8.4. Preference-Based Engineering Design methodology#
Using both conceptualizations we use the following formal definition of PBED:\
Preference-Based Engineering Design is the search for the optimal design solution, defined as the combination of design variable values that yields the highest overall preference rating for the group of decision makers and does not violate any physical or social constraints. Preference ratings for candidate design solutions are defined using preference functions that relate to each decision maker’s objective(s).\
By relating design variables (physical design parameters of the to be designed object) to objectives (subjective interests of decision makers) and integrating design constraints we ensure that the the most preferred design solution is also feasible.
Given the above conceptualization we can formulate the steps to find the engineering design that is best fit for purpose:
Specify the design variables.
Retrieve decision maker’s objectives.
Determine the preference functions for each objective.
To each objective assign decision maker’s weights.
Determine the design constraints.
Find the optimal design having the highest preference score.
Step 1: Specify the design variables
Design variables are those variables that are under the control of the designer. They relate to the to be designed object, f.i. When designing a building, its geometry, materials, etc. are under the control of the designer. In the case of urban planning design variables are the types of real estate to be developed, the roads, parks, etc. which are under the control of the urban planner. When applied to management the design variables are usually relate to the planning of activities where the manager still has control over when they take place in time.
Step 2: Retrieve decision maker’s objectives
Objectives functions relate to the variables that a decision maker is interested in and wants optimized (minimized or maximized). Such variables can be directly or indirectly related to design variables. For instance, when designing a building its height is a design variable but could also be of interest to the municipality whilst the construction costs are indirectly related to this variable but of interest to another decision maker. All design variables and objectives are elements of a system and therefore need all to be related, directly or indirectly.
Step 3: Specify the preference functions for each objective
Preference functions relate a range of objective values to preference scores. Because preference scores are thus functions of objective values we are able to determine the preference score for each objective of a candidate design solution and therefore also able to determine the overall preference score of that candidate solution.
Step 4: To each objective assign decision maker’s weight
By assigning weights to objectives we can express the relative importance of each objective. In case there are more than one decision maker these can also be used to express a decision maker’s relative importance or power.
Step 5: Determine the design constraints
Design constraints can be used to limit the allowed range of design variables but can also be part of the problem context such as societal rules and regulations or the laws of physics. Design constraints determine the feasible set, the set of allowed design solutions.
Step 6: Find the optimal design having the highest preference score
Given the design constraints, objective and preference functions and weights attached to objectives we can create a mathematical model that can be used to find the candidate design solution that has the highest overall preference rating and is also feasible.