Open Engineering Framework

Initial Development to Create a Framework for Open
Engineering
John M. Vogel
AE 8900 MAV - Special Problems, Spring 2011
Abstract
The purpose of this paper is to introduce a framework for effective open engineering, a concept
that brings open source philosophy to product design. Open design and crowd-sourcing have
become popular and successful techniques used by individuals and upstarts, but have not yet
been integrated with traditional design methods used by Original Equipment Manufacturers
(OEM). A description of an open design environment and review of a basic product design
process as it pertains to complex mechanical or aerospace systems developed at Georgia
Institute of Technology is presented. Also described are the interfacing tool, and benefits and
drawbacks of pursuing open engineering.
1
Aerospace Systems Design Laboratory
School of Aerospace Engineering, Georgia Institute of Technology

I. Introduction
lthough the concept of openly sharing information is far from new, open initiatives have
gained traction to create solutions for difficult and complex problems within the last two
decades. The advent of the Internet has enabled the exchange of great amounts of
information at a low cost to the far reaches of the globe. The Linux operating system (OS) and
Wikipedia free encyclopedia are just two examples of open source software and openly editable
reference that have had great success using the open philosophy. The challenge is how to
translate the open source philosophy into open design, where the products are physical objects
instead of immaterial, like software.
A
There have been small successes in the mechanical realm with open design. Websites
Makezine.com and Instructables.com, are full of user based content of do-it-yourself projects,
how-to’s, hacks, and mods. There is also MakerBot, an open hardware 3D printer, but no
significant strides have been made towards designing a complex system of systems needed to
design a helicopter or hypersonic aircraft. As of the writing of this paper, there is one group
testing the open engineering concept as part of a competition. Team FREDNET is an Open
Source and Open Participation competitor trying to win the Google Lunar X PRIZE, a
competition to “land a robot on the surface of the Moon, travel 500 meters over the lunar
surface, and send images and data back to the Earth.”1
The focus described here is to produce a framework so that classic product design
processes can interface effectively with the open environment to create open engineering.
II. Open Design
Open design has come to have the same philosophy as open source, but pertains to the
creation of material products as opposed to software. It is the “development of physical
products, machines and systems through the use of publicly shared design information.”2
Open
design, like open source, has been enabled by the success of the Internet, which removed
geographical constraints and allows globalized communication. Mass communication, in turn,
enables mass collaboration, “a form of collective action that occurs when large numbers of
people work independently on a single project.”3
Open design does not have a linear hierarchy
of personnel which makes roles and responsibilities flexible. Designs are freely copied,
modified, and redistributed without having to pay royalties or fees. Individuals contribute as few
or as many resources as they wish. The open environment discussed here does not have
logistical constraints such as computer server hosting, differences in communication language,
location, culture, or design platform. This will assume that every individual is enabled to create
value by having access to an Internet capable device with a web browser and can operate basic
web-based programs like a text editor and spreadsheet application.
III. Design Process
While each large aerospace OEM has its own specialized product design process, we
will look at the generalized product design process. The Georgia Institute of Technology teaches
the Integrated Product/Process Development (IPPD) decision making process as a general
2

design framework capable of handling complex problems while incorporating quality engineering
methods, system engineering methods, and aspects of the product life-cycle.4
A graphical
representation of the general system life cycle can be seen in Figure 1 below.
Figure 1. Generaleralized product life-cycle4
The life-cycle figure shows the cradle-to-grave stages of a product or process. This
discussion will move the product from the initial concept definition stage to the end of the
architectural design stage where the product is ready for detailed design.
The IPPD process is seen in Figure 2.
COMPUTERINTEGRATED ENVIRONMENT
PRODUCT DESIGN DRIVEN
PROCESS DESIGN DRIVEN
REQUIREMENTS
& FUNCTIONAL
ANALYSIS
SYSTEM DECOMPOSITION
&
FUNCTIONAL ALLOCATION
SYSTEM SYNTHESIS
THROUGH MDO
SYSTEM ANALYSIS
&
CONTROL
ESTABLISH
THE NEED
DEFINE THE PROBLEM
ESTABLISH
VALUE
GENERATE FEASIBLE
ALTERNATIVES
EVALUATE
ALTERNATIVE
7 M&P TOOLS AND
QUALITY FUNCTION
DEPLOYMENT (QFD)
ROBUST DESIGN
ASSESSMENT &
OPTIMIZATION
ONLINE QUALITY
ENGINEERING &
STATISTICAL
PROCESS
MAKE DECISION
SYSTEMS
ENGINEERING METHODS
QUALITY
ENGINEERING METHODS
TOPDOWN DESIGN
DECISION SUPPORT PROCESS
Figure 2. Georgia Institute of Technology’s generalized Integrated Product/Process Development4
The IPPD graph shows three pillars: quality engineering methods, systems engineering
methods, and a top-down design decision support process. The arrows indicate interaction
between the three pillars including how step outputs support or are inputs for other steps in the
process.
An example of how IPPD methodology can be integrated with existing tools to complete
the design process is shown in Figure 3.
3

Baseline 1st
Option 2nd
Option
Engine Type MFTF Mid-Tandem
Fan
Turbine Bypass
Fan 3 Stage 2 Stage No Fan
Combustor Conventional RQL LPP
Nozzle Conventional Conventional +
Acoustic Liner
Mixer Ejector
Nozzle
Aircraft
Technologies
None Circulation
Control
Hybrid Laminar
Flow Control
alt. concepts
criteria
HOWs
Morphological Matrix
Best
Alternative
Tech. Alternative
IdentificationQFD
MADMMADM
Weights
Pugh Evaluation Matrix
Subjective EvaluationSubjective Evaluation
(through expert opinion,(through expert opinion,
surveys, etc.)surveys, etc.)
Figure 3. Example concept selection process using IPPD4
This shows how information flows from tool to tool and will be the example design process
referenced in the following sections.
A. Concept Generation
There are two phases of design: concept generation and evaluation. We can see from
the first box under the top-down design decision support process of the IPPD that we must
Establish the Need. The need can come from a number of sources such as a Request for
Proposal (RFP) or a personal need. To Establish the Need, the Seven Management and
Planning Tools (7M&P) and Quality Function Deployment (QFD) based on quality engineering
methods and Requirements and Functional Analysis from system engineering tools are
employed.5
They turn requirements into the engineering problem definition. Not all the tools
need to be used each time a design is pursued and are subject to scrutiny of necessity.
1. Seven Management and Planning Tools
We begin by utilizing several of the 7M&P tools that help foster an understanding of the
problem, encourage creativity, and organize issues without special knowledge of the tools.
When a problem is not familiar to the group, the first step is to create an affinity diagram.
An affinity diagram begins as a brainstorming session where a goal question is posed to the
group and the group comes up with as many ideas as possible. Then the ideas are grouped into
over arching themes. This is the bottom-up approach.
A tree diagram is similar to the affinity diagram in that it uses the top-down approach.
The goal question is divided into over arching themes and the themes are decomposed into
smaller issues by asking each member to list requirements to achieve a particular goal
statement for each theme.
To get an indicator of which issues are the most important, the interrelationship digraph
is used. The issues are displayed and relationships are indicated by using arrows. A root cause
issue will have more originating arrows and a key indicator will have more arrows pointing to it.
The interrelationship digraph gives a good graphical indication of which ideas are the most
important.
4

2. QFD
At this point, there are a lot of notions and ideas of what is necessary to fulfill the
requirements but no engineering values. To further refine the problem and turn generalities into
a ranking of product attribute importance, the QFD is used. The QFD takes the requirements, or
the “what’s”, and product characteristics, the “how’s”, and ranks the characteristics by
significance based on importance to the customer and risk.
A generalized QFD is shown in Figure 4 and consists of a relationship matrix of problem
issues, or “what’s”, rows and solution issues, or “how’s”, columns.
Customer
Requirements
System
Product and
Process
Charactaristics
"Hows"
"Whats"
Competative
Assessment
Relationship
Matrix
Correlation
Matrix
Direction Of ImprovementCustomerRanking
Target Values
Absolute Importance
Affinity and
Tree
Diagrams
Interrelationship
Matrix
Priortization
Matrix
Risk Ranking
Strong Relationship
Medium Relationship
Weak Relationship
Figure 4. Generalized Quality Function Deployment5
A design process can consist of one or more QFDs by cascading the “how’s” of the
parent matrix to the “what’s” of the child matrix and repeating the QFD process for the child
matrix. The values or properties of the problem or solution issues are displayed in the
correlation matrix, target values, and competitive assessment. These are sometimes referred to
as “rooms” as the QFD is sometimes called the House of Quality.
The relationship matrix is the main room and the matrix elements contain a ranking of
the importance of the “what’s” to the “how’s”. For the initial QFD, these will be the customer
requirements and ways to achieve or measure the requirements, respectively. The importance
matrix is an absolute or relative ranking of the “what’s” and, in our example, is the customer
ranking of importance.
5

The element values represent either a weak, medium, or strong relationship of the
“what’s” to the “how’s”. The correlation matrix, sometimes referred to as the roof in the house of
quality, is the interrelationship of the “how’s” to each other and just below that, a set of arrows
indicates direction of improvement. Below the relationship matrix is the absolute and relative
importance. A summation of the “how’s” element values is the absolute importance and the
summation of the “how’s” element values multiplied by its corresponding importance value is the
relative importance. The competitive assessment gives a relative comparison of existing
systems to meet “what’s” and technical assessment gives a relative comparison of existing
systems on the “how’s”.
Target values are given below the relative importance. Competitive assessment,
technical assessment, and target values are gleaned from benchmarking from existing systems.
The risk ranking is listed below the target values and given a relative score of the difficultly in
achieving the target value.
The Pareto chart shown in Figure 5 is the graphical summary of the product attributes.
Figure 5. Example Pareto chart6
The product attribute importance values are in descending order left to right and are
represented by vertical bars. The line shows the cumulative distribution of the product attribute
importance values. The Pareto chart summarizes the most important engineering requirements
based on the customer wants and risks.
B. Establish Value
Once the problem has been defined, the next step in the IPPD process is to establish
value. This is done through the Overall Evaluation Criteria (OEC). The OEC is the correlation
between system effectiveness, or benefit, and cost to give a qualitative assessment of how well
a concept meets design requirements. An example OEC for military systems follows.4
     
 Cost
ityDependabiltyAvailabiliCapability
OEC
 

(1)
6

Here, the benefits are capability, availability, and dependability, and cost is monetary. The
coefficients, α, β, and γ, represent relative weights for each attribute and sum to unity. The
attributes can themselves be described by equations whose value is normalized against a
baseline value.
To create a set of feasible alternatives, the morphological matrix tool is used. An
example is shown in Figure 6.
Figure 6. Example morphological matrix6
The purpose is to identify possible new combinations for a system. First, perform a functional
decomposition of the product by listing the “what’s” of the QFD, in the first column of the matrix.
Identify all the possible ways in which the function might be satisfied and list them across the
columns as the alternatives. For example, one of the “what’s” in Figure 6 is “Engine” and the
possible alternatives listed to satisfy this function are turboshaft, turbo-diesel, hybrid, or electric.
C. Evaluation
The second phase of the design process and the next step in the IPPD process is the
evaluation of alternatives. There are a multitude of ways and processes to evaluate design
alternatives, so a comprehensive approach is not possible. It is up to the designer to explore the
possibilities and choose the best approach for the design, but there are tools to help make this
decision. For the example here, we will discuss the deterministic but subjective analysis of Pugh
diagrams and Multi-Attribute Decision Making (MADM) and touch on probabilistic analysis
methods that use modeling and simulation for Joint Probabilistic Decision Making (JPDM).
1. Pugh Matrix
7

The Pugh matrix is a relative comparison - positive, negative, or same - of each concept
to a reference or datum concept for each criterion for values in the matrix elements. Positive,
negative, and same evaluations are totaled for each concept and evaluated subjectively. A good
candidate design has many positives and few or no negatives and sames. If there is no clear
superior option, a new datum is chosen and the process is repeated. This is a qualitative
evaluation of the concepts.
2. TOPSIS
Multi-Attribute Decision Making (MADM) describes a family of evaluation tools used to
select a concept in an analytical, deterministic way when attributes are subjective and
numerous. One particular tool that is useful when you have information about the characteristic
values and the ranks concepts cardinally is the Technique for Ordered Preference by Similarity
to Ideal Solution (TOPSIS).
The first step of the TOPSIS technique is to gather engineering characteristic data on all
compared concepts and turn qualitative quantities into numerical representation. Next, all
values should be normalized and multiplied by a weighting factor - the relative risk from the
QFD. This will determine the optimal solution for each criterion and calculate the distance from
the optimum for each criterion for each concept. The best alternative has the shortest distance
to the ideal solution and is farthest away from the negative-ideal solution.
3. JPDM
JPDM combines multi-attribute decision making with probabilistic engineering
characteristic values and their uncertainty. The objective of JPDM is to create a visual
representation of the probability of concepts meeting two requirements criteria simultaneously
that are evaluated through an objective function. These functions are enabled by modeling,
simulation, and Monte Carlo. Models can include surrogate models like Response Surface
Equations (RSE), which are determined with (Design of Experiments) and higher fidelity
simulations. The outputs are cumulative distribution functions of concepts versus criteria and
joint probability density functions of two criteria.
IV. Open Engineering
The ideal open engineering framework would enable the open environment to interface
with the pragmatic approach of product design process. For this, we will envision a web based
portal that can be accessed through the Internet from a desktop or equivalent computer. This
portal should support established mass communication tools such as forums, editable reference
pages, instant messaging, email, teleconferencing, videoconferencing, and new design specific
tools based on the product design process. A project will host the portal where users can log on
to use these tools and to view and modify design documents. These documents would include,
but not be limited to, texts, spreadsheets, 3D models, analysis input and output files, as well as
a revision history of the documents. Ultimately, a user can take all the content and recreate the
project on another host to create branches from same project trunk.
8

The design process should be integrated from front to back. When changes occur to
numerical values or additions are made to a listing, these should propagate automatically as
necessary. This is the Integrated Design (ID) concept and should be incorporated when
possible.
The portal will follow the elements of the IPPD process: Establish the Need, Define the
Problem, Establish Value, Generate Alternatives, and Evaluate Alternatives.
A. Establish the Need
The 7M&P and QFD tools are already made with group interaction and input in mind, so
only minor adjustments are needed to adapt them to the mass collaborative environment.
Affinity, tree diagram, and interrelation digraph should be openly editable as reference material
but do not necessarily have outputs to be linked to the QFD, Pareto chart, or other tools
downstream.
B. Define the Problem
An example QFD was shown previously in Figure 4. Each room can easily be
represented in a spreadsheet. Google already maintains an openly editable spreadsheet in the
Google Docs suite. This tool, or something similar, can be recast as the tool to support mass
collaboration on the QFD. Users will be able to add, delete, or change relationship and
correlation strength in their respective matrices. Users can also change values for technical
assessment, customer assessment, and target values. Creation of new requirements should
automatically propagate to the Pareto chart and evaluation tools. Once values are input into the
QFD, the Pareto chart should update automatically with the relative risks and values in the
correct descending order, and be easily displayed or viewed.
C. Establish Value
Equations to calculate the OEC and the attributes in OEC will be openly editable to
users as well as the weights for each criterion. Criterion weight should be derived from the
importance values in the QFD causing these values to propagate initially, but is still left editable
for the users.
D. Feasible Alternatives
The Morphological Matrix can also be openly editable as a Google Doc style
spreadsheet. The components of functional decomposition come from the “how’s” of the QFD,
so the matrix can already be populated with values defined from work accomplished on the
QFD. Possible alternatives to satisfy the functions, the column values, should be openly
editable for users to add, delete, modify.
E. Evaluate Alternatives
To reiterate, evaluation methods are numerous and subjective. Here, we only touched on
three evaluation concepts, the Pugh matrix, TOPSIS, and JPDM. These tools range from little-
9

information opinion evaluation to lots-of-information, modeling, and simulation based evaluation
to give a broad sense of possibilities on how an open environment will interact during this stage.
1. Pugh Matrix
The Pugh matrix will be an openly editable spreadsheet where users can edit relative
comparisons of each concept versus the datum for each criterion. Users will be able to
comment on comparisons to create a conversation. Positive, negative, and same evaluations
versus datum can be totaled automatically for each concept. Several Pugh matrices will exist
with different concepts chosen as the datum.
2. TOPSIS
Users gather data on all concepts and populate an openly editable spreadsheet. The
spreadsheet is set up to do the necessary calculations: normalize values, multiply by the
weights imported from the QFD, determine the optimal solution per criteria, and calculate the
distance from the optimum for each criterion for each concept. A ranking of the concepts by
ordered distance from the ideal positive would be displayed.
3. JPDM
To evaluate concepts in the open environment using JPDM, users would define objective
functions for requirements criteria, parameterize concept alternatives from morphological matrix,
and create models to connect these physical parameters to the criteria. Objective functions can
be drawn from the OEC and modeling equations and concept parameters can be openly edited
in a spreadsheet. Multiple cross plots of 2D joint probability distribution plots and cumulative
distribution functions are displayed for evaluation.
V. Effective Design
There are several concepts built into this framework making it effective, some inherent in
the design process itself and others from leveraging the open community. The design process is
effective because it brings knowledge to the early stages of the process, has high transparency,
is integrated, and has a high degree of flexibility.
By bringing knowledge to the front of the decision making process, issues can be
avoided during later stages. Making changes late in the design process can be costly and cause
delays in product delivery as the changes have to be incorporated, meaning analysis has to be
redone. This was shown in the analysis of Japanese car manufacturers in the 1980’s as
compared to the American car manufacturers. The Japanese were making the majority of their
changes early in the design process when design flexibility is high and committed costs are low
as opposed to American manufactures that made changes late in the design which contributed
to delays and higher design costs. Front loaded knowledge during the design process reduces
overall design cycle time.
The IPPD based process inherently has high visibility due to a step based process and
utilization of visual tools. The high visibility allows stakeholders such as the customer, regulatory
body, manufacturing team, and others to be involved at any point and review all processes at
10

any level they feel is important. The process emphasizes teamwork and integration, and high
visibility enables issues to be identified quickly and early so that they may be worked out.
The process is independent of preconceived solutions and biases which makes it
effective for novel missions and novel concept solutions that have little to no historical data
available. People tend to rely on personal experience and known configurations because they
have worked in the previous efforts, but this directs away from novel solutions that may actually
be the most robust solution.
Integrated design, where values are linked between tools and changes propagate
appropriately, allows the process to calculate and create outputs for evaluation quickly. Design
games such as changing requirements weightings in the QFD or changing mission scenario can
be calculated near real-time to evaluate how robust a design concept is to possible variations.
Three tools for concept evaluation were discussed previously: the Pugh matrix, TOPSIS,
and JPDM. The Pugh matrix is very simple and requires the least amount of information about
the concepts.
Increasing in information but still simple to implement is the TOPSIS tool. It is analytical,
deterministic, and capable of comparing a large number of concepts quickly, when compared to
the Pugh matrix process, but it is highly subjective to the weights of the criterion. These weights
always depend on “who’s in the room” and the tools used to agree on the values or criteria can
be of a subjective nature, such as complexity or color, which has to be assigned a
corresponding numerical value.
The most complex tool to implement is JPDM. It is effective for evaluating large and
complex design space analytically. It brings the highest fidelity to the evaluation step but
requires increased resources and computational capabilities. Resources like access to and
understanding of complex simulation codes for creating RSEs can be difficult to obtain. There is
the possibility to incorporate other tools to maintain effectiveness and minimize project
computational resources. The JPDM tool can be created to be processed offline at the end
users computer or the project can set up a distributed computing network like the Berkeley
Open Infrastructure for Network Computing (BOINC) where programs like SETI@home run.
Summary plots would be uploaded to the project for viewing and commenting. Users would
have the flexibility to choose the evaluation process that fits their needs to the degree of fidelity
necessary.
Other factors that relate to engaging the open environment that make this tool effective
are speed of evaluation, comprehensive scope, and inclusiveness.
Evaluation speed is a disparity evident between software design and a more classic
complex machine design and is a key enabler to successful open systems operation. For
example, the Linux OS is free download and operational within a few minutes. Programmers
have the capability to change the source code, compile, reload, and evaluate changes within a
matter of minutes. Successful changes can be uploaded to a code repository for others to
evaluate and can be incorporated into future builds. Wikipedia is another example of a
successful product of open environment as previously mentioned. Anybody can be an editor
and has the ability to read and evaluate changes quickly in near real-time. The online
collaboration has collectively created a very successful encyclopedia and is as accurate as the
Encyclopedia Britannica in science articles.
11

The design process has the ability to comprehensively evaluate all possible
configurations. This way, users will not become disenfranchised by the fact that their favorite
concept was not evaluated. Another option is to copy the entire project which allows groups with
contrary opinions to create separate branches. This enables choice to the users which
increases engagement.
Open engineering also allows users to collaborate independent of background or
experience. The 7M&P tools are already built for this and the assumed open environment has
low entry requirements. The users have the ability to comment, email, and instant message to
foster an environment of open communication and enable conversation.
Up front knowledge, transparency, integration, flexibility, speed, comprehensive scope,
and inclusion are all qualities that enable the open engineering frame work to be effective.
VI. Reasons for Open Design
Open design engineering may not be suitable for every project. Below are some
reasons projects might benefit from open engineering as well as issues that need more
discussion.
The open engineering process provides for effective brainstorming during the
‘establishing the need’ process. Here, brainstorming users all have equal access and come from
a global community that brings in a diverse set of ideas. The point needs to be made that there
is no criticism and finger-pointing during this part of the process. Unfortunately, there are
drawbacks to mass collaboration during brainstorming because there is no inherent hierarchy.
Guides like a timeline and goals have to be in place for brainstorming to remain effective.
Without these guidelines, too many resources could be spent on the initial stages, to the point of
diminishing return.
Various voting techniques can be employed to resolve conflicts among users for
weighting values or equation form. With increased user base, the possibility of gaining
consensus for user input values decreases. Techniques such as Delphi, Analytical Hierarchical
Programming, or Utility Theory can be employed to capture the voice of the users. For instance,
normally it would be wise to heavily weight SME opinion on design parameter inputs as, by
definition, they are the experts. However, there is no hard and fast rule that the project
community has to or will lend gravity to SME chosen values. It is up the users to incorporate
SME chosen values, which can end up being based on human factors like credentials or
persuasive skills, which moves away from the pragmatic approach we were trying to achieve.
Open engineering has relatively low costs for the same processes at an OEM because
people volunteer their time and resources to be a part of the project. Users join for recreational
purposes as a hobby or sometimes there is the lure of a monetary prize, as is the case for Team
FREDNET and the Google Lunar X PRIZE. Other times, users have a need not being filled by
commercial entities and through mass communication, are able to find each other and work
together without pay to find a solution.
OEMs are usually adverse to change and less technically nimble than groups that spawn
out of an open environment. These groups can rapidly adapt to their surroundings, are high-risk,
and high-reward, all of which are seeds for creativity and innovation.
12

VII. Conclusion
A framework that will allow the open design community to use proven concept selection
techniques for complex system of systems as well as described reasons why this framework will
be effective has been laid out. The open environment brings low cost and highly motivated
talented people to a structured approach to decision making leveraging the process’ inherent
flexibility of information available, comprehensiveness of scope, and quick feedback to keep
people engaged. This framework gets the product ready for the detailed design phase in the
product life-cycle. Future work to be done includes how to refine the modeling and simulation
models used in JPDM evaluation in an open environment and how to accomplish other phases
in the product life-cycle leveraging open design. There are already tools for detailed design in
collaborative CAD modeling and open manufacturing enabled by rapid prototyping, but there is
still work to be done to establish open test and evaluation, and open product support to
complete Product Life-cycle Management (PLM) in an open environment.
13

References
1
“About the Prize,” Google Lunar X PRIZE, URL:http://www.googlelunarxprize.org/ [cited 22
April 2011]
2
”Open Design,” Wikipedia, The Free Encyclopedia, URL:
http://en.wikipedia.org/wiki/Open_design [cited 27 April 2011]
3
”Mass Collaboration,” Wikipedia, The Free Encyclopedia, URL:
http://en.wikipedia.org/wiki/Mass_collaboration [cited 27 April 2011]
4
Schrage, D., ”Aerospace Systems Engineering – Product Life-cycle Engineering (PLE)
supported with Product Life-cycle Management (PLM)”, AE6372 Class Notes. Fall 2007.
5
Schrage, D., “Defining The Problem Continued: Quality Function Deployment”, AE6372 Class
Notes. Fall 2007.
6
Cooper, T., Minor, J., Mosig, T., Narisetti, R.K., Vogel, J., ”Advanced VTOL Concept
“SMART – COPTER”, AE6372 Final Presentation. 14 December 2007.
14

Open Engineering Framework

More Related Content

What's hot (18)

Similar to Open Engineering Framework (20)

Open Engineering Framework