| Step 1: Form Integrated Product Team | cntrl-y to open |
| Step 2: Negotiate Desirements | cntrl-x to open |
| Step 3: Generate Alternatives | cntrl-k to open |
| Step 4: Value Analysis | cntrl-v to open |
| Step 5: Document Results | cntrl-m to open |
Systems Engineering Tailored for Science and Technology (SETFST) is a structured approach to identifying optimal solutions to complex problems. Applied to research and development, SETFST contributes to sound systems engineering over the life cycle of those products that emerge from the effort. SynGenics developed the SETFST process to help ensure that customer requirements/desirements are understood and to support good decision-making in the absence of complete information at each stage—from conceptual design to critical design review. The strengths of the SETFST process include its conceptual simplicity and its analytical rigor. At its simplest, desirements are identified and alternatives that might satisfy those desirements are evaluated, rated, and compared. It is a robust process with multiple benefits.
SETFST is a collaborative effort. The process begins by creating a clear understanding
of the problem. Often, that requires looking at the project from several standpoints.
The first step is to form an Integrated Product Team (IPT). The knowledge
and expertise of team members will bring a number of perspectives to the problem,
creating a fuller, more well-rounded understanding.
Project teams work best
when they contain a relatively small number of Subject Matter Experts (SMEs)—e.g., engineers, scientists, regulators, logisticians, financial analysts, customers, and users. The team creates systematic and traceable inputs
that encourage “ what if” scenarios as they work through the definition of
desirements and perform Value Analysis.
Once the full IPT is in place, it begins capturing desirements—definition and documentation of what the
final product, technology, or system must do rather than how it can be done. Customers are identified,
initial desirements are defined, Key Performance Parameters (KPPs) and Evaluation Criteria (EC) are established.
The customers provide criteria related to performance, reliability, portability, schedule to completion, life-cycle cost,
maintainability, etc. Each criterion is documented with a description, a unit of measure, customers or applications to
which it pertains, and assumptions made. Details specified for each criterion for each customer include priority,
objective, desirability limit, desirability function (d-curve), and a weighting factor. Each criterion is assigned
a numerical weighting factor reflecting its importance relative to other desirements of the same type. Types are
weighted for each customer's view of that type's relative importance. The d-curve, with its objective and
desirability limit(s) defines one measure of "goodness" for a particular customer. Individual desirements are documented for every application or customer. A composite set of desirements is identified comprising
desirements for all customers, i.e., for the "Constructed Customer" . Exit criteria encompass the subset of the constructed set of desirements that define successful completion; they include Key Performance
Parameters (KPPs) and all other criteria that must be met.
An objective function is generated that incorporates all desirements. The evaluation and analysis
activities enable prediction of where, within the multidimensional solution space, desirements are met,
and provide pointers to the best solutions. This approach may be used to guide investment of program
resources to the candidate technologies most likely to provide an affordable solution—meeting customer's
needs at a price those customers are willing to pay.
The team identifies possible solutions (called alternatives) to satisfy the desirements.
These may be new technologies; combinations of existing materials, technologies, or
components; novel approaches; or other systems with potential to satisfy the set of desirements.
Identifying requirements—what the final product, technology, or system must do, is crucial.
The customer provides information such as: performance, reliability, portability, schedule,
life-cycle cost, and maintainability. Criteria are documented with description, unit of measure,
customer, priority, objective, threshold, desirability, and type. Requirements are documented
for each customer, and exit or solution criteria are determined.
Value Analysis evaluates the alternatives against the desirements.
The measures of merit are desirability and risk. Desirability is a measure of
" goodness". Risk is the probability of failure to exceed the desirability limit.
Theoretical models, simulations, expert opinion, statistically designed
experiments, or other quantitative assessment methods may be used to predict
how each alternative will fare with respect to a desirement. A worksheet
for each alternative evaluates it against every criterion of a given type.
Tabulated in the worksheet for each desirement are the predicted mean response
value of the desirement's unit of measure and some measure of spread of the
prediction interval for that alternative. The predicted mean value is
translated into a desirability value, di, using the desirability function
for desirement i. The di values are aggregated into a composite desirability
(Dtype) for the alternative with respect to all desirements of the given type.
A measure of risk is calculated for each alternative for each desirement and
aggregated to produce a risk measure for the alternative against all desirements
of a given type. Similarly, top-level desirability is calculated using Dtype values and weighting factors, while the overall risk measure is generated from
the lower-level probabilities of failure to meet desirability limits.
From these worksheets, a Value Scorecard is compiled for each customer—one scorecard for each type of desirement and one Affordability Scorecard
showing top-level results. Scorecard metrics include expected value, the
desirability value to which it maps, and risk. The Value Scorecard presents
desirability and risk for multiple alternatives for all desirements for a
single customer. Desirability and risk are displayed at three levels:
individual desirement, desirement type, and overall. The scorecard assists
in identifying risk drivers and technical challenges in meeting the desirements.
The Value Scorecard for a single customer is used to communicate with
that customer when reporting progress or raising issues. The scorecard
for the Constructed Customer supports program decisions such as
selecting a single technology alternative for further development.
Once the team has focused on a particular technology alternative,
the process moves to refining it to best meet the program objectives
and exit criteria. Levels of design variables are evaluated in
different combinations to produce response-value predictions
using statistically designed experiments and response-surface
methodology. Regression analysis or other quantitative methods
are performed to generate predictive functional relationships.
The Multiattribute Desirability Optimization Methodology applies
an optimization formula to establish values of design variables
that relate to the best-value, most robust solution.
The SETFST process documents the history of the work done to determine the project direction.
Applying the SETFST process early in a program permits the team to focus
on the few alternatives most likely to lead to successful product development.
The process can be efficiently and quickly revisited periodically to refine
the assessment or reevaluate alternatives in light of criteria that may have
changed due to customer involvement and insight gained into the realm of
the possible, a changing world, or as a result of knowledge derived through further research.
The SETFST process is an intense approach to a project. It involves
dedicated commitment on the part of the project leadership and
the team members, but the results are worth the time and effort spent.
Well-researched desirements and carefully analyzed alternatives
define the direction toward the best solution, so that the
project may move ahead without the delays inherent in less
well-defined research projects. Repeatedly, teams report
that they have developed better products in a shorter time using this process.