Computing the value of information from periodic testing in holistic decision making under uncertainty

Abstract

Abstract Periodic tests are often considered before making intervention decisions in the life-cycle of an engineering system. Value of information (VoI) is a strong tool to quantify the added value from future tests and support rational decision making under uncertainty. While intervention decisions can be made sequentially, the benefits of holistic decision making (HDM) are worth exploration. This paper proposes a holistic decision modelling and optimization approach, considering combined effects of interventions and dependencies in intervention decisions. An algorithm is developed for quantifying the VoI from periodic tests in holistic optimization of multiple interventions (i.e. holistic VoI), based on an extended Bayesian decision analysis framework. The proposed approach is exemplified on a decision problem in structural management, compared with a sequential decision making (SDM) approach. The proposed approach yields optimal decisions associated with higher utilities, due to capturing decision dependencies and combined effects. In addition, the proposed approach yields optimal decisions reliably (whether future tests add value or not) and without the need to prescribe decision rules. Moreover, the relationship between the holistic VoI and the VoI from each test is investigated for the first time. Sensitivities to the parameters characterizing decision contexts and capacities of testing methods are given.