What kind of people make good systems engineers? Kossiakoff
and Sweet give their opinion:
p.24"A systems engineer should have a creative bent and must like
to solve practical problems... The following characteristics are commonly found in successful systems engineers. They
1. Enjoy learning new things and solving problems
2. Like a challenge
3. Are skeptical of unproven assertions
4. Are open minded to new ideas
5. Have a solid background in science and engineering
6. Have demonstrated technical achievement in a specialty area
7. Are knowledgeable in several engineering areas
8. Pick up new ideas and information quickly
9. Have good interpersonal and communications skills"
How does a systems engineer learn the necessary skills to function
well on the job? Other than classroom training there is an opportunity to learn on the job, often from professionals that
are the best in their field. A systems engineer should be able to figure out which principles in each engineering profession
are the most important. Knowledge of the various branches of engineering is a powerful tool in solving systems-related problems.
Additionally, the ability to perform approximate calculations (making estimates) is valuable in making sure that everything
is progressing smoothly towards completion.
p.25"a systems engineer does not need to spend a lifetime becoming
an expert in associated disciplines, but rather can accumulate a working knowledge of related fields through selected readings,
and more particularly, discussion with colleagues knowledgeable in each field. The important thing is to know which principles,
relationships, acronyms, and the like are important at the system level and which are details. The power of multidisciplinary
knowledge is so great that to a systems engineer the effort required to accumulate it is well worth the learning time.
The Power of Approximate Calculation
The practice of systems engineering requires another talent besides multidisciplinary knowledge.
The ability to carry out 'back of the envelope' calculations to obtain a 'sanity check' on the result of a complex calculation
or test is of inestimable value to the systems engineer... Most systems engineers have the ability, using first principles,
to apply basic relationships, such as the communications equation or other simple calculation, to derive an order of magnitude
result to serve as a check."
Systems engineering becomes important when building large complex machines. Such development efforts
usually have these characteristics:
p.50"A typical major system development exhibits the following characteristics:
- It is a complex effort
- It meets an important need
- It usually requires several years to complete
- It is made up of many interrelated tasks
- It involves several different disciplines
- It is usually performed by several organizations
- It has a specific schedule and budget"
New design approaches should usually be tested in some format before a commitment is made to build
them.
p.83"When a new design approach is undertaken it is unwise to wait until the design is fully implemented
before determining whether or not the approach is sound. Instead, testing should first be done on a theoretical or experimental
model of the design element, which can be created quickly and at minimum cost."
One stage of developing a new system involves looking at the possible design concepts and making early
decisions about which concepts are feasible and which are not.
p.135"The principal objective of the concept exploration phase, as defined here, is to convert the
operationally oriented view of the system derived in the needs analysis phase into an engineering-oriented view required in
the concept definition and subsequent phases of development. This conversion is necessary to provide an explicit and quantifiable
basis for selecting an acceptable functional and physical system concept, and then guiding its evolution into a physical model
of the system... As in the case of operational requirements, the derivation of system performance requirements must also simultaneously
consider system concepts that could meet them. However, to ensure that the performance requirements are sufficiently broad
to avoid unintentionally restricting the range of possible system configurations, it is necessary to conceive not one, but
to explore a variety of candidate concepts."
Kossiakoff and Sweet offer this advice when looking at alternative
design concepts:
p.145"Guidelines for Defining Alternative Concepts [section title]
As noted in the next section, conceiving new candidate approaches to satisfying a set of requirements is an inductive process,
and hence requires a leap of the imagination. For such a process it is helpful to postulate some guidelines for selecting
alternatives:
1. Start with the existing (predecessor) system as a baseline
2. Partition the system into its major subsystems
3. Postulate alternatives that replace one or more of the subsystems essential to the mission with
an advanced, less costly, or otherwise superior version
4. Vary the chosen subsystems [or superior system] singly or in combination
5. Consider modified architectures, if appropriate
6. Continue until you have a total of four to six meaningful alternatives."
Looking at the various design concepts before building a new system helps set system performance
requirements and ensures that a viable design architecture is not overlooked.
p.154"the objective of exploring implementation concepts is to consider a sufficient variety of
approaches to support the definition of a set of system performance requirements that are feasible of realization in practice
and do not inadvertently preclude the application of an otherwise desirable concept. To that end the exploration of system
concepts needs to be broadly based.
Alternative Implementation Concepts
The predecessor system, where one exists, forms one end of the spectrum to be explored. Given the
operational deficiencies of the predecessor system to meet projected needs, modifications to the current system concept should
first be explored with a view to eliminating these deficiencies. Such concepts have the advantage of being relatively easier
to assess from the standpoint of performance, development, risk, and cost than are radically different approaches... The other
end of the spectrum is represented by innovative technical approaches featuring advanced technology."
A model of the system is a useful way to think about operational problems and should be
based on key attributes and relationships. Perhaps this concept is also useful for building a computer chess program. What
key attributes and relationships exist among the pieces on a chess board? How would we model those? What approximations can
we make that preserve the essential features of these attributes and relationships but allow for fast calculation?
p.409"Another powerful strategy for dealing with decisions in the face of complexity and uncertainty
is the use of modeling throughout the system development process. In broad terms, modeling is used to focus on particular
key attributes of a complex system and to illuminate their behavior and relationships apart from less important system characteristics.
The objective is to reveal critical system issues by stripping away properties that are not immediately concerned with the
issue under consideration. Simulation is the modeling of dynamic behavior. Trade-off analysis is the modeling of decisions
among alternatives."
Approximate calculations can simplify complex relationships and help people to understand the functional
relationships among the various parts of a system. Perhaps this is also true for the pieces on a chessboard. If we can find
an approximate calculation for the positional pressure exerted by the chess pieces, we can use it to focus our search
for the best lines of play in a chess game.
p.419"Approximate calculations represent the use of mathematical models, which are abstract representations
of selected functional characteristics of the system element being studied. Such models capture the dominant variables that
determine the main features of the outcome, omitting higher-order effects that would unduly complicate the mathematics. Thus,
they facilitate the understanding of the primary functionality of the system element."
