Incorporation of Decision and Game Theories in Early-Stage Complex Product Design to Model End-Use
Mesmer, Bryan L.
MetadataShow full item record
The need for design models that accurately capture the complexities of products increase as products grow ever more complicated. The accuracies of these models depend upon the inputs and the methods used on those inputs to determine an output. Product designers must determine the dominant inputs and make assumptions concerning inputs that have less of an effect. Properly capturing the important inputs in the early design stages, where designs are being simulated, allows for modifications of the design at a relatively low cost. In this dissertation, an input that has a high impact on product performance but is usually neglected until later design stages is examined. The end-users of a product interact with the product and with each other in ways that affect the performance of that product. End-users are typically brought in at the later design stages, or as representations on the design team. They are rarely used as input variables in the product models. By incorporating the end-users in the early models and simulations, the end-users' impact on performance are captured when modifications to the designs are cheaper. The methodology of capturing end-user decision making in product models, developed in this dissertation, is created using the methods of decision and game theory. These theories give a mathematical basis for decision making based on the end-users' beliefs and preferences. Due to the variations that are present in end-users' preferences, their interactions with the product cause variations in the performance. This dissertation shows that capturing the end-user interactions in simulations enables the designer to create products that are more robust to the variations of the end-users. The manipulation of a game that an individual plays to drive an outcome desired by a designer is referred to as mechanism design. This dissertation also shows how a designer can influence the end-users' decisions to optimize the designer's goals. How product controlled information, data in which end-users use to modify their beliefs, can be manipulated by the designer to drive the product to a desired performance is discussed in this dissertation. This dissertation explores multiple examples of how designers can optimize product robustness to end-users and how the designer can use mechanism design on end-users in a product. In order to provide simulated examples of robustness and mechanism design in products with end-users, a complex product simulator is created with end-user interactions modeled using decision and game theory. The products used in this dissertation are that of a building and of an aircraft, with a focus on the structure's evacuations. The model created in this dissertation, Vacate-GT, captures the interactions of the end-users, being the evacuees. The interactions captured are the end-users' exit path selection and velocity vector selection, resulting in a movement decision. This dissertation develops methodologies using decision and game theory to capture these decisions. Vacate-GT is validated against experimental data of building and aircraft evacuations, showing that the end-users' decisions in an evacuation can be accurately modeled using decision and game theory. The validations show how the use of decision and game theories in evacuation simulators result in natural value driven models where individual and group behaviors emerge. Vacate-GT is used to provide simulated examples of how a designer can optimize product robustness to end-user variations and manipulate end-user information to optimize performance. The dissertation shows how end-users can be modeled to create a more accurate product model. Examples using Vacate-GT provide numerical evidence that the use of decision and game theories to incorporate end-user uncertainties into models enables the designer to create more robust products. The simulated examples also show how mechanism design enables the designer to optimize their product goal.