Sampling methods for uncertainty in life cycle cost analysis in product design

Abstract

Life cycle cost (LCC) analysis was initiated from the necessity to calculate the cost of products from ‘cradle to grave’. LCC is not a new tool, but its use is still limited and requires further studies, especially concerning all the uncertainty involved. This project seeks to study and understand the difference between the two most common sampling methods utilized to measure this uncertainty: Latin Hypercube sampling (LHS) and Random sampling (also known as Monte Carlo sampling). The LLC model was developed in SEER-H Galorath Software and the samples from two different types of variable and in two sizes (100 and 1000) were collected using Microsoft Excel and the Risk Solver Add-in. The data was analyzed with a range of statistical tools using the R package software: t-Test, boxplot, scatterplot and the cumulative distribution function. The results point to the fact that LHS seems to be more accurate using small samples, but when comparing big samples the outcome is quite similar in both sampling methods