Chapter 7-Chapter Decision Making and Concept Selection
judgement
integration of a persons basic mental processes and ethical standards good judgement is clearly understanding the realities of the situation
Ratio Scale
interval scale in which 0 is used as anchor needed to establish meaningful weighting factors all operations allowed
Ordinal Scale
items are placed in rank order comparisons made whether two items are equal, greater than, or less than no addition or subtraction does not say how far apart can determine mode
what phase is evaluating and selecting a concept?
last step in conceptual design phase
utility
measure of preference order for a particular user
utility
measure of satisfaction (or preference order) that is associated with each outcome
weighted decision matrix
method of evaluating competing concepts by ranking the decision criteria with weighting factors and scoring the degree to which each design concept meets the criterion
analog models
models based on an analogy or similarity between different physical phenomenon EX. ordinary graph, process flow charts
control volume
models boundaries 1. finite 2. differential
maximax decision rule
should select the alternative that maximizes the maximum value of the outcomes; one with the smallest possible loss look at best outcome OPTIMISTIC approach
continuous media
solids or fluids assume that their medium transmitting a stress or flow does not contain holes
preference
statement of relative value in the eyes of the decision maker Subjective
objective
statement of which decision maker wants to achieve
Decision Under Conflict
states of nature are replaced by courses of action determined by an opponent who is trying to maximize his or her objective function
direct assignment
team decides how to assign 100 points between the different criterion only recommended for teams where there are many years of experience designing the same product line
evaluation
type of process in which alternatives are first appraised according to some standard 1. absolute criteria 2. go/no-go scenario 3. relative criteria, pugh decision matrix, analytic hierarchy 4. best concept
marginal utility
understanding what is gained from adding one more unit to what is already possessed Law of Diminishing Marginal Utility
Complete Criteria Comparison Matrix [C]
use 1-9 ratings 1/# if B is greater than A A=row B=column total each column
point scale
used with decision matrix simplest way to convert the values of different design criteria into a consistent set of values
pugh chart
useful method for identifying the most promising design concepts among alternatives generated relative comparison compares each concept relative to a reference (datum) concept and determines whether the concept in question is the same (S), less than (-), or greater than (+) the datum
Utility Theory
value preference utility marginal utility
objective tree
weighting factors determined by using a hierarchial objective tree some experience needed multiply (Q1)(Q11)(Q111)=Q111
Normalize Matrix [C] to give [NormC]
normalize matrix by dividing each box by the column total
decision point
on decision tree, square
transient (dynamic)
parameters change with time
defensive avoidance
procrastinate, shift responsibility, remain inattentive to corrective information
vigilance
search for relevant information, appraise carefully and unbiasedly before making a decision
hypervigilance
search frantically for immediate solution
similitude
the condition of physical response is similar 1. geometric 2. kinematic 3. dynamic
unconflicted change
uncritically adopt whichever courses of action is most strongly recommended
decision tree
a graphical and mathematical model for decision making under uncertainty probabilities of outcomes are known!!
Nominal Scale
a named category or identifer can only compare if they are the same or not EX. "thick or thin"
iconic model
a physical model that looks like the real thing but is scaled representation aka geometric representations
greater variability in decision variables is associated with
greater risk
Question: Using expected values only, which contract would the decision maker choose?
A: (100,000)(0.6)+(5,000)(0.1)-(40,000)(0.3)=$62,700 B: (60,000)(0.5)+(30,000)(0.3)-(10,000)(0.2)=$37,000 would choose A b/c maximizes utility
Analytic Hierarchy Process
AHP least arbitrary a problem solving methodology for making a choice among a set of alternatives when the selection criteria represent multiple objections, have a natural hierarchical structure, or consist of qualitative and quantitate measurements
Decision Theory
based on utility theory and probability theory 1. alternative courses of action 2. states of nature 3. outcome 4. objective 5. utility 6. states of knowledge
scale models
can be made cheaper and quicker similitude
model validation
checking to see that the model gives an accurate representation of the real world
model verification
checking to see that the model works as intended
Decision Under Certainty
-Has all necessary information to evaluate the outcome of their choices -Each action results in a known outcome which will occur with a probability of 1 -Just choose lowest values on loss table
Perform Consistency Check on [C]
1. Calculate weighted sum vector {Ws}=[C][W] in excel 2. Calculate consistency vector {Cons}={Ws}/{W} 3. Find λ as average values in {Cons} 4. Find Consistency Index (CI)=(λ-n)/(n-1) 5. Calculate Consistency Ratio (CR)=CI/RI -RI is Random Index value found on table 6. If CR<0.1, the {W} is valid
Steps to Make a Pugh Chart
1. Choose Criteria 2. Formulate Decision Matrix -Columns: concepts - Rows: criteria 3. Clarify the Design Concepts 4. Choose Datum 5. Complex the Matrix (+,-,S) 6. Evaluate Ratings 7. Establish new datum (highest rated) and rerun 8. Examine Selected Concept for Improvement Opportunities
AHP Process for Pairwise Comparison of Selection Criteria
1. Complete Criteria Comparison Matrix [C] 2. Normalize Matrix [C] to give [NormC] 3. Average Row Values of NormC to give Criteria Weights [W] 4. Perform Consistency Check on [C]
AHP Process for Pairwise Comparison of Design Alternatives
1. Complete Criteria Comparison Matrix [C] 2. Normalize Matrix [C] to give [NormC] 3. Average Row Values of NormC to give Priority Vector {P1} 4. Perform Consistency Check on [C]
Decision Making Models
1. Decision Under Certainty 2. Decision Under Uncertainty 3. Decision Under Risk 4. Decision Under Conflict
Design Selection Based on Absolute Criteria
1. Evaluation Based on Judgement of Functional Feasibility 2. Evaluation Based on Assessment of Technology Readiness 3. Evaluation based on Go/No-Go Screening of Constraints
To Determine Best of Design Alternatives
1. Final Rating Matrix [FRating] are the values from {P1} 2. Calculate [FRating}{W}={Alternative Value} 3. Select Alternative with the Highest Rating
Models In Evaluation
1. Iconic model 2. Analog model 3. Symbolic model
Measurement Scales
1. Nominal Scale 2. Ordinal Scale 3. Interval Scale 4. Ratio Scale
Finite Element Analysis (FEA)
1. Preprocessing 2. Computation 3. Postprocessing
Steps to Build Mathematical Model
1. Problem Statement 2. Define Boundaries of Model (Control Volume) 3. Determine Pertinent Physical Laws and Available Data 4. Identify Assumptions 5. Construct the Model 6. Computation and Verification 7. Validation of Model
Patterns by Which People Cope with Challenges
1. Unconflicted Adherence 2. Unconflicted Change 3. Defensive Avoidance 4. Hypervigilance 5. Vigilance
utility function conclusions
1. can determine a preference ordering of two different amounts 2. decision makers attitude towards risk
psychological stress arises from:
1. decision makers concern about material and social losses that will result from either course of action chosen 2. reputation as competent decision maker is at risk
aids in mathematical modelling
1. dimensional analysis 2. scale models
systematic methods for determining weighting factors
1. direct assignment 2. objective tree 3. AHP
steps in good decision making:
1. establish objectives 2. classify objectives by importance 3. develop alternative actions 4. evaluate alternatives against objectives 5. alternative that hold most promise=tentative decision 6. explore tentative decision for adverse consequences 7. take action to prevent consequences
requirements for selecting a design
1. set of design selection criteria 2. set of alternatives believed to satisfy the set of criteria 3. means to evaluate the design alternatives with respect to each criterion
situation requiring action:
1. should 2. actual 3. must 4. want
Characteristics of Mathematical Models
1. steady state or transient (dynamic) 2. continuous media or discrete events 3. deterministic or probabilistic 4. Lumped or Distributed
Interval Scale
Difference etween arbitrary paris of values can be meaninglfully compared type needed to determine how much worse A is than D addition and subtraction possible no multiplication or division central tendency
data associativity
ability to share digital design data wiht other applications such as finite element analysis without each application having to translate or transmit data
symbolic models
abstractions of teh important quantifiable components of the physical system that use symbols to represent properties of the real system most important model
chance events
aka states of nature on decision tree circles
dimensional analysis
allows you to express a problem with a minimum number of design variables
value
an attribute of an alternative that is implied by choice EX. if a is chosen over b, a has more value than B money is used to express value
maximin decision rule
choose the alternative that maximizes the minimum payoff that can be obtained select alternative that minimizes the maximum loss WORST CASE SCENARIO
outcome
combination of an action and a state of nature
absolute comparison
concept directly compared with a fixed and known set of requirements (PDS)
relative comparison
concepts are compared with each other on the basis of a metric
unconflicted adherence
continue with current action and ignore information about risks of losses
discrete model
deals with individual entities
states of knowledge
degree of certainty associated with each state of nature expressed as probabilities
Decision Under Risk
each action can result in two or more outcomes, but the probabilities of outcomes are unknown (1) Maximin Rule (2) Maximax Rule (2) Combined Criterion
Decision Under Uncertainty
each state of nature has an assigned probability of occurence -using probability of occurence of each state of nature table, find expected value of each material, lowest is answer
states of nature
environment of the decision model usually out of decision makers control
simulation
exercise the model by inputting a series of values to determine the behavior of the proposed design under a stated set of conditions
minimax regret criterion
finds the maximum opportunity loss for each alternative
steady state
input variables are their properties do not change wiht time
intuition
instinctive feeling as to what is probably right (educated guess)