Generating Design Alternatives.

1
EST 200, Evaluating Design Alternatives

• MEC

2
Contents

• Decision Making.
• Choosing the Best Alternative.
• Numerical Evaluation Matrix.
• Priority Checkmark Method.
• Best-of-Class Chart.
• Design Evaluation.

3
Decision Making

• A selection process.
• The best or most suitable course of action
  finalized from among several available
  alternatives.
• Intellectual and Goal Oriented.
• Involves imagination, reasoning, evaluation and
  judgement.
• Involves commitment of resources.

4
Effectiveness of decision making enhanced by
participation.
5

• Our job is to give the client, on time and on
  cost, not what he wants, but what he never
  dreamed he wanted and when he gets it, he
  recognises it as something he wanted all the
  time
• - Sir. Denys
  Louis Lasdun,
• Architect.

6
Choosing the Best Alternative

• Resources are limited eg time, money, expertise
  etc.
• Rarely are the resources available to fully
  develop more than one design scheme.
• Never mind all of our alternatives.
• Must choose the best alternative for further
  elaboration, testing, and evaluation.

7
Choosing the Best Alternative

• Limit the analysis to the clients most important
  objectives.
• Avoid drowning useful information in a sea of
  relatively unimportant data.
• Establish metrics with common sense of scale.
• Do not mistakenly over/underemphasize some
  results.

8
Choosing the Best Alternative

• Information must necessarily reflect a fair
  amount of subjectivity.
• Many of the metrics reflect qualitativenot
  measurable, quantitativeresults.
• Metrics should be thought of more as indicating a
  clear sense of direction than an algorithm or
  numerical solution.

9
Choosing the Best Alternative

• Check that each alternative satisfies all of the
  applicable constraints.
• Design alternatives that dont meet constraints
  considered infeasible.
• Three selection methods to link design
  alternatives to ordered unweighted design
  objectives.

10
Pareto Optimality

• Named after Vilfredo Pareto.
• Pareto improvement Resources can be re-allocated
  to make at least one person better off without
  making other individuals worse off.
• Pareto optimal/Pareto efficient Situation where
  no individual/preference criterion can be better
  off without making at least one
  individual/preference criterion worse off or
  without any loss thereof, means no scope for
  further Pareto improvement.

11
A, B - Pareto inefficient , Pareto improvement
since total output increases. C,D Pareto
Efficient, No improvement possible.
Upper Limit
12
Pareto Efficiency

• Implies resource allocation in the most
  economically efficient manner.
• Does not imply equality or fairness.
• Resources cannot be reallocated to make one
  individual better off without making at least one
  individual worse off.
• Economy in a Pareto optimum state when no
  economic changes can make one individual better
  off without making at least one other individual
  worse off.

13
Methods of Choosing an Alternative

• Three selection methods
• - numerical evaluation matrix.
• - priority checkmark method.
• - best of class chart.
• Ordered objectives cannot be scaled on a
  mathematically meaningful ruler.
• May bring order to judgments and assessments that
  are subjective at their root.

14
Numerical Evaluation Matrix

• Constraints (upper rows) and objectives (lower
  rows) in the left-hand column.
• Limit the number of decisive objectives to the
  top two or three.
• Difficult to mediate among more than two or three
  objectives at one time.
• Reflects the application of the metrics to the
  design alternatives.
• To see if one design is Pareto optimal - superior
  in one or more dimensions, and at least equal in
  all the others.

15
Numerical Evaluation Matrix

• Values can be used to work with the client (and
  perhaps users) to revisit the objectives.
• Client may change their mind about relative
  rankings to get a very strong winner.
• Metrics and associated testing procedures not to
  change with whoever is applying or making
  measurements for the metrics.

16
Numerical Evaluation Matrix- Juice Container
Design
17
Priority Checkmark Method

• A simpler, qualitative version of the numerical
  evaluation matrix.
• Easy to use, makes the setting of priorities
  simple, readily understood by clients and other
  parties.
• Ranks the objectives as high, medium, or low in
  priority.
• However, considerable information is lost that
  may be useful in differentiating between
  relatively close alternatives.

18
Priority Checkmark Method

• Objectives with high priority given three checks,
  those with medium priority given two checks,
  objectives with low priority given only one
  check.
• Design alternative that meets an objective in a
  satisfactory way marked with one or more checks.

19
Priority Checkmark Method

• Metric results assigned as 1 if they are awarded
  more than the target value (eg 70 points on a
  0100 scale), and as 0 if their award is less
  than the target value.
• Choice of a target value (threshold) very
  important.
• May lead to results that appear to be more
  disparate than they really are.

20
Priority Checkmark Method - Juice Container
Design
21
Best-of-Class Chart

• For each objective, assign scores to each design
  alternative.
• 1 for the alternative that meets that objective
  best, 2 for second-best and so on..
• Alternative that met the objective worst given a
  score equal to the number of alternatives being
  considered.

22
Best-of-Class Chart

• Two alternatives can be considered best,
  handled by splitting available rankings.
• Eg two firsts would each get a score of
  (12)/2 1.5, second and third would get
  (23)/2 2.5.
• Scores help to see if the design is Pareto
  optimal (best in all categories), or at least
  best in the most important (i.e., highest ranked)
  objectives.

23
Best-of-Class Chart- Juice Container Design
24
Best-of-Class Chart - Merits

• Allows us to evaluate alternatives with respect
  to the results for each metric.
• No binary yes/no decisions.
• Easy to implement and explain.
• Ranking methods allow for qualitative evaluations
  and judgments.
• Can be done by individual team members or by a
  design team as a whole.

25
Best-of-Class Chart - Merits

• Helpful if there are many alternatives to choose
  among from.
• Can be used if we want to narrow our consultative
  and thoughtful process to the top few.

26
Best-of-Class Chart - Drawbacks

• Encourages evaluation based on opinion rather
  than testing or actual metrics.
• Shows only the rankings, not the actual score.
• May lead to a moral hazard akin to that attached
  to priority checkmarks.
• Temptation to fudge the results or cook the books.

27
Best-of-Class Chart - Drawbacks

• May not provide information on whether two
  results are close or not.
• Eg we do not know if the first and second
  results are close or not, which could be
  important information.

28
Design Evaluation

• Design evaluation and selection demand careful,
  thoughtful judgment.
• Ordinal rankings of the objectives obtained using
  PCCs cannot be meaningfully scaled or weighted.
• Cannot simply sum the results.
• Use common sense when evaluating results.

29
Design Evaluation

• Metrics results for two alternative designs that
  are relatively close to be treated as effectively
  equal, unless there are other unevaluated
  strengths or weaknesses.
• Results should meet the expectations.
• No excuse for accepting results blindly and
  uncritically.

30
Design Evaluation

• If results do meet our expectations, ask whether
  evaluation was done fairly.
• Should not reinforce biases/preconceived ideas.
• Check whether the constraints used to eliminate
  designs are truly binding.

31
Thank You