Spring 2005 Seminar Series

DATE: Thursday, February 24, 2005
LOCATION: E40-298
TIME: 4:15pm
Reception immediately following in the Philip M. Morse Reading Room, E40-106

SPEAKER:
Olivier L. de Weck
N. Noyce Assistant Professor of Aeronautics & Astronautics
and Engineering Systems
Massachusetts Institute of Technology

TITLE
Supersize Me? ... or Architecting the Evolution of Engineering Systems

ABSTRACT
Engineering Systems such as telescope arrays, space exploration systems and satellite constellations used to be planned and deployed as a large single monolithic project. Typically, "best guess" assumptions about the necessary capacity and performance (size) of these systems at some future time were used to define requirements and find optimal architectures. Recent business failures, increasing uncertainties about future demand patterns and limited resources for mega-projects drive the need for a new approach to architecting the evolution of these large systems. Supersizing is not the right answer in most cases.

A theoretical framework is emerging that first applies multi-objective optimization to find non-dominated architectures in a deterministic performance-cost-capacity design space. Instead of choosing a single "best" configuration, one decomposes the system into stages, potentially with real options embedded at critical decision branch points. These stages can then be deployed over time, either for capacity/performance expansion or contraction. The external uncertainties are modeled mathematically using Geometric Brownian Motion (GBM), binomial lattices or discrete scenarios. The potential evolution paths of the system are then evaluated against the set of future exogenous evolution paths to find the most valuable initial stage as well as the set of real options to embed ab initio.

Two case studies will be presented in this context. The first retraces the design assumptions and history of Iridium and Globalstar, two billion-dollar class communications satellite constellations dating back to the mid-1990's. It is shown how deliberate staged deployment of these systems could have led to value increases on the order of 30%. The second case describes the architecting and design of the next generation of ground-based radio telescope arrays. These arrays are comprised of hundreds of stations and seek to simultaneously maximize imaging performance (uv-density) while minimizing the cost of connecting stations with fiber optic cables. Surprising results and new topologies emerge when these systems are optimized both with genetic algorithms and via simulated annealing in a static sense. The principles of forward–versus-backwards staging are also compared to allow for staged growth of the arrays over time.