Chapter 8
Optimising Under Uncertainty: Economics and computational challenges
Bringing costs inside the engineering models is essential from a decision-making perspective, particularly when the final goal is ‘Select’. As will be discussed later in this chapter, engineering economics share mathematical similarities with phenomenological or physical modelling, thus opening up an interesting continuum between engineering risk and reliability considerations on the one hand, and economic, financial modelling or actuarial science on the other. Adding then an optimisation layer on top of uncertainty models brings another dimension of CPU complexity to real physical models: design reliability algorithms or adaptive response surface are being developed, but robustness of estimation remains difficult. This chapter does not introduce in depth the economics behind utility theory or non-linear rank transformed utilities, but rather offers some thoughts about the implementation of the computational challenges associated with the embedding of large physical models inside technico-economics. The role of time inside the formulation of the risk measure is discussed, as well as an extension to dynamic decision-making with the real options approach. The chapter concludes with a perspective on the promise of High Performance Computing for modelling in the context of risk and uncertainty.
