Guidelines for Process

Selection
The initial purpose of rapid prototyping technology was to create parts as a
means of visual and tactile communication. Since those early days of rapid
prototyping, the applications of additive manufacturing processes have
expanded considerably.

According to Wohlers and Associates , parts from AM machines are used

for a number of purposes, including:
• Visual aids

• Presentation models

• Functional models

• Fit and assembly

• Patterns for prototype tooling

• Patterns for metal castings

• Tooling components

• Direct digital/rapid manufacturing

AM processes, like all materials processing, are constrained by material
properties, speed, cost, and accuracy. The performance capabilities of
materials and machines lag behind conventional manufacturing technology
(e.g., injection molding machinery), although the lag is decreasing. Speed
and cost, in terms of time to market, are where AM technology
contributes, particularly for complex or customized geometries.
With the growth of AM, there is going to be increasing demand for
software that supports making decisions regarding which machines to use
and their capabilities and limitations for a specific part design. In particular,
software systems can help in the decision-making process for capital
investment of new technology, providing accurate estimates of cost and
time for quoting purposes, and assistance in process planning.
Selection Methods for a Part
Decision Theory

there are three elements of any decision

• Options—the items from which the decision maker is selecting

• Expectations—of possible outcomes for each option

• Preferences—how the decision maker values each outcome

Assume that the set of decision options is denoted as A = {A1, A2,. . .,An}.
In engineering applications, one can think of outcomes as the performance
of the options as measured by a set of evaluation criteria. More specifically,
in AM selection, an outcome might consist of the time, cost, and surface
finish of a part built using a certain AM process, while the AM process itself
is the option. Expectations of outcomes are modeled as functions of the
options, X = g(A), and may be modeled with associated uncertainties.
Preferences model the importance assigned to outcomes by the decision maker. For
example, a designer may prefer low cost and short turn-around times for a concept
model, while being willing to accept poor surface finish. In many ad hoc decision support
methods, preferences are modeled as weights or importances, which are represented as
scalars. Typically, weights are specified so that they sum to 1 (normalized). For simple
problems, the decision maker may just choose weights, while for more complex decisions,
more sophisticated methods are used, such as pair-wise comparison . In utility theory,
preferences are modeled as utility functions on the expectations. Expectations are then
modeled as expected utility. The best alternative is the one with the greatest expected
utility.
Decision Support Problem (DSP) Technique

The advantages of DSPs, compared with other decision formulations, are that they
provide a means for mathematically modeling design decisions involving multiple
objectives and supporting human judgment in designing systems . The formulation
and solution of DSPs facilitate several types of decisions, including:

Selection—the indication of preference, based on multiple attributes, for one among

several alternatives.

Compromise—the improvement of an alternative through modification.

Coupled and hierarchical—decisions that are linked together, such as selection–

selection, compromise–compromise, or selection–compromise.
The selection problems will be divided into two related selection
subproblems.

First, it is necessary to generate feasible alternatives, which, in this case,

include materials and processes.

Second, given those feasible alternatives, a quantification process is applied

that results in a rank-ordered list of alternatives.

The first subproblem is referred to as “Determining Feasibility,” while the

second is simply called “Selection.”
Approaches to Determining Feasibility

The problem of identifying suitable materials and AM machines with which to fabricate
a part is surprisingly complex. As noted previously, there are many possible applications
for an AM part. For each application, one should consider the suitability of available
materials, fabrication cost and time, surface finish and accuracy requirements, part size,
feature sizes, mechanical properties, resistance to chemicals, and other application-
specific considerations. To complicate matters, the number and capability of
commercial materials and machines continues to increase. So, in order to solve AM
machine and process chain selection problems, one must navigate the wide variety of
materials and machines, comparing one’s needs to their capabilities, while ensuring
that the most up-to-date information is available.
knowledge-based system for the generation, selection, and process planning of production
AM processes

Problem defintion

“To propose, from a detailed functional specification, different alternatives

of rapid manufacturing processes, which can be ordered and optimized
when considering a combination of different specification criteria (cost,
quality, delay, aspect, material, etc.).”

This approach utilized two reasoning methods, case- based and the
bottom-up generation of processes; the strengths of each compensated for
the other’s weaknesses.
The problem of determining process and material feasibility can be
represented by the Preliminary Selection Decision Support Problem (ps-
DSP) .This is a structured decision formulation and corresponds to a formal
decision method based on decision theory. Qualitative comparisons among
processes and materials, with respect to decision criteria, are sufficient to
identify feasible alternatives and eliminate infeasible ones. After more
quantitative information is known, more detailed evaluations of
alternatives can be made.
Preliminary selection decision support problem word formulation
Approaches to Selection

The standard Selection Decision Support Problem (s-DSP) has been applied to
many engineering problems and has recently been applied to AM selection . The
word formulation of the standard s-DSP is given in Fig. Note that the decision
options for AM selection are feasible material-process combinations.
Expectations are determined by rating the options against the attributes.
Preferences are modeled using simple importance values. Rank ordering of
options is determined using a weighted-sum expression of importance and
attribute ratings. An extension to include utility theory has recently been
accomplished.
Word formulation of the selection decision support problem
Selection Example

In this section, we present an example of a capital investment decision related to the application of metal
AM processes to the production manufacture of steel caster wheels. This selection problem is very similar
to a quotation problem, but includes a range of part dimensions, not single dimension values for one part.
In this scenario, the caster wheel manufacturer is attempting to select an AM machine that can be used
for production of its small custom orders. It is infeasible to stock all the combinations of wheels that they
want to offer, thus they need to be able to produce these quickly, while also keeping the price down for
the customer. The technologies under consideration are Direct Metal Deposition, Direct Metal Laser
Sintering, Electron Beam Melting, Laser Engineered Net Shaping, Selective Laser Melting, and Selective
Laser Sintering. A readily available stainless steel material (whatever was commercially available for the
process) was used for this example. The processes will be numbered randomly (Processes 1–6) for the
purposes of presentation, since this example was developed in the mid-2000s and, as a result, the data
are obsolete.
Before beginning the selection process, the uncertainty involved in the
customization process was considered. Since these caster wheels will be
customized, there is a degree of geometric uncertainty involved.
A model of a caster wheel is displayed in the Fig. 13.4a, while its main
dimensions are shown in Fig. 13.4b. In this example, we have decided to
only allow customization of certain features. Only standard 12 mm
diameter X 100 mm length bolts will be used for the inner bore, therefore,
these dimensions will be constrained. Customers will be allowed to
customize all other features of the caster wheel within allowable ranges for
this model wheel, as displayed in Table 13.1.
Model of steel caster wheel
Caster wheel dimensions
The alternative AM technologies will be evaluated based on 7 attributes
that span a typical range of requirements, as shown in the following
section. Scale type refers to the method used to quantify the attribute. For
example, ultimate tensile strength is a ratio scale, meaning that it is
represented by a real number, in this case with units of MPa. Geometric
complexity is an example of an interval scale, in this case with ratings
between 1 and 10, with 1 meaning the lowest complexity and 10 meaning
the greatest amount of complexity.
• Ultimate Tensile Strength (UTS): UTS is the maximum stress reached before a material fracture. Ratio scale [MPa].

• Rockwell Hardness C (Hard): Hardness is commonly defined as the resistance of a material to indentation. Ratio scale
[HRc].

• Density (Dens.): The density refers to the final density of the part after all processing steps. This density is proportional
to the amount of voids found at the surface. These voids cause a rough surface finish. Ratio scale [%].

• Detail Capability (DC): The detail capability is the smallest feature size the technology can make. Ratio scale [mm].

• Geometric complexity (GC): The geometric complexity is the ability of the technology to build complex parts. More
specifically, in this case, it is used to refer to the ability to produce overhangs. Interval scale (1–10).

• Build Time (Time): The build time refers to the time required to fabricate a part, not including post-processing steps.
Ratio scale [h].

• Part Cost (Cost): The part cost is the cost it takes to build one part with all costs included. These costs include
manufacturing cost, material cost, machine cost, and operation cost. Ratio scale [$].
In this example, we examine two weighting scenarios (relative importance
ratings). In Scenario 1, geometric complexity was most heavily weighted because
of the significant overhangs present in the build orientation of the casters. Build
time and part cost were also heavily weighted because of their importance to
the business structure surrounding customization of caster wheels. Because of
the environment of use of the caster wheels, UTS was also given a high
weighting. Detail capability was weighted least because of the lack of small,
detailed features in the geometry of the caster wheels. In Scenario 2, all
selection attributes were equally weighted.
Table 13.2 shows the results of the evaluation of the alternatives with
respect to the attributes. Weights for the two scenarios, called Relative
Importances, are included under the attribute names
Table 13.2 Rating alternatives with respect to attributes
On the basis of these ratings, the overall merit for each alternative can be
computed. Merit values for each scenario are given in Table 13.3, along with
their rankings. Note that slightly different rankings are evident from the
different scenarios. This indicates the importance of accurately capturing
decision maker’s preferences. Process 4 is the top ranking process in both
scenarios. However, the second choice could be Process 2, 3, or 6,
depending upon preferences. In cases like this, it is a good idea to run
additional scenarios in order to understand the trade-offs that are relevant.
Table 13.3 Merit values and rankings