Illuminating Tradespace Decisions Using Efficient Experimental Space-Filling Designs for the Engineered Resilient System Architecture
Abstract
System Engineers rely on a variety of models to help understand different viewpoints in several domains throughout a system's life-cycle. These domain models include operational simulations, life-cycle cost models, physics-based computational models, and many more. Currently, there is a technical gap with regard to our ability to untangle the system design drivers within these system life-cycle domains. This technical report proposes a procedural workflow that addresses the technical gap by leveraging the methods of experimental design in order to clearly identify tradable variables and narrow the search for viable system variants. Domain models have their unique set of inputs and outputs. Model inputs represent value properties that define a system alternative configuration or environmental conditions that represent uncertain factors within the system boundary. Model outputs represent measures of performance or effectiveness that allow us to assess alternatives and understand the tradeoffs among several objectives. In order to illuminate the tradeoffs that exist in a complex system design problem we propose an approach that approximates model input and output behavior using the functional form of statistical metamodels. After performing an experimental design we can fit a metamodel that has a functional form known as a response surface. We utilize contour profilers that show horizontal cross sections of multiple response surfaces to visualize where key trade decisions exist. Our purpose is to illuminate trade decisions across several different viewpoints by integrating metamodels that approximate the behavior of multiple domain models. Our research supports the tradespace analytics pillar for the development of the Engineered Resilient System (ERS) Architecture. ERS is a Department of Defense initiative developed by the US Army Engineered Research Development Center.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 30, 2015
- Accession Number
- AD1008897
Entities
People
- Alex Maccalman
- Coleman Grider
- Hunter Wood
- Hyangshim Kwak
- Mary Mcdonald
- Paul Evengelista
- Robert Hill
- Steve Upton
Organizations
- USMA Operations Research Center (ORCEN)