all the parts that are within a predefined
range from active parts are analyzed. For the clearance
analysis, we have imposed to the calculator to use NURBS
for the calculation of the points of contact in order to obtain a
better approximation respect to the JT tessellated data.
If a collision occurs, the colliding path has to be modified in
order to make it collision-free. There are some collisions
which are acceptable and that the engineers should be capable
to identify (e.g. a screw that collides with its screwed hole).
To make the assembly simulation more reliable, we also
developed the tools used in the assembly process. In
particular, the tools have been modelled in NX CAD
environment, according to the Italian UNI standard and using
NX parts families. For each type of tool, a part family was
developed to allow the user to obtain the necessary tool by
varying some typical parameters. Naturally, parameters are
selected by the user in a way to obtain tools available on the
market. Figure 5 shows the sketch of part families realized in
NX and which represents single open end wrench. The
process of importing from NX to VisMockUp has been
described in paragraph 3.1.
Fig. 5. Single open end wrench used in assembly simulation
Once an assembly step has been verified, it is possible to add
the assembly instructions that the operator must follow in
order to execute that step. We keep separated assembly path
verification from assembly text instruction. In fact, assembly
instructions are contained in a separated event executed in
parallel with its associated assembly event. An assembly
instruction contains the part number of components to be
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12

mounted, the fasteners’ part number, and the assembly tool to
be used. It is also possible to specify the tightening torque for
fasteners such as a nuts or bolts and any other technical
information. The complete assembly simulation, along with
assembly instructions, can be exported in a standard video
file format (e.g. avi, mpeg, etc.) and used for company
training purposes.
optimization made by the DFA phase contained in the entire
AVIO process. NoR and NoST have been equal to zero
whereas the NCA will be estimated at the test and
maintenance phases for each item being produced. The
research team is currently involved in the definition of a
metric related to the safety of operators involved in assembly
tasks.

5. DISCUSSION OF RESULTS A key element in assessing the benefits of technology-

enabled improvements in product design is considering how
time savings are used. Time savings can be used by
The DPA process and the customization of the DMU tool
management in different ways: a) reducing overall process
allowed to meet the initial requirements in terms of overall
costs by using fewer engineering hours during the design; or
assembly time (OAT), number of reworks (NoR), number of
b) re-investing time savings to design products with more
Special Tools used for Assembly (NoST) and number of non-
features or enhanced quality. In many company settings, the
correct assemblies (NCA).
second choice is more common. Investing automation-
enabled time savings into added features or quality
OAT can be minimized by reducing design parts through the enhancement can increase the benefits associated with
application of DPA. Besides, the learning curve of the
technology implementation.
assembly operator can be reduced thanks to the instructional
videos which are obtained as output of the DPA process.
At AVIO, the introduction of the DPA process and the
NoR can be also reduced since issues can be detected at early
minimization of physical prototypes have reduced the overall
design stages, due to the verification of assembly operations.
time-to-market and product development costs. As a result,
In case of non-feasible assembly, it is possible to re-design
engineers had more time to devote to critical design and
assembly procedures or the product itself. NoST
manufacturing activities, like increasing the exploration of
minimization is ensured by the use, in the DPA process, of
new product features and configurations. More time devoted
the tools (or their envelopes) used for assembly if available
to critical design activities, in turn, led to enhanced quality of
from some tool library or after their modelling in the CAD
the product developed. Facilitating the handoff among
environment. Concerning NCA, DPA procedures can bring
different engineering disciplines is crucial for manufacturing
sensible improvements allowing the designer to follow
firms to be competitive today. DMU can be a useful approach
specific guidelines. Besides, product parts are designed in a
in the challenge to build tighter interaction between design
way to minimize the possibility of human mistakes. The
and manufacturing departments. DMU can drive more
designer is asked to make symmetrical components to avoid
consistent product quality, increased efficiency through
orientation problems. In particular, symmetry is along the
reduced NPD time, easier checking and real-time availability
insertion axis. For asymmetric components, the asymmetry
of simulation data. Moreover, design errors resulting from the
has to be defined unambiguously. At the end of the design
inability to make verifications of all product configurations
process, the training video which is obtained from the DPA
and variants can be avoided.
process can provide additional support to the operator.
There are some limitations in the version of VisMockUp used
Besides efficiency and effectiveness objectives, the new
for the application in AVIO. For example, in the linear path
process introduced has also addressed the issues related to the
creation it is not possible to obtain a feature-based alignment
safety of assembly operator. At this purpose, an add-on of the
for the path nodes definition and it is not possible to realize a
VisMockUp tool, i.e. Vis Jack, provides a static manikin (it
simple screw hole alignment using a linear path pick mode
cannot simulate motion) to be placed in the mock-up
option. Indeed, it’s not possible to select the hole and the
environment in order to evaluate the space available to the
screw centres as inferred points. A possible workaround is to
operator for the assembly. However, there is not an automatic
include in CAD models some datum points in
way to evaluate if the manikin’s position is ergonomically
correspondence of the inferred points of interest, so that they
correct or not.
are present and selectable in the relative JT files. Datum
points insertion, however, would involve extra work time for
With specific reference to the FREMM fuel system, the designers who own CAD models. Another limitation is the
satisfaction of initial requirements has allowed to achieve
fact that flexible components are not supported since all the
important benefits in terms of reduced design time and costs. components are treated as rigid bodies. To address this
Based on interviews with AVIO managers and engineers, the problem, it could be supposed that if a path exists for a
research team has evaluated the benefits achieved in terms of
flexible component treated as a rigid body then it also exists
the four key metrics identified.
the path for the flexible component itself. However, this is
not always true and in particular for complex wiring systems
For the FREMM transducer plate, OAT has been equal to that need different assembly procedures respect to rigid
about 6 hours and this represents, as highlighted by AVIO components.
engineers, an excellent performance respect to previous
experiences made on assemblies of similar level of
Although DMU is a valuable tool, some limitations exist
complexity. This data is derived from the assembly time

10
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12
which suggest cautions in its use. DMU allows the user to
make geometric verifications of the product and its sub-
systems and is thus focused on all kind of simulations
concerning the geometric shape, kinematics or other design
aspects. However, simulations regarding product intrinsic
functionalities (e.g. performances) which are not related to
geometric envelopes or component spatial position are
excluded form DMU verifications. In this case, other tools
are needed.
Full-size, physical mock-up remain therefore an invaluable
tool as it lets engineers see exactly the behaviour of product
parts and help confirm clearances or take better design
decisions linked with assembly, maintenance, and servicing.
Moreover, physical mock-up can be a better tool for junior
designers as it gives them a much practical sense of how to
design product components.
6. CONCLUSION
This paper provided an integrated process and technology
perspective of product assembly optimization. Three points
are to be highlighted. First, the methodology defined supports
the explicit definition of knowledge used in the assembly
process; second, it was conceived as a methodology to
streamline knowledge sharing in a multi-disciplinary design
environment; third, it helps disseminate information about the
task and process knowledge to concerned users.
DMU has been applied to a ship fuel system, with assembly
sequence definition, simulation and verification. The virtual
verifications made during the design evolution process are
presented, along with a high-level integration of the DFA
procedure with the virtual verification process.
The method applied enables an early identification of
potential assembly issues and allows to optimize four critical
variables represented by the overall assembly time (reduced
to 6 hours), the number of reworks due to non-feasible
assembly (0), the number of special tools used for assembly
(0), and the number of non-correct assemblies (0).
Immersive technologies have been available for CAD and
visualization for decades. However, virtual reality is not yet
pervasive in industrial settings. DMU would benefit sensibly
from potentialities of virtual reality. An interesting evolution
of this research could be thus channelled towards the
integration, within the DPA environment, of 3D projection to
enhance the ability to visualize complex assembly scenes, or
towards developing new display technologies.
Moreover, next research will be dedicated to apply process-
based performance measurement methods to assess how the
new technology impacted assembly processes. A taxonomy
of key process metrics will support the evaluation of the
overall business impact of product assembly optimization.
The insights from this study can be re-used to extend the
application of the methodology to other programs within
AVIO and other companies as well. Besides, further research
will address the integration of the DPA analysis in the AVIO
11
Product Design Management (PDM) environment and the
creation of a system for distributing training videos (e.g.
through company intranet, streaming, podcast, etc.) to
assembly departments and specialized workers.
REFERENCES
Abarbanel, B. (1996). Boeing 777 - concurrent engineering
and digital pre-assembly. Proceedings of the 8th
Conference on Innovative Applications of Artificial
Intelligence, August 4-8, Invited talk 1589.
Baba, Y. and Nobeoka, K. (1998). Towards knowledge-based
product development: the 3-D CAD model of knowledge
creation. Research Policy, 26 (6), 643-659,
Boothroyd, G. and Dewhurst, P. (1986). Product design for
assembly. Boothroyd Dewhurst, Wakefield, RI.
Boothroyd, G., Knight, W. and Dewhurst, P. (2001). Product
design for manufacture and assembly. Dekker, New
York, NY.
Choi, A.C.K., Chan, D.S.K. and Yuen, A.M.F. (2002).
Application of virtual assembly tools for improving
product design. The International Journal of Advanced
Manufacturing Technology, 19 (5): 377-383
Döllner, G., Kellner, P. and Tegel, O. (2000). Digital mock-
up and rapid prototyping in automotive product
development. Journal of Integrated Design and Process
Science, 4 (1), 55-66.
Holbrook, A.E.K. and Sackett, P.J. (1988). Design for
assembly - guidelines for product design. Proceedings 9th
International Conference on Assembly Automation,
London, UK.
Hyeon, H-J., Parsaei, H-R. and Wong, J-P. (1991).
Concurrent engineering: the manufacturing philosophy
for the 90’s. Computers & Industrial Engineering, 21 (1-
4), 35-39.
Ishikawa, H., Yuki, H. and Miyazaki, S. (2000) Design
information modeling in 3-D CAD system for concurrent
engineering and its application to evaluation of
assemblability/disassemblability in conceptual design.
Concurrent Enginering 8 (1), 24-31.
Khan, Z. (2008). Design for assembly. Assembly Automation,
28 (3), 200-206.
Kuo, T.C., Huang, S.H. and Zhang, H-C. (2001). Design for
manufacture and design for ‘x’: concepts, applications,
and perspectives. Computers & Industrial Engineering,
41: 241-260.
Miyakawa, S. and Ohashi, T. (1986). The Hitachi assembly
evaluation method (AEM). Proceedings of the
International Conference on Product Design for
Assembly, Newport, RI.
Molloy, E., Yang, H., Browne, J. and Davies, B.J. (1991).
Design for Assembly within concurrent engineering.
CIRP Annals -Manufacturing Technology 40 (1), 107-
110.
Nan, F-Q. et al. (2008). Research on collaborative technology
of networked collaborative product digital pre-assembly
system. Journal of System Simulation, 20 (23), 6418-
6423.
Rooks, B. (1998). Shorter product development time with
digital mock-up. Assembly automation, 18 (1), 34-38.
 Journal of Modelling and Simulation of Systems (Vol.1-2010/Iss.1)
Corallo et al. / Digital Mock-up to Optimize the Assembly ... / pp. 4-12

Won Don, K., Jong-Ho, N. and Ju yong, P. (2007). A digital

mock-up system for construction of product information
model in shipbuilding process. Journal of Ship
Production, 23 (1), 7-16.
Wu, T. and O’Grady, P. (1999). A concurrent engineering
approach to design for assembly. Concurrent
Engineering, 7 (3), 231-243.

AUTHORS PROFILE

Angelo Corallo is assistant professor and researcher at

eBMS - S. S. ISUFI of the University of Salento. His
research and teaching activities are focused on technology
and organizational strategies in complex industries. He is
involved in activities related to knowledge management and
collaborative working environments in project and process-
based organizations, with specific reference to the aerospace
industry and interest towards languages, methodologies and
technologies for knowledge modeling. He has coordinated
efforts in the EU research programs for DBE (Digital
Business Ecosystem) and OPAALS (Open Philosophies for
Associative Autopoietic Digital Ecosystems).

Alessandro Margherita is assistant professor and researcher

at eBMS - S. S. ISUFI of the University of Salento. He
received a Ph.D. in “e-Business Management” in 2007 and in
2006 he was visiting at the MIT Sloan Center for Digital
Business. His research and teaching activities are
characterized by a cross-disciplinary focus with major
interest in the fields of IT-enabled process innovation, new
product development and business engineering. He’s
involved as teacher in international Masters’ and Ph.D.
programs at S.S. ISUFI and in corporate Master’s programs
such as Finmeccanica FHINK.

Giampaolo Pascali is a research fellow at eBMS - S. S.

ISUFI of the University of Salento eBMS of Scuola
Superiore ISUFI. He received a cum-laude degree in
“Computer Science Engineering” from the University of
Salento. He is currently involved in different projects, in
collaboration with an Italian aerospace company, with the
purpose to re-engineer CAD/CAM/CAE processes. His
activities are mainly focused on the customization of CAx
(Computer Aided tools) and PLM (Product Lifecycle
Management) platforms for improving different phases of the
New Product Development process.

12