This project has significantly improved
the accuracy of laser stereolithography, a key rapid prototyping technique.
It has also succeeded in advancing the integration of rapid prototyping
into the development cycle.
A new data standard offering better links to CAD (computer-aided design)
tools has improved stability and automation. The selection of appropriate
modelling techniques has been supported by new front-end software.
Understanding and take-up of rapid prototyping by development engineers
has been promoted by computer-based learning materials.
With the involvement of three car manufacturers, as well as industrial
and academic specialists, the project focused on automotive applications,
but results are readily transferable to other industries.
The speed with which improvements make
the journey from drawing-board to catalogue or showroom is a critical
factor in the struggle to maintain competitive advantage in the
car industry and beyond.
Rapid prototyping will soon be seen as an indispensable weapon in
the manufacturer's armoury. By producing functional or design models
directly from computer-aided design (CAD) systems, rapid prototyping
techniques (RPT) can dramatically shorten the development cycle
for new parts or assemblies.
Yet the use of rapid prototyping remains low, at least in Europe.
Focusing specifically on its application within the car industry,
this project set out to identify and overcome both the technical
and the cultural obstacles to its widespread adoption.
The three-year project, initiated as a collaborative venture by
Fiat and BMW, began by analysing the currently available rapid prototyping
processes. Weaknesses were identified by assessing the processes
against the practical needs of the car industry, as represented
by the partners themselves.
The partners found that the techniques available at the time were
too slow and too unstable. They often failed, needed active supervision
and did not produce sufficiently accurate models. In particular,
engineers at Fiat, BMW and Mercedes-Benz needed laser stereolithography
tools which were robust and fully automated. They wanted to be able
to start the prototyping process as they left their offices in the
evening, confident that an accurate model would be waiting for them
when they returned in the morning.
Layer by layer
Laser stereolithography builds up a solid model by illuminating
successive layers of liquid photopolymer resin with a UV laser beam.
Each point touched by the beam hardens and is bonded to the layer
At the start of the project, however, the accuracy of stereolithographic
modelling was not good enough for regular use. When resin solidifies
it shrinks. As each new layer was hardened by the laser, it tried
to 'pull up' the layer below, and the accumulation of internal stresses
caused unacceptable deformation of the completed model.
The German company EOS, working with the Institute for Polymer Testing
and Polymer Science (IKP) at the University of Stuttgart, investigated
the potential for minimising deformation by adjusting the sequence
in which different areas of the resin layer were hardened.
Conventionally, the laser beam simply passed up and down the layer
in strips. The partners came up with an alternative strategy which
treats the layer as a grid of cells, typically two millimetres square.
In the first step, all these cells are illuminated while maintaining
a small gap between each, creating a layer of unconnected cells.
Next, these cells are connected to each other, resulting in a stable,
but less shrinking layer. After the part has been taken off the
platform, the liquid resin in the gaps is polymerised in the post-curing
The approach worked well. Combined with advances in resin quality
outside of this project, it has dramatically improved modelling
Another successful strategy is called skin&core, a technique that
divides the model into two parts - an outer skin and an interior
core. There are two applications using skin&core. On the one hand,
massive models can be built much faster by building a 2 mm skin
and filling the core with a truss-like structure. On the other,
it can be used for building light, hollow models which can be burned-out
in the investment casting process.
An new standard from CAD to RP
The second obstacle which the partners tackled was the format in
which data is passed from CAD software to rapid prototyping equipment.
A de facto standard, called STL, did exist. But this only described
the inner and outer surfaces of the model with triangles. Separate
software tools were needed to 'slice' the description into the layers
required for stereolithography.
At the same time, STL was not compatible with all CAD systems. BMW
itself was unable to generate STL files directly. Once more, intermediate
software was needed to manipulate the data, slowing down the process
and tending to compound inaccuracies.
The partners decided to cut through this complexity. BIBA, at the
University of Bremen, joined forces with EOS to develop a new data
format, and an accompanying set of software tools. Their Common
Layer Interface (CLI) format enables a CAD system to generate model
layers and pass them to a rapid prototyping tool in a single step.
Speed, ease of use, reliability, and accuracy are all improved.
For BMW, Fiat, and Mercedes-Benz, the CLI standard also makes it
much easier to exchange part geometries between design, development
and production processes. Designed as an open standard in order
to facilitate the development of new software by third parties,
the CLI format is already supported by a large number of CAD and
rapid prototyping systems.
Support and persuasion
The final but crucial obstacle to the adoption of rapid prototyping
techniques had been identified at the outset as a cultural one.
Stories of inaccuracy and instability persisted. Resistance to rapid
prototyping was widespread, and was in danger of becoming entrenched.
The partners would only reap the commercial rewards of the project's
technological developments when rapid prototyping techniques were
actually put to use. Now that their technical failings had been
addressed, it was essential to show development engineers exactly
how these techniques could help them.
First, the selection of appropriate modelling materials and technologies
had to be simplified. Choices depend upon the use to which the model
is to be put, but are often highly complex. Stereolithography, for
example, gives good surface quality, suitable for a single design
part which is to be finished and painted. Laser sintering, on the
other hand, uses a more stable material and produces a surface quality
that is adequate for a mock-up.
BIBA built expert system software with a front end designed to help
non-specialists to choose the right rapid prototyping process for
a particular task. Their 'Rapid Prototyping System Selector' package
matches modelling materials and build technologies to information
about the intended application of the model.
Secondly, engineers had to be persuaded that rapid prototyping could
really help them. The three car manufacturers produced a self-running
computer-based demonstration to explain the technologies, and their
uses and benefits, to engineers and others within their companies.
The effectiveness of rapid prototyping was amply demonstrated by
the production during the project of nearly 300 separate model parts
by BMW alone. It is still too early to tell how successful the partners
will be in promoting rapid prototyping internally, but they have
certainly put in place all the elements required for its full integration
into the car development cycle.
Both the project's technical developments and its integrative methodology
have excellent potential for transfer to industries such as aeronautics,
consumer electronics, and medical technology.
Transfer will be hastened by the project's products themselves.
EOS has incorporated the new illumination strategy into all its
laser stereolithography equipment. By simplifying and standardising
model description data handling routines, the CLI format, also supported
by all EOS equipment, will lower the entry cost for new users of
rapid prototyping techniques.