NASA-Systems Engineering

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: be solved by analytic methods alone. A Monte Carlo simulation is a way of sampling input points from their respective domains in order to estimate the probability distribution of the output variable. In a simple Monte Carlo analysis, a value for each uncertain input is drawn at random from its probability distribution, which can be either discrete or continuous. This set of random values, one for each input, is used to compute the corresponding output value, as shown in Figure 28. The entire process is then repeated k times. These k output values constitute a random sample from the probability distribution over the output variable induced by the input probability distributions. For an example of the usefulness of this technique, recall Figures 2 (in Chapter 2) and 24 (this chapter), which show the projected cost and effectiveness of three alternative design concepts as probability "clouds." These clouds may be reasonably interpreted as the result of three system-level Monte Carlo simulations. The information displayed by the clouds is far greater than that embodied in point estimates for each of the alternatives. An advantage of the Monte Carlo technique is that standard statistical tests can be applied to estimate the precision of the resulting probability distribution. This permits a calculation of the number of runs (samples) needed to obtain a given level of precision. If computing time or costs are a significant constraint, there are several ways of reducing them through more deliberate sampling strategies. See MSFC-HDBK-1912, Systems Engineering (Volume 2), for a discussion of these strategies. Commercial software to perform Monte Carlo simulation is available. These include add-in packages for some of the popular spreadsheets, as well as packages that allow the systems or program analyst to build an entire Monte Carlo model from scratch on a personal computer. These packages generally perform the needed computations in an efficient manner and provide graphical displays of the results, which is very helpful in communicating probabilistic inform...
View Full Document

This document was uploaded on 02/26/2014 for the course E 515 at University of Louisiana at Lafayette.

Ask a homework question - tutors are online