Managing the Computational Cost in a Monte Carlo Simulation by Considering the Value of Information Univ. of Toledo

Format:

Conference/Event

Author/Creator:

Nikolaidis, Nikolaidis, author.

Contributor:

Mourelatos, Zissimos

Pandey, Vijitashwa

Conference Name:

SAE 2012 World Congress & Exhibition (2012-04-24 : Detroit, Michigan, United States)

Language:

English

Physical Description:

1 online resource

Place of Publication:

Warrendale, PA SAE International 2012

Summary:

Monte Carlo simulation is a popular tool for reliability assessment because of its robustness and ease of implementation. A major concern with this method is its computational cost; standard Monte Carlo simulation requires quadrupling the number of replications for halving the standard deviation of the estimated failure probability. Efforts to increase efficiency focus on intelligent sampling procedures and methods for efficient calculation of the performance function of a system. This paper proposes a new method to manage cost that views design as a decision among alternatives with uncertain reliabilities. Information from a simulation has value only if it enables the designer to make a better choice among the alternative options. Consequently, the value of information from the simulation is equal to the gain from using this information to improve the decision. A designer can determine the number of replications that are worth performing by using the method. The study in this paper suggests that one may need much fewer replications than one expects in order to make an informed design decision

Notes:

Vendor supplied data

Publisher Number:

2012-01-0915

Access Restriction:

Restricted for use by site license